Practical Applications of Digital Pathology Molecular Technologies in the Clinical Diagnostic Laboratory Molecular Assays In

Home , Immunofluorescence

H LEE MOFFITT CANCER CENTER & RESEARCH INSTITUTE INC Nonprofit Org. 12902 Magnolia Drive Tampa FL 33612-9416 PAID Jacksonville, FL

Permit No. 4390

Index Medicus/MEDLINE Index

is included in included is Control Cancer

Special Technologies for Ex Vivo Analysis of Cancer of Analysis Vivo Ex for Technologies Special

Lesions of the Breast the of Lesions

Epithelial in Stains Immunohistochemical of Use

Pathway Toward Personalized Medicine Personalized Toward Pathway

Advances in the Molecular Analysis of Breast Cancer: Cancer: Breast of Analysis Molecular the in Advances

in Invasive Micropapillary Colorectal Carcinoma Colorectal Micropapillary Invasive in

Overexpression of Vascular Endothelial Growth Factor A A Factor Growth Endothelial Vascular of Overexpression

Adult Diffuse Gliomas Diffuse Adult

Impending Impact of Molecular Pathology on Classifying Classifying on Pathology Molecular of Impact Impending

Adenocarcinoma

as a Predictive Marker in Lung Lung in Marker Predictive a as EGFR in Advances

Role in Clinical Practice Clinical in Role

Molecular Pathology of Soft-Tissue Neoplasms and Its Its and Neoplasms Soft-Tissue of Pathology Molecular

Surgical and Molecular Pathology of Barrett Esophagus Barrett of Pathology Molecular and Surgical

Development and Therapeutic Effectiveness Therapeutic and Development

Circulating Tumor Cells: A Window Into Tumor Tumor Into Window A Cells: Tumor Circulating

of Molecular Testing in Hematopathology in Testing Molecular of

Biomarkers in Hematological Malignancies: A Review Review A Malignancies: Hematological in Biomarkers

Molecular Assays in Cytopathology for Thyroid Cancer Thyroid for Cytopathology in Assays Molecular

Laboratory

Molecular Technologies in the Clinical Diagnostic Diagnostic Clinical the in Technologies Molecular

Practical Applications of Digital Pathology Digital of Applications Practical

H. LEE MOFFITT CANCER CENTER & RESEARCH INSTITUTE, AN NCI COMPREHENSIVE CANCER CENTER CANCER COMPREHENSIVE NCI AN INSTITUTE, RESEARCH & CENTER CANCER MOFFITT LEE H.

cancercontroljournal.org Vol. 22, No. 2, April 2015 April 2, No. 22, Vol. MOFFITT PERFORMS THE MOST BLOOD & MARROW TRANSPLANTS IN FLORIDA

TOGETHER WE CAN OFFER YOUR PATIENTS:

• Transplant as an option. With lower treatment toxicities and better donor matching, candidacy is no longer limited solely by chronological age

• Expertise in donor selection and state-of-the-art transplant strategies

• The ability to have post-transplant care, including immunosuppression suppression, managed close to home. Our commitment to you is close communication regarding our shared patients

• Transplant at the optimal time. Timing of referral for transplant may impact outcomes

TO REFER A PATIENT OR DISCUSS OPTIMAL REFERRAL TIMING, SPEAK TO ONE OF OUR TRANSPLANT PHYSICIANS.

CALL 1-888-MOFFITT

MOFFITT PERFORMS THE Editorial Board Members

Editor: Rami Komrokji, MD Production Staff: Associate Member Lodovico Balducci, MD Veronica Nemeth Malignant Hematology MOST BLOOD & MARROW Senior Member Editorial Coordinator Conor C. Lynch, PhD Program Leader, Senior Adult Oncology Program Don Buchanan Assistant Member Moffitt Cancer Center Graphic Designer TRANSPLANTS IN FLORIDA Tumor Biology Deputy Editor: Amit Mahipal, MD Consultants: Assistant Member Julio M. Pow-Sang, MD Sherri Damlo Senior Member Clinical Research Unit Gastrointestinal Oncology Medical Copy Editor Chair, Department of Genitourinary Oncology and Art Coordinator Director of Moffitt Robotics Program Kristen J. Otto, MD Kelly Young Moffitt Cancer Center Assistant Member Graphic Designer Head & Neck Oncology Editor Emeritus: Michael A. Poch, MD Associate Editor of Assistant Member John Horton, MB, ChB Educational Projects Professor Emeritus of Medicine & Oncology Genitourinary Oncology John M. York, PharmD Jeffery S. Russell, MD, PhD Akita Biomedical Consulting Moffitt Cancer Center Assistant Member 1111 Bailey Drive Endocrine Tumor Oncology Journal Advisory Committee: Paso Robles, CA 93446 Aliyah Baluch, MD Elizabeth M. Sagatys, MD Phone: 805-238-2485 Assistant Member Assistant Member Fax: 949-203-6115 Infectious Diseases Pathology - Clinical E-mail: [email protected] Dung-Tsa Chen, PhD Jose E. Sarria, MD Associate Member Assistant Member For Consumer and Biostatistics Anesthesiology General Advertising Hey Sook Chon, MD Saïd M. Sebti, PhD Information: Assistant Member Senior Member Veronica Nemeth Gynecological Oncology Drug Discovery Editorial Coordinator Jasreman Dhillon, MD Bijal D. Shah, MD Cancer Control: Assistant Member Assistant Member Journal of the Moffitt Cancer Center Pathology - Anatomic Malignant Hematology 12902 Magnolia Drive – MBC-JRNL Tampa, FL 33612 Jennifer S. Drukteinis, MD Lubomir Sokol, MD, PhD Phone: 813-745-1348 Associate Member Senior Member Fax: 813-449-8680 Diagnostic Radiology Hematology/Oncology E-mail: [email protected] Timothy J. George, PharmD Hatem H. Soliman, MD Pharmacy Residency Director Assistant Member Clinical Pharmacist - Malignant Hematology Breast Oncology Clement K. Gwede, PhD Jonathan R. Strosberg, MD Associate Member Assistant Member Health Outcomes & Behavior Gastrointestinal Oncology Sarah E. Hoffe, MD Sarah W. Thirlwell, RN Associate Member Nurse Director Radiation Oncology Supportive Care Medicine Program Cancer Control is a member of John V. Kiluk, MD Eric M. Toloza, MD, PhD the Medscape Publishers’ Circle®, Associate Member Assistant Member an alliance of leading medical Breast Oncology Thoracic Oncology publishers whose content is Richard D. Kim, MD Nam D. Tran, MD featured on Medscape Associate Member Assistant Member (www.medscape.com). Gastrointestinal Oncology Neuro-Oncology Bela Kis, MD, PhD Jonathan S. Zager, MD Most issues and supplements of Assistant Member Associate Member Cancer Control are available at Diagnostic Radiology Sarcoma and Cutaneous Oncology cancercontroljournal.org

CANCER CONTROL: JOURNAL OF THE MOFFITT CANCER CENTER (ISSN 1073-2748) is published by H. Lee Moffitt Cancer Center & Research Institute, 12902 Magnolia Drive, Tampa, FL 33612. Telephone: 813-745-1348. Fax: 813-449-8680. E-mail: [email protected]. Internet address: cancercontroljournal.org. Cancer Control is included in Index Medicus ®/MEDLINE® and EMBASE®/ Excerpta Medica, Thomson Reuters Science Citation Index Expanded (SciSearch®) and Journal Citation Reports/Science Edition. Copyright 2015 by H. Lee Moffitt Cancer Center & Research Institute. All rights reserved. Cancer Control: Journal of the Moffitt Cancer Center is a peer-reviewed journal that is published to enhance the knowledge needed by professionals in oncology to help them minimize the impact of human malignancy. Each issue emphasizes a specific theme relating to the detection or management of cancer. The objectives of Cancer Control are to define the current state of cancer care, to integrate recently generated information with historical practice patterns, and to enlighten readers through critical reviews, commentaries, and analyses of recent research studies. DISCLAIMER: All articles published in this journal, including editorials and letters, represent the opinions of the author(s) and do not necessarily reflect the opinions of the editorial board, the H. Lee Moffitt Cancer Center & Research Institute, Inc, or the institutions with which the authors are affiliated unless clearly specified. The reader is advised to independently verify the effectiveness of all methods of treatment and the accuracy of all drug names, dosages, and schedules. Dosages and methods of administration of pharmaceutical products may not be those listed in the package insert and solely reflect the experience of the author(s) and/or clinical investigator(s).

April 2015, Vol. 22, No. 2 Cancer Control 129 Table of Contents

Editorial

Personalizing the Attack on Cancer: Zeroing in on Targets 133 Anthony M. Magliocco, MD

Articles

Practical Applications of Digital Pathology 137 Daryoush Saeed-Vafa, MD, and Anthony M. Magliocco, MD

Molecular Technologies in the Clinical Diagnostic Laboratory 142 Victor E. Zota, MD, and Anthony M. Magliocco, MD

Molecular Assays in Cytopathology for Thyroid Cancer 152 Pablo Valderrabano, MD, Victor E. Zota, MD, Bryan McIver, MD, PhD, Domenico Coppola, MD, and Marino E. Leon, MD

Biomarkers in Hematological Malignancies: 158 A Review of Molecular Testing in Hematopathology Mohammad Hussaini, MD

Circulating Tumor Cells: A Window Into Tumor Development 167 and Therapeutic Effectiveness Gisela Cáceres, PhD, John A. Puskas, PhD, and Anthony M. Magliocco, MD

Surgical and Molecular Pathology of Barrett Esophagus 177 Sherma Zibadi, MD, PhD, and Domenico Coppola, MD

Molecular Pathology of Soft-Tissue Neoplasms and Its Role in Clinical Practice 186 Evita B. Henderson-Jackson, MD, and Marilyn M. Bui, MD, PhD

130 Cancer Control April 2015, Vol. 22, No. 2 Table of Contents

Advances in EGFR as a Predictive Marker in Lung Adenocarcinoma 193 Farah K. Khalil, MD, and Soner Altiok, MD, PhD

Impending Impact of Molecular Pathology on Classifying Adult Diffuse Gliomas 200 Robert J. Macaulay, MD

Overexpression of Vascular Endothelial Growth Factor A in 206 Invasive Micropapillary Colorectal Carcinoma Marilin Rosa, MD, Maisoun Abdelbaqi, MD, Katherine M. Bui, Aejaz Nasir, MD, MPhil, Marilyn M. Bui, MD, PhD, David Shibata, MD, and Domenico Coppola, MD

Advances in the Molecular Analysis of Breast Cancer: 211 Pathway Toward Personalized Medicine Marilin Rosa, MD

Use of Immunohistochemical Stains in Epithelial Lesions of the Breast 220 Laila Khazai, MD, and Marilin Rosa, MD

Special Technologies for Ex Vivo Analysis of Cancer 226 Jenny M. Kreahling, PhD, and Soner Altiok, MD, PhD

Departments

Editorial: Should We Preclude Phase 1 Clinical Trials From Enrolling Elderly Individuals? 232 Lodovico Balducci, MD Original Research: Effect of Age on Clinical Outcomes in Phase 1 Trial Participants 235 Amit Mahipal, MD, Aaron C. Denson, MD, Benjamin Djulbegovic, MD, PhD, Richard Lush, PhD, Ambuj Kumar, MD, Tzu-Hua Juan, MS, Michael J. Schell, PhD, and Daniel M. Sullivan, MD

Case Report: Primary Enteropathy-Associated T-Cell Lymphoma Type 2: An Emerging Entity? 242 Nicole M. Grigg-Gutierrez, MD, Rodolfo Estremera-Marcial, MD, William W. Cáceres, MD, and Doris H. Toro, MD

April 2015, Vol. 22, No. 2 Cancer Control 131 Table of Contents

Special Report: Cost Effectiveness of Colorectal Cancer Screening Strategies 248 Shaan S. Patel and Meredith L. Kilgore, PhD

Editorial: Colorectal Cancer: The Last Nail in the Coffin for Androgen Deprivation Therapy for Prostate Cancer? 259 Lodovico Balducci, MD, and Julio M. Pow-Sang, MD

Editorial: Does Androgen Deprivation Therapy for Prostate Cancer Increase the Risk of Colorectal Cancer? 261 Anna Martling, MD, PhD

Special Report: Risk of Colorectal Cancer by Subsite in a Swedish Prostate Cancer Cohort 263 Yunxia Lu, MD, PhD, Rickard Ljung, MD, PhD, Anna Martling, MD, PhD, and Mats Lindblad, MD, PhD

Ten Best Readings Relating to Molecular Biomarkers 271

About the art in this issue: Craig Damlo is an innovation expert specializing in creating a culture of innovation and design thinking. With a background in aerospace product and innovation development, Craig helps companies understand how to best utilize their resources in order to maximize innovation within their own groups. Craig has always been interested in how things work and problem solving, constantly taking things apart as a child, and this desire drove him to use photography as a way to capture what he was observing around him. He uses his knowledge of physics to tell a photographic story of how time and movement affect image capture and how the passing of time—as well as the human hand— works to reclaim objects, as showcased in this collection. Craig also uses photography to hone his innovation skills through careful, patient observation and the capture of still moments to observe just how things operate. Learn more about Craig and his work at Getty Images, www.flickr.com/craigdamlo, and www.soapboxrocket.com. Cover: Quality Paint

Pages: 130-131 Life in Stone The Chair Stands Alone Home Is Where the Heart Is Sam's The Closest Exit May Be Behind You Three Layers of Paint A Quiet Sound The Old Barn Flight Deck Circa 1969 This Way to Gas The Broken Hall The Oven Vent Past Flights .

132 Cancer Control April 2015, Vol. 22, No. 2 Editorial

Personalizing the Attack on Cancer: Zeroing in on Targets

The “War on Cancer” continues with some recent has provided the knowledge underlying the classical encouraging victories. classification of malignancy based on the organ and Advances in genomics and other laboratory-based cells of origin along with the assessment of biological diagnostic technologies have transformed our under- appearance, which is used for grading. Microscopic standing of the fundamental molecular basis of cancer, examination of surrounding tissues and lymph nodes and applying this knowledge has led to new genera- create the current staging systems. tions of targeted therapeutics. Pathologists are at the Advances in analytics, staining reagents, digital center of this revolution of personalized medicine microscopy, and computer-assisted vision have creat- as they are faced with the challenges of deploying ed new opportunities to analyze tissue to extract more emerging technologies into routine patient care. In ad- detailed information, including more precise measure- dition, pathologists are grappling with the challenge ments of prognostic and predictive biomarkers, such of combining molecular knowledge with long-estab- as protein and nucleic acids. Improvements in digital lished systems of tumor classification based on light image analysis will help reduce errors in pathology microscopy. and provide more accurate information for treatment Unfortunately, pathology practices are often selection. To this point, Dr Zota and I review key ad- also battling diminishing reimbursements and in- vances in the analytical methods of molecular biology creasing regulatory burdens. The latest technologies that recently transformed pathology. The ability to are often considered “research use only,” forcing accurately analyze key mutations in driver mutations clinical laboratories into complex in-house valida- such as EGFR in lung cancer and BRAF in melanoma tion processes to qualify their use under the Clinical from routinely acquired tissue specimens has enabled Laboratory Improvement Amendments as laborato- personalized medicine and targeted therapies to be ry-developed tests (LDTs). Frequently, third-party routinely implemented. payers deny coverage of new diagnostic technol- The decreased costs of next-generation sequenc- ogies, putting some laboratories out of business. ing now enable hundreds of genes — up to whole Further complicating matters is the intention of the genomes — to be cost-effectively analyzed in the hos- US Food and Drug Administration to more closely pital laboratory. In fact, one of the greatest current scrutinize hospital laboratory–developed LDTs. challenges is the management of “information over- Despite these challenges, there is no question load.” The so-called “long tail” of cancer mutations that molecular analysis will become routine and fun- illustrates that cancer is a diverse disease character- damental to the diagnosis and management of cancer. ized by a multitude of diverse driver mutations and The application of the knowledge derived from a degree of genomic liquidity that results in varied the completion of the Human Genome Project and genomic “landscapes” and evolutionary “trees” within the rapid collapse in costs for deep genetic and tran- tumors. Some driver mutations occur in small numbers scriptomic analysis are fundamentally transforming of tumors, making it difficult to conduct convention- the way pathologists understand and classify cancer al, large-scale clinical trials. This fact has led to the and will certainly impact the way oncologists manage creation of basket-type trials in which tumors are and treat malignancies in their patients. analyzed and then matched to selective therapeutic In this issue of Cancer Control, members of the inhibitors. anatomical and clinical pathology departments of We now conceptualize cancer as a 4-dimensional the H. Lee Moffitt Cancer Center & Research Insti- disease in a patient that evolves over time with pop- tute review several topics of interest to oncologists ulations of malignant cells that react, migrate, and and other health care professionals, focusing on the adapt to the therapies oncologists unleash on them. transformative effects of molecular pathology on the We have also learned that some mutational drivers and contemporary practice of pathology. events cross anatomical locations, forcing health care Dr Saeed-Vafa and I present an update on devel- professionals to consider novel molecular classifica- opments in digital and analytical microscopy. Light tions that transcend conventional histological ones. microscopy has been the fundamental tool for pathol- On a positive note, the deployment of novel tools, ogists to analyze cancer for more than 100 years and such as digital polymerase chain reaction, will help

April 2015, Vol. 22, No. 2 Cancer Control 133 us measure trace amounts of biomarkers in biosam- methods may enable us to better manage this diffi- ples, such as blood and urine, potentially allowing cult clinical condition and avoid overtreatment and more accurate screening, monitoring, and effective overscreening. treatment selection, including the discontinuation of Drs Henderson-Jackson and Bui review new medically futile and toxic therapies and the switch to developments in the area of soft-tissue neoplasms and more effective ones. the application of molecular classification approaches Furthermore, the advances in immunotherapy are to improve the classification of these heterogeneous challenging pathologists to provide new, accurate, and lesions and, perhaps more importantly, to uncover predictive biomarkers to improve patient selection therapeutically useful targets that could be used for and understand response. Some of these biomarkers treatment. may include analysis of the complex composition of Drs Khalil and Altiok review how the treatment the tumor microenvironment. course for patients with lung cancer has been altered The paper by Dr Valderrabano and colleagues in recent years with the uncovering of the role of illustrates the impact of molecular analysis on con- key molecular drivers, including EGFR mutation, in ventional optical microscopy for the assessment of a subset of cases. The uncovering of these common thyroid lesions. The accurate presurgical analysis and mutations in lung cancer has expanded the demand classification of thyroid lesions are essential to avoid for the routine molecular testing of solid tumors unnecessary morbidity. Conventional cytopathology and provided the impetus for routine solid tumor has difficulty distinguishing certain benign from ma- molecular diagnostic laboratories in many specialized lignant lesions based on light microscopy, thus leading hospitals. In addition, the study of acquired resis- to frequent overtreatment in some patients and pos- tance mechanisms has uncovered new insights into sible undertreatment in others. The identification and how tumors evolve and evade personalized treatment routine deployment of molecular markers to improve options, opening up new strategies for overcoming classification of indeterminate lesions will be of great these resistance mechanisms. benefit to patients and surgeons and will reduce costs Dr Macaulay discusses how advances in complex to the health care system by limiting unnecessary molecular testing have significantly impacted the sub- surgical procedures. specialty of neuropathology and the classification and Researchers in the field of molecular technolo- understanding of brain tumors. The uncovering of gies have made inroads into the diagnosis and man- driver mutations in IDH1 and other molecular events, agement of malignant hematological conditions. including chromosomal alterations and DNA epigen- Dr Hussaini reviews current trends in hematological etic events, have driven the development of routine malignancy testing. These methods and approaches molecular analysis of brain cancer to improve the often foreshadow similar advances in the approach classification of tumors and drive treatment selection. to solid tumors and provide a roadmap for effectively In an article by Dr Rosa and coauthors regarding implementing novel technology platforms. the overexpression of vascular endothelial growth Dr Cáceres and others introduce us to the con- factor A in invasive micropapillary colorectal carci- cept of the “liquid biopsy” and circulating tumor noma, we are reminded that immunohistochemistry cells (CTCs). CTCs hold promise for the sequential is a molecular technology — essentially “in situ pro- temporal examination of malignancy prior to, during, teomics” — that enables us to evaluate the molecular and after therapy. CTC technologies coupled with characteristics of histological structures. Dr Rosa and other molecular methods to analyze cell-free DNA colleagues show that spaces originally thought to be in blood are expected to provide us with tools to “preparation artifacts” are actually neovascular struc- better select therapy for metastatic disease and en- tures. This aspect of the biology of micropapillary able health care professionals to determine earlier carcinoma may explain its clinical aggressiveness and if a particular therapy is succeeding or failing, thus potentially offer new therapeutic targets. allowing for the potential real-time modulation of Breast cancer is one of the strongest examples treatment strategies. Further refinements to the sen- of how molecular methods have transformed our sitivity rates of liquid-biopsy technologies are also understanding of the pathology and classification of expected to lead to improved screening and the ear- this disease. Historically, classification was exclusively lier detection of cancer, improved molecular-based based on microscopic examination. However, this clas- staging, and novel and practical methods to monitor sification has now been transformed with extensive minimal residual disease. basic, translational, and clinical research, resulting Drs Zibadi and Coppola review the complex chal- in a new molecular understanding of breast cancer. lenges of diagnosing and managing Barrett esopha- The molecular analysis of breast cancer has uncov- gus, which represents a problematic lesion that can ered “molecular portraits” that help define distinct progress to malignancy. Use of novel molecular testing subtypes of the disease with different clinical behav-

134 Cancer Control April 2015, Vol. 22, No. 2 iors that each require different treatment strategies. In 2 papers, Drs Rosa and Khazai review current methods and trends in the molecular classification of breast cancer and the implications for the improved, enhanced quality of treatment and better outcome for patients. Classical pathology methods typically require “fixing” tumor tissue, which essentially is killing and preserving the cancer as a frozen moment in time. Although this method has yielded spectacular divi- dends, it is still a “snapshot” of a very dynamic living system composed of a complex mixture of tissue types, including malignant cells, and a variety of host cells. Drs Kreahling and Altiok discuss alternate methods to maintain cancer cells ex vivo that may enable the detailed evaluation of drug-sensitivity profiles, thus arming health care professionals with the ability to select more effective treatments for their patients. Also included in the April issue of Cancer Control is original research from Dr Mahipal and colleagues presenting data on the effect of age on clinical outcomes among patients enrolled in phase 1 clinical trials at Mof- fitt Cancer Center. Dr Grigg-Gutierrez and others share a case report on primary enteropathy-associated T-cell lymphoma type 2. In 2 Special Reports, Mr Patel and Dr Kilgore provide a systematic review of the cost effectiveness of colorectal cancer screening strategies, and Dr Lu and colleagues discuss the risk of colorectal cancer by subsite in a nationwide prostate cancer cohort in Sweden. Pathology has been profoundly affected by these changes and pathologists — frequently unrecognized by patients — serve as critical members of the health care team. Pathologists remain key to deploying new diagnostic molecular technologies in the ongoing battle against cancer, providing the coordinates and advanced surveillance required for correctly aiming the powerful next generation of targeted therapeutic weaponry in the ongoing war. In summary, we are living in revolutionary times in terms of how we understand, diagnose, and treat cancer. I hope you enjoy and benefit from reading this issue of Cancer Control.

Anthony M. Magliocco, MD Senior Member Department Chair, Anatomic Pathology H. Lee Moffitt Cancer Center & Research Institute Tampa, Florida [email protected]

April 2015, Vol. 22, No. 2 Cancer Control 135 Department of Anatomic Pathology H. Lee Moffitt Cancer Center & Research Institute The Department of Anatomic Pathology at the H. Lee Moffitt Cancer Center & Research Institute is an academic department that participates in the teaching of medical students, residents, and anatomical pathology fellows in its accredited fellowship programs. Research is also a significant component of the academic mission of the department and its pathologists are engaged in collaborating with other Moffitt Cancer Center investigators and researchers on clinical trials and the publication of study results. One of the main focuses of the department has been the identification and application of new prognostic markers for the analysis of solid tumors. This includes an extensive immunohistochemistry laboratory, an advanced molecular diagnostics facility that recently introduced clinical next-generation sequencing, digital microscopy, and technologies for the capture and analysis of circulating tumor cells. The pathologists also work with other Moffitt programs in creating algorithmic pathways that provide a standardized tool to guide patient care and treatment plans as well as translate new discoveries into clinical diagnostics. New testing and methodologies are continuously being added.

Department Chair Anthony M. Magliocco, MD

Vice Chair, Clinical Services Barbara A. Centeno, MD

Programs and Services

Fellowships

Autopsy Education Cytopathology Robert Macaulay, MD Marilyn M. Bui, MD, PhD Marilyn M. Bui, MD, PhD Medical Director Program Leader Program Director Clinical Testing Development Mentoring and Career Dermatology Pathology Anthony M. Magliocco, MD Development Jane L. Messina, MD Medical Director Jane L. Messina, MD Program Director Cytopathology Program Leader Gastrointestinal Pathology Barbara A. Centeno, MD Research and Development Domenico Coppola, MD Medical Director Soner Altiok, MD, PhD Program Director Histology Program Leader Molecular Pathology Farah K. Khalil, MD Anthony M. Magliocco, MD Medical Director Program Director Surgical Pathology Surgical Pathology Ardeshir Hakam, MD Ardeshir Hakam, MD Medical Director Program Director

Faculty Members by Specialty

Breast Dermatopathology Gastrointestinal Genitourinary Gynecology Laila Khazai, MD Timothy W. McCardle, MD Barbara A. Centeno, MD Jasreman Dhillon, MD Ardeshir Hakam, MD Anthony M. Magliocco, MD Jane L. Messina, MD Domenico Coppola, MD Shohreh Dickinson, MD Ozlen Saglam, MD Marilin Rosa, MD Masoumeh Ghayouri, MD Kun Jiang, MD, PhD

Head and Neck Molecular Neuropathology Sarcoma Thoracic Marino E. Leon, MD Domenico Coppola, MD Robert J. Macaulay, MD Marilyn M. Bui, MD, PhD Soner Altiok, MD, PhD Jason G. Savell, MD Anthony M. Magliocco, MD Evita Henderson-Jackson, MD Farah K. Khalil, MD Dahui Qin, MD, PhD Prudence V. Smith, MD

For more information about the Department of Anatomic Pathology, please call 813-745-3741 (during normal business hours). www.MOFFITT.org

136 Cancer Control April 2015, Vol. 22, No. 2 Digital pathology will continue

to advance and provide precise

prognostic information and

help guide treatment decisions.

Life in Stone. Photograph courtesy of Craig Damlo. www.soapboxrocket.com.

Practical Applications of Digital Pathology Daryoush Saeed-Vafa, MD, and Anthony M. Magliocco, MD

Background: Virtual microscopy and advances in machine learning have paved the way for the ever-expanding field of digital pathology. Multiple image-based computing environments capable of performing automated quantitative and morphological analyses are the foundation on which digital pathology is built. Methods: The applications for digital pathology in the clinical setting are numerous and are explored along with the digital software environments themselves, as well as the different analytical modalities specific to digital pathology. Prospective studies, case-control analyses, meta-analyses, and detailed descriptions of software environments were explored that pertained to digital pathology and its use in the clinical setting. Results: Many different software environments have advanced platforms capable of improving digital pathology and potentially influencing clinical decisions. Conclusions: The potential of digital pathology is vast, particularly with the introduction of numerous software environments available for use. With all the digital pathology tools available as well as those in development, the field will continue to advance, particularly in the era of personalized medicine, providing health care professionals with more precise prognostic information as well as helping them guide treatment decisions.

Introduction for the digital era of pathology. This has also led to Advances in molecular biology are transforming im- the development of multiple, image-based computing age-based specialties like pathology from primarily environments capable of quantitative and morpho- subjective-based analyses to quantitative-based digital logical analyses. It is possible that we may see auto- analyses. The technologies used to capture high-res- mated systems in the future assisting pathologists in olution images, create digital versions of glass slides, examining tissue by performing some level of triage, and make virtual microscopy possible, as well as ad- similar to what has been done with the Papanicolaou vance machine-based learning, have paved the way test in cytopathology. At the time of publication, several digital software products have been developed, including AQUA (HistoRx, New Haven, Connecticut), From the Departments of Imaging Research (DS-V) and Anatomic Pathology (AMM), H. Lee Moffitt Cancer Center & Research Institute, Tissue Studio (Definiens, Munich, Germany), HALO Tampa, Florida. (PathXL, Belfast, Northern Ireland), ImageJ (National Submitted February 18, 2015; accepted April 3, 2015. Institutes of Health, Bethesda, Maryland), and many Address correspondence to Anthony M. Magliocco, MD, Department others with analytical capabilities that could be ap- of Anatomic Pathology, Moffitt Cancer Center, 12902 Magnolia plied to digital pathology images. Drive, Tampa, FL 33612. E-mail: [email protected] No significant relationships exist between the authors and the companies/organizations whose products or services may be Digital Image Analysis referenced in this article. The key area in which digital microscopy has an ad-

April 2015, Vol. 22, No. 2 Cancer Control 137 vantage over a human observer — even an expert targets a molecule and then a secondary antibody — is in the precise quantification of the features in coupled to a fluorophore (a fluorescent chemical a digital image. This might include counting features compound that can re-emit light if it is excited) rec- in a defined area, such as nuclei or mitotic figures, ognizes and binds to the primary antibody. Using measuring precise distances, or recording the intensity different antibodies and fluorophores, different com- of a stain in defined cell populations. The analysis is partments and targets can be specifically marked. only as good as the quality of the algorithms for iso- Typically, a digital camera will capture sequential lating the feature of interest (eg, separating malignant images of the multiplex-stained tissue by capturing nuclei from benign ones) or mapping areas of cancer an exposure through successive filters, thus isolating from surrounding normal areas. In addition, machine each of the target probes. AQUA then creates tis- algorithms may be stymied by preanalytical variables sue “masks” defined by specifically stained features. (eg, inadequate fixation, necrosis, warm ischemia). For example, the diamidino-phenylindole stain will However, the ability of computers to measure features identify nuclei, so AQUA can be used to define a with great precision lends itself to applications such nuclear mask, which can then be used to isolate as screening large numbers of slides for small areas of the key pixels where the target antibody signal is abnormality, precisely quantifying important features captured. Combinations of masks can be created and in malignant cells (eg, proliferative rate or presence arranged so that signal intensity in various tissue of tumor-associated lymphocytes), and quantifying components (eg, estrogen receptor within nuclei of important protein markers used in therapy selection keratin-positive cells) can be digitally isolated and (eg, tissue levels of estrogen receptor in breast can- precisely quantified. The technology then provides a cer). With emerging interest in immunotherapies, score for the particular target in any area of interest computer-assisted approaches are being developed (Figs 1 and 2A–C). This AQUA score is calculated to better quantify the immune cell components of from the sum of the target pixel intensity divided the microenvironment. More sophisticated analyses by the compartment area. All AQUA scores are also include the development of artificial intelligence or normalized for image exposure times and are directly machine learning capable of recognizing key mor- proportional to molecules per unit area or protein phological features in tissue sections. concentration.2 Providing an objective and quantitative method for determining protein expression AQUA has applications in the research setting and clinical In 2005, McCabe et al1 developed an approach for practice in terms of providing precise information AQUA using immunofluorescent-stained slides. De- regarding the subcellular location and quantity or termining protein expression via traditional immuno- concentration of a particular antigen — quantities histochemistry (IHC) is a technique that allows the that may have important implications regarding the direct visualization of protein expression in tissue sections in situ. However, IHC has many disadvantages, including the subjectivity of interpretation due to a variety of reasons (eg, interobserver variability and variability from one assay to another).2 Despite its limitations, IHC is useful and is standard in the surgical-pathology laboratory and most hospitals. Dig- ital analysis of fluorescent-labeled, IHC-stained tissue specimens may make it possible to address some of the limitations inherent in classical chromogenic IHC. The dynamic range for immunofluorescent labeling is far larger than for standard insoluble product assays, thus allowing for more precise quantification. In addition, the ability to stain the same tissue for multiple targets, multiplexing by using different fluorescent labels, opens up the possibility of isolating subcompo- nents of tissue (eg, nuclei, keratin-positive cells) and potentially normalizing quantification across different assay runs. AQUA is based on these key features of fluores- Fig 1. — Urothelial carcinoma tissue core stained via an immunoflu- cent IHC.2,3 It analyzes digital images captured from orescence assay as follows: blue = nuclear (diamidino-phenylindole), green = cytoplasmic (pan-cytokeratin, AE1/AE3), and red = target antibody. slides stained with the typical immunofluorescence The MRE11 AQUA (HistoRx, New Haven, CT) scores are numerical represen- protocol in which an unlabeled primary antibody tations of the quantity of the target in each different compartment.

138 Cancer Control April 2015, Vol. 22, No. 2 mance of in situ–based tissue biomarkers.

Tissue Studio Tissue Studio is an image analysis environment that takes a context-based, relational approach, rather than a pixel-based, pattern recognition approach. Its segmentation algorithms begin by grouping pixels into objects called image object primitives. These objects are then merged into what the user recognizes as the microanatomy of the tissue and are then related to the super objects (eg, white space, lumen, tissue, background, glands). Because of this hierarchical relationship, which links the smallest elements in an image to the largest, Tissue Studio A provides a means for detecting and quantifying hundreds of relational attributes of the tissue that could be related to clinical outcome. Doing so enables pathologists to ask more advanced questions of the underlying biology. A novel example demonstrating the power of this tool came from Beck et al4 in which the researchers used the Developer XD (Definiens) image analysis environment to create a processing pipeline consisting of basic image processing and feature construction, training and application of the epithelium/stroma classifier, and construction of higher-level features to a create a computational pathologist (C-Path) to analyze microscopic images taken from patients with breast cancer. C-Path automatically quantified morphological features derived from the images to create a prognostic model.4 The researchers discovered B that the score of the prognostic model was strongly associated with overall survival rates and was independent of clinical, pathological, and molecular factors.4 Their analysis was not predicated on features predefined by an expert pathologist; thus, their image analysis system was mainly automated and had no manual steps.4 Furthermore, C-Path identified features in the stroma as being better predictors of patient survival rates than the conventional use of the tumor epithelial cells themselves.4 That stromal features provided statistically significant prognostic information was a novel finding because morphological characteristics of the stroma had not been previously identified by pathologists as providing clinically significant information.4 Another example of the utility of this software C came in 2013 when Definiens and Metamark Ge- netics (Cambridge, Massachusetts) agreed to use Fig 2A–C. — Urothelial carcinoma AQUA (HistoRx, New Haven, CT) component masks. These figures represent the (A) nuclear, (B) cytoplasmic, the image analysis platform for a dual-marketed, and (C) tumor masks used to calculate the AQUA score from the urothelial multiplex, immunofluorescence prognostic test carcinoma tissue core in Fig 1. for prostate cancer.5 The companies claim that the effectiveness of a particular therapy. AQUA scores are technology they marketed together automatically reproducible and provide continuous variable data classifies tumors of the prostate as “regions of inter- that can be used to define superior marker classifiers est” and will quantify the expression of proteomic and improve the prognostic and predictive perfor- markers in biopsy specimens.5 This technology is

April 2015, Vol. 22, No. 2 Cancer Control 139 likely to help health care professionals make better types of cancers. However, these counts have many of informed treatment decisions. the same disadvantages as the aforementioned Ki67 proliferation index. Ciresan et al14 implemented a deep Fractal Analysis neural network to detect mitosis in slides stained with Mandelbrot6 introduced the concept of fractals in hematoxylin and eosin from samples gathered from 1967. A fractal is a complex, never-ending pattern patients with breast cancer. The deep neural network self-similar across different scales and generated by operates in a feedforward manner, with simple pixel a process in a recursive feedback loop.7 Formally, patterns becoming successively more advanced rep- fractals are defined by measurements of their shape, resentations as they pass through the network. This such as the degree of their boundary fragmentation or method was powerful for detecting mitotic figures and their irregularity across multiple scales. Fractal analy- outperformed all competing approaches examined sis can characterize irregular structures that maintain a by Ciresan et al.14 constant level of complexity across a range of scales.8,9 A practical application of fractal analysis in the field Circulating Tumor Cells of medicine is analysis of the morphological com- Circulating tumors cells (CTC) are an ever-expanding plexity of tumors. For example, an automated frac- field of personalized medicine and are the front line tal analysis technique has been used to quantify the of cancer treatment. CTCs are released into the blood morphological complexity of breast epithelium and via malignant tumors and can be detected in the blood prostate cancer,10-12 and the morphological complex- with a highly sensitive and specific capture method. ity of the breast epithelial architecture was strongly Current techniques face the challenge of detecting the and significantly associated with disease-specific and minute numbers of CTCs among the plethora of white overall survival rates.10 Thus, the use of automated and red blood cells circulating in the blood. A novel fractal analysis provides a novel approach to devel- application related to CTCs for digital pathology might oping better objective prognostic indicators for health be in the detection of these CTCs. Through the use care professionals. of immunofluorescence-staining protocols, various CTC subtypes can be identified through sophisticated Machine Learning algorithms written to interpret stained results. This Pathologists must be able to recognize morpholog- would create an objective and reproducible method ical patterns. This ability depends on training: the for subtyping and quantifying the CTCs. more slides pathologists review in their training, the For more information about CTCs, please review better they become. Through this training, patholo- the article written by Cáceres and colleagues in this gists learn to associate visual patterns with disorders. issue of Cancer Control. Pattern-recognition algorithms operate on a similar scheme, with recent advances showing that combin- Conclusions ing unsupervised (implicit) and supervised (explicit) In the early era of personalized medicine, digital learning can train algorithms to learn statistical regu- pathology will continue to advance, provide precise larities for nearly any data. Therefore, the application prognostic information, and help guide treatment de- of machine learning to digital pathology is a natural cisions.15,16 Digital image analysis is likely to improve fit and offers potential for advancing the field. the assessment of predictive biomarkers in tissue sec- The Ki67 proliferation index is a valuable tool tions and overcome the challenges associated with for evaluating the progression of neuroendocrine subjective analyses. With such a significant interest tumors and predicting therapeutic response. Modal- in immunotherapy, digital image analysis is likely to ities to quantify the Ki67 proliferation index, the provide solutions to better assess the tissue microen- ratio of the number of immunopositive tumor cells vironment for immune cell components, much like to all tumor cells, are subjective and imprecise. To digital analysis for in situ flow cytometry. create an objective and reproducible quantification The potential of digital pathology is vast. In process, Xing et al13 developed an automatic Ki67 the field of cytopathology, automated digital an- counting method. Their algorithm creates a “dictio- alyzers exist for reviewing Papanicolaou staining; nary” of visual features of cells and then uses those however, they are costly and their use is limited, features to identify cells in images and subsequently thus making analyzers less cost effective than cyto- count them with high precision.13 Their method technologists.17 However, with the rapid increases performs comparably to the manual count of the in computing power and concomitant decreases in pathologist, beating out any existing method that cost, it may be only a matter of time before such utilizes machine learning.13 analyzers become more common. Mitotic counts have been a staple in the field of Beck et al4 demonstrated that automated mor- pathology for determining the prognosis of different phometric analyses can play a major role in teaching

140 Cancer Control April 2015, Vol. 22, No. 2 medical professionals about important prognostic morphological features, but will such analyses become cost effective enough for routine clinical use? No doubt these tools will continue to further our understanding of disease and improve the abilities of pathologists to make better informed decisions related to prognosis and treatment. In addition, as machine learning advances, so does the field of digital pathology. In the future, software may be able to handle large throughput volume and highly complex, mul- tidimensional data, thus assisting pathologists with reducing diagnostic errors and rendering objective, accurate, and reproducible diagnoses, all of which are likely to improve clinical care and decrease the cost of health care.

References 1. McCabe A, Dolled-Filhart M, Camp RL, et al. Automated quantitative analysis (AQUA) of in situ protein expression, antibody concentration, and prognosis. J Natl Cancer Inst. 2005;97(24):1808-1815. 2. Gustavson M, Dolled-Filhart M, Christiansen J, et al. Aqua technology and molecular pathology. In: JS Platero, ed. Molecular Pathology in Drug Discovery and Development. Hoboken, NJ: John Wiley & Sons; 2009:295-334. 3. Camp RL, Chung GG, Rimm DL. Automated subcellular localization and quantification of protein expression in tissue microarrays. Nat Med. 2002;8(11):1323-1327. 4. Beck AH, Sangoi AR, Leung S, et al. Systematic analysis of breast cancer morphology uncovers stromal features associated with survival. Sci Transl Med. 2011;3(108):108ra113. 5. Definiens. Definians and Metimark enter multi-year deal for prostate cancer prognostic test. Published November 20, 2013. http://www.definiens. com/about-definiens/news-room/press-releases/view/article/definiens-and -metamark-enter-multi-year-deal-for-prostate-cancer-prognostic-test.html. Accessed April 14, 2015. 6. Mandelbrot B. How long is the coast of Britain? Statistical self-similarity and fractional dimension. Science. 1967;156(3775):636-638. 7. Fractal Foundation Web site. http://fractalfoundation.org/resources /what-are-fractals. Accessed April 16, 2015. 8. Zmeškal O, Veselý M, Nežádal M, et al. Fractal analysis of image structures. http://www.fch.vutbr.cz/lectures/imagesci/download_ejournal/01 _O.Zmeskal.pdf. Accessed April 14, 2015. 9. Lopes R, Betrouni N. Fractal and multifractal analysis: a review. Med Image Anal. 2009;13(4):634-649. 10. Tambasco M, Eliasziw M, Magliocco AM. Morphologic complexity of epithelial architecture for predicting invasive breast cancer survival. J Transl Med. 2010;8:140. 11. Tambasco M, Costello BM, Kouznetsov A, et al. Quantifying the architectural complexity of microscopic images of histology specimens. Micron. 2009;40(4):486-494. 12. Tambasco M, Magliocco AM. Relationship between tumor grade and computed architectural complexity in breast cancer specimens. Hum Pathol. 2008;39(5):740-746. 13. Xing F, Su H, Neltner J, et al. Automatic Ki-67 counting using robust cell detection and online dictionary learning. IEEE Trans Biomed Eng. 2014;61(3):859-870. 14. Ciresan DC, Giusti A, Gambardella LM, et al. Mitosis detection in breast cancer histology images with deep neural networks. Med Image Comput Comput Assist Interv. 2013;16(pt 2):411-418. 15. Słodkowska J, Rojo MG. Digital pathology in personalized cancer therapy. Folia Histochem Cytobiol. 2011;49(4):570-578. 16. Guzman M, Judkins AR. Digital pathology: a tool for 21st century neuropathology. Brain Pathol. 2009;19(2):305-316. 17. Bengtsson E, Malm P. Screening for cervical cancer using automated analysis of PAP-smears. Comput Math Methods Med. 2014;2014:842037.

April 2015, Vol. 22, No. 2 Cancer Control 141 New technologies for molecular analysis

are being routinely used in diagnostics.

The Chair Stands Alone. Photograph courtesy of Craig Damlo. www.soapboxrocket.com.

Molecular Technologies in the Clinical Diagnostic Laboratory Victor E. Zota, MD, and Anthony M. Magliocco, MD

Background: New technologies for molecular analysis are increasing our ability to diagnose cancer. Methods: Several molecular analysis technologies are reviewed and their use in the clinical laboratory is discussed. Results: Select key technologies, including polymerase chain reaction and next-generation sequencing, are helping transform our ability to analyze cancer specimens. As these technological advances become more and more incorporated into routine diagnostic testing, our classification systems are likely to be impacted and our approach to treatment transformed. The routine use of such technology also brings challenges for analysis and reimbursement. Conclusion: These advances in technology will change the way we diagnose, monitor, and treat patients with cancer.

Introduction molecular subcategories. Insights gained from deep Molecular analysis is transforming the way health care sequencing helps us understand tumor evolution and professionals approach cancer. Fundamentally, cancer the emergence of new subclones that may be resistant is primarily a disease of DNA. Driver mutations drive to therapy. Applying methods of molecular analysis to clonal proliferation. Some cancers acquire disrupted cancer treatment is likely to become routine practice, genomes, often as a consequence of defective DNA and the information derived from such analysis can repair processes that may be either inherited or ac- then help guide classification and treatment selection. quired. Some of these driver mutations present at- Furthermore, advances in trace DNA measurements tractive targets for therapy, such as mutant EGFR and will allow the molecular monitoring of patients and BRAF, whereas others may predict a more aggressive the effective analysis of tumor progression, thus en- course of cancer, such as mutations in p53. The clas- abling the creation and application of more adaptive sification of cancer is trending away from histomor- therapies for patients. phological determinants and toward functional and Polymerase Chain Reaction From the Department of Anatomic Pathology, H. Lee Moffitt Cancer Polymerase chain reaction (PCR) has revolutionized Center & Research Institute, Tampa, Florida. molecular biology. This single advance enabled the Submitted February 18, 2015; accepted April 10, 2015. trace amounts of DNA to be amplified so as to en- Address correspondence to Anthony M. Magliocco, MD, Department able other techniques, such as sequencing, to be per- of Anatomic Pathology, Moffitt Cancer Center, 12902 Magnolia formed. The technology was based on the discovery Drive, Tampa, FL 33612. E-mail: [email protected] of a thermally stable polymerase enzyme taq isolated No significant relationships exist between the authors and the companies/organizations whose products or services may be referenced from Thermus aquaticus, which can live at tempera- in this article. tures approaching the boiling point in hot streams.1

142 Cancer Control April 2015, Vol. 22, No. 2 The key aspect of this thermally stable enzyme is that then amplified by PCR, fluorescently labeled, and read it survives repeated episodes of heating and cooling via automated droplet flow cytometry. Each droplet and can still carry out DNA polymerase activity. PCR is assigned a positive or negative value based on the is based on sequentially and exponentially denaturing rate of fluorescent intensity. The number of positive double-stranded DNA and synthesizing new strands and negative droplets is read via flow cytometry and in target areas defined by “primers” to exponentially is used to calculate the concentration of an allele. amplify areas of interest in each round. Typically, 20 to An advantage of digital PCR is low DNA template 30 rounds of DNA amplification cycles are performed requirements. In the era of personalized medicine, this that enable the massive amplification of a target.2 fact is of particular importance because laboratories Specific areas for targeted amplification are defined are provided with minimal amounts of tissue triaged by the primer sequence. Consequently, the target and for diagnostic purposes and molecular characteriza- flanking sequences must be known and the primers tion. In addition, digital PCR typically does not require are artificially synthesized. a preamplification step for increasing the abundance PCR has numerous key applications, including of all sequences of interest and does not alter the identifying rare sequences (eg, presence of viral DNA relative abundances of these sequences. However, in a tissue sample) or amplifying a target gene so unbiased amplification is difficult to achieve; thus, that a sufficient quantity can be applied for standard PCR biases are sometimes introduced.4,5 sequencing methods. PCR has improved following Clinical Application: Digital PCR can be used the introduction of real-time or quantitative PCR for any testable DNA sequence that can be detected (qPCR), which incorporates a probe that either emits with standard PCR; digital PCR also has the advan- or quenches a signal after each round of PCR. This tage of target sequence quantification, which can be modification enables the direct monitoring of the pro- used to detect rare variants in a population of DNA gression of PCR, typically by photoelectrical means to molecules and to estimate the frequency of a variant precisely determine the quantity of the starting target sequence or the relative copy number of separate and product. This method is valuable for quantitatively sequences in template DNA.6-8 determining targets such as the number of viruses or Digital PCR is likely to have widespread use for the mutational load in a biosample. screening and the monitoring of minimal residual disease. Its limitation is that a specific mutation or Digital Polymerase Chain Reaction target sequence must be known in advance, and a In 1999, Volgelstein and Kinzler3 introduced the term digital PCR assay must be designed for this specific “digital PCR,” which is a modification of the original requirement. PCR method in which the reactions are partitioned into subcompartments and the reaction is carried to Detecting Clinically Significant Rare Variants completion. The goal is to dilute the specimen to the The genetic makeup of tumors is heterogeneous, point where 1 target is present in each compartment, meaning that tumors are composed of various cell thus enabling precise, single-target quantification with populations with distinct mutations.9 Although most either a positive or negative result for each partition somatic mutations are present in all tumor cells, ad- (hence the term “digital”). This analytical method is ditional mutations are acquired with every tumor cell based on the principles of the Poisson distribution. division and, if that mutation confers a survival advan- Oftentimes, the compartmentalization is accomplished tage, then distinct subclones will evolve. The ability by creating an emulsion. This method has improved to detect the presence of targeted therapy–resistant accuracy rates for sensitivity and specificity of the subclones in a tumor could alter our approach to treat- reaction and is capable of detecting rare targets. ment. It may be possible to accurately predict which Droplet-digital PCR is qPCR in a droplet. In each tumors are intrinsically resistant to therapy prior to droplet is a mini-PCR reactor that contains a single treatment commencement.10 It may also be possible DNA molecule, primers, Mg2+ molecules, dinucleo- to identify tumors at risk for developing resistance tides, a polymerase molecule, and a fluorescent-la- during the course of treatment.4,11 beled, allele-specific detection probe. qPCR functions Several known key “actionable” mutations are on the same principle as PCR, but a fluorescently linked to therapies approved by the US Food and labeled detection probe is added to the process. The Drug Administration (FDA). These include mutations probe incorporates a 5' reporter dye and a 3' nonflu- of EGFR, PIK3CA, and KRAS.5,12,13 Digital PCR may orescent quencher and is designed to be complemen- help detect these key actionable mutations in biofluid tary to a specific allele (wild-type or mutant). During specimens such as plasma or urine, leading to a more the amplification process, DNA polymerase removes rapid diagnosis and initiation of treatment when tis- the quencher from the probe, thus unmasking the flo- sue is lacking; in addition, the use of digital PCR may rescent signal. A single DNA molecule in a droplet is also help us accurately monitor treatment response

April 2015, Vol. 22, No. 2 Cancer Control 143 through the quantitative assessment of mutational Sanger Sequencing load in the biofluid.5,12-15 Molecular pathology owes its existence to our ability With its sensitivity rate, digital PCR enables the to sequence DNA. Specific sequence information is tracking of individual tumor clones that may result in a prerequisite for designing the primers and probes tailoring therapies to delay disease recurrence. Within used in diagnostic tests. The DNA sequencing method the era of personalized medicine, it is insufficient to originally developed by Sanger et al24 is the basis for detect mutant alleles alone. Instead, one should quan- most DNA sequencing performed in clinical laborato- tify the allele burden to suggest the clinically relevant ries. It is also the backbone technology that enabled threshold at which targeted therapies are the most the sequencing of the human genome. effective and subsequently modify therapies to avoid Sanger et al24 used the principles of DNA repli- or delay the emergence of resistance. One example cation to develop the process now known as Sanger of this approach is the quantification of KRAS-mutant dideoxy sequencing, which is a method based on a alleles in metastatic colorectal cancer. KRAS muta- combination of DNA polymerase reaction primed with tions are predictive of nonresponse to anti–epidermal specific sequencing primers, along with either radio- growth factor receptor (EGFR) therapies in metastatic active nucleotide or dye termination. The products of colorectal cancer, but 50% of patients without the the reaction are electrophoretically analyzed, which mutations benefit from these therapies.16 When us- enables a direct sequence to be determined.25 Detec- ing digital PCR for the quantification of the mutant tion software generates an electropherogram of the KRAS allele, a reverse correlation was seen between DNA sequence, correlating each rate of fluorescence the proportion of mutated DNA and the frequency of intensity of each dye to a specific dideoxynucleotide anti-EGFR response (P < .001).16 In addition, patients triphosphate migration time. Capillary electrophoresis with fewer than 1% of mutant KRAS alleles have sim- facilitated the widespread incorporation of sequenc- ilar progression-free and overall survival rates than ing and fragment analysis into the clinical laboratory. those with wild-type KRAS tumors.16 Pyrosequencing is a method that allows the rap- Evidence is encouraging that the supreme sensi- id sequence determination of short DNA fragments. tivity of digital PCR could be exploited as a screening The reaction is based on detecting the release of the tool to detect cancer in earlier stages. For example, the pyrophosphate (PPi) byproduct after the nucleotide detection of KRAS, APC, and TP53 mutations in fecal, is incorporated into the growing DNA chain. Several urine, or blood samples may be a strong indicator of enzymes, as well as DNA polymerase, are required, an underlying colonic malignancy.3,6,7,17,18 including adenosine triphosphate (ATP) sulfurylase, Digital PCR can also be used to estimate copy luciferase, and apyrase. After primer hybridization, number variation (CNV), which is a known mech- deoxynucleotide triphosphates (dNTPs) are sequen- anism employed by cancer cells to alter gene ex- tially added to the growing chain by DNA polymerase, 19 pression. and this process results in the release of PPi. Because Digital PCR is expected to revolutionize the mon- dNTPs are individually added to the reaction, their in- itoring of minimal residual disease for diseases with a corporation can be determined based on the presence key molecular indicator, such as the BCR-ABL fusion or absence of PPi. Therefore, for each dNTP incorpo- 20 in chronic myelogenous leukemia. In addition, the rated, an equimolar quantity of PPi is detected. The technology may be able to monitor the emergence enzyme apyrase is used to degrade unincorporated of resistance clones to key targeted therapies, such dNTPs and excess ATP prior to the next dNTP. Detec- as EGFR T790M mutations in non–small-cell lung tion is based on the conversion of PPi to ATP by the cancer, which may predate the emergence of thera- enzyme ATP sulfurylase. The resulting ATP provides py-resistant metastasis.18,21,22 These applications form the energy required for luciferase to convert luciferin the basis of the “liquid biopsy,” enabling the real-time to oxyluciferin, thus generating visible light, which monitoring of the evolution and progression of the is detected using a charge-coupled device camera. tumor as well as the emergence of therapy resistance. The sequence is then displayed. Access to real-time data may lead to the consideration of adaptive treatment designs and the more rapid Mass Spectrometry for DNA Mutational discontinuation of potentially toxic and medically Analysis futile treatments.21 The MassARRAY System (Sequenom, San Diego, Cal- Researchers have demonstrated that detecting ifornia) capitalizes on the previous work of many a mutant allele at a ratio of 1:100,000 is possible scientists.26,27 In 1988, Karas and Hillenkamp28 first when using a microdroplet system.5,23 This techno- reported the use of an organic molecule as a matrix logical advance may lead to opportunities to study to assist desorption/ionization of other small mole- minimal residual disease and real-time responses cules under ultraviolet laser irradiation. At the same to therapeutic interventions. time, Tanaka et al29 showed that analytes can be ion-

144 Cancer Control April 2015, Vol. 22, No. 2 ized using a mixture of metal powder and glycerol. tions.40 This system is used in many clinical labora- This method became known as soft ionization. This tories because of its high sensitivity and quantitative advance enabled the technique to be applied to larg- aspects.41 It is also flexible in the input samples and er biomolecules.30 Karas et al31 devised a method of can analyze DNA derived from frozen tissue, forma- matrix-assisted laser desorption/ionization (MALDI), lin-fixed paraffin-embedded tissue, fresh tissue, and and, subsequently, Tanaka et al29 showed that coupling cell lines.42 It can also efficiently and cost effectively MALDI to a time-of-flight (TOF) mass analyzer allowed survey numerous known mutation targets in thou- the detection of macromolecules (in particular, pro- sands of samples.41 Data analysis is performed using teins). This method of MALDI-TOF mass spectrometry Typer Analyzer (Sequenom), an analysis software with (MS) for the analysis of macromolecules created new customizable parameters that can be adjusted on de- opportunities for applying MS to biomedical research. sired probabilities, including mutation frequency. In MALDI-MS, a sample is embedded in the crys- talline matrix and deposited on a conductive sample Digital Molecular Barcode Technologies support. A nanosecond laser beam then disrupts the The NanoString Protocol (NanoString Technologies, sample embedded in a matrix.28 The resulting ions are Seattle, Washington) quantifies nucleic acid molecules then manipulated by an electrical field.32 The ions are and is frequently applied to the quantification of ri- accelerated and sent in a vacuum flight tube where bonucleic acid (RNA) expression. The system uses a they are separated according to their speed, which combination of isolation probes to capture targets is related to the mass:charge ratio. Spectra are then of interest coupled with a detection system with a analyzed and the ionic masses are isolated based on directly observable signal generated by an attached TOF. 33 The spectra produced by this method are gen- reporter bar code that can be optically observed by a erally simple and do not require excessive processing scanner system. This technology directly isolates and for interpretation.33 quantifies the target molecules without the need for In 1992, Nordhoff et al34 first demonstrated the an amplification step. This is an advantage because use of MALDI-TOF MS to detect and measure the amplification biases and errors can be avoided and masses of nucleic acids. The most commonly used targets can be digitally and precisely counted so that method for MALDI-TOF MS is detection and quan- measurements of expression are accurate over several tification of single-base, primer-extension products orders of magnitude.43 The technology utilizes a simi- for qualitative and quantitative analysis of DNA cop- lar approach to measure messenger RNA expression ies containing single nucleotide polymorphisms by and DNA CNVs.44,45 Up to 800 regions of interest can MALDI-TOF MS.35 This method is applied to a select- be studied with 1 probe mix/1 hybridization.46 ed region, which is amplified by PCR along with a The Prosigna (NanoString Technologies) assay uses single-base primer extension reaction, enabling the a proprietary algorithm to determine risk for breast resolution of 4 different base results.36 The data are cancer recurrence using multiplex gene expression presented in Sequenom proprietary software. determination.47-50 The assay is based on the Prediction The single-base primer extension assay can be ap- Analysis of Microarrays (PAM50; NanoString Technol- plied to the diagnosis and screening of any patholog- ogies) assay, which uses 50 genes to classify breast ical conditions known to have mutations in a specific cancer into intrinsic subgroups and calculates risk gene or a specific set of genes easily identified and of recurrence at 10 years. It is available in Europe quantified by this method.37 This assay was used to and has received clearance by the FDA. The assay detect and quantify the frequency of EGFR-activating can be used for estrogen receptor–positive breast mutations in the tissue of a patient with non–small-cell cancer in postmenopausal women who have fewer lung cancer to predict response to tyrosine kinase in- than 3 positive axillary lymph nodes. The assay has hibitors.38 This is a powerful method for resequencing similar performance characteristics to Oncotype DX projects as it overcomes many of the limitations of (Genomic Health, Redwood City, California), Mam- gel capillary–based electrophoretic methods because mostrat (Clarient, Aliso Viejo, California), and the im- of its resolution. It is also ideal for detecting frame- munohistochemical 4 score.51 shift or single-base polymorphisms (single nucleotide For various subtypes of leukemia, the nCounter polymorphisms).39 Leukemia Fusion Gene Expression Assay Kit (NanoS- tring Technologies) detects fusion genes resulting Somatic Mutation Analysis from balanced translocations.52 In addition, the as- Several multigene panels have also been developed say detects 25 fusion genes and the expression of for the MassARRAY System, and these panels sur- 23 biomarkers, including TP53, RB1, NRAS, EVI1, vey select key oncogenes for selected mutations. For MN1, MLLT11, FLT3, and WT1, among others.52 The example, LungCarta (Agena Bioscience, San Diego, kit can be broadly applied to DNA extracted from California) surveys 26 genes for 214 different muta- select tissues and specimens, including formalin-fixed,

April 2015, Vol. 22, No. 2 Cancer Control 145 paraffin-embedded samples as well as from blood uses semiconductor sequencing technology. After the and bone marrow biopsy specimens. The test is cost nucleotides are incorporated into the DNA molecules effective, less sensitive to the quality of RNA, and less via polymerase, a proton is released, resulting in min- challenging in data analysis, thus making it preferable ute changes in pH recognized by the PGM, which in for the screening of fusion events with known break turn interprets whether a nucleotide has been add- points as well as for examining CNV. ed or not. The biofluidic semiconductor chip is se- NanoString Technologies is also working on ways quentially flooded with one nucleotide after another, to mark antibodies to enable the capture and quan- resulting in voltage spikes when incorporation has tification of targeted proteins.53 occurred. If a dinucleotide or polynucleotide repeat is observed, then the voltage spike will be double in Next-Generation Sequencing the target microscopic well.57,58 Because the PGM does Until recently, molecular diagnostic laboratories re- not require fluorescence and camera scanning for sig- lied on low-throughput molecular and cytogenetics nal detection, the instrument is small in size and can methods designed to interrogate the most frequent sequence at high speeds and at a low cost per base. mutations (hot-spot mutations) in cancer. The meth- The major limitation of semiconductor sequencing is a odology limited the number of mutations or target relatively small sequencing length (200 bp) and poor regions tested in an assay. With the growing number nucleotide resolution in long homopolymeric DNA of mutations and genes being discovered in a variety regions. The Ion Torrent PGM can also be customized of cancers, such as lung, breast, melanoma, and brain with an Ion Ampliseq (Life Technologies) panel of cancers, molecular diagnostic laboratories have be- 200 to 300 actionable genes (single nucleotide vari- gun to implement next-generation sequencing (NGS) ant, selected translocation, gene amplifications) and to interrogate numerous mutations in a timely and is being used for eligibility testing for the Molecular cost-effective manner. Analysis for Therapy Choice program of the National Several major NGS technologies are available, and Cancer Institute.59 sequencing by synthesis technology from Illumina Massively parallel (next-generation) DNA se- (San Diego, California) dominates the market. Its NGS quencing, combined with declining sequencing costs, technology uses solid-based amplification of small as well as increased technical feasibility and maturing DNA fragments and sequentially identified composing to support the bioinformatics infrastructure, is being bases detected during synthesis of a complementary used in hospital laboratories because the technolo- DNA strand. Performing NGS is not dissimilar from gy cost-effectively evaluates multiple targets for the other sequencing technologies in that the procedure presence of mutations in a variety of specialties, in- requires the standard steps of nucleic acid isolation, cluding oncology, genetics, infectious diseases, and performance of the assay, followed by detailed data others.60-62 Targeted panels are the most suited for the analysis.54 The first step is DNA isolation followed by clinical application of NGS assays and have improved fragmentation and enrichment of the target genes,54 the diagnostic yields of molecular testing. However, producing a set of short DNA fragments (100–500 base for a new assay to become effectively utilized in the pairs) flanked by oligonucleotide adapters during the health care system, it must have high rates of sensi- library generation process.55 The adapters are comple- tivity, accuracy, precision, and specificity, and it must mentary to immobilized oligonucleotides on a flow demonstrate effective clinical utility. cell, and the resultant libraries are subjected to a clonal So, does NGS make a difference in patient care? amplification step prior to sequencing.56 The propri- Tests have been developed that rapidly analyze tu- etary technology utilizes bridge amplification to form mor DNA to detect hundreds of variants that may template clusters on a flow cell. The company has drive cell growth and to provide clues about treatment developed proprietary sequencing by synthesis chem- options.62 In addition, the sequencing of circulating istry in which fragments or “clusters” are sequenced complementary DNA has detected chromosomal ab- base by base by adding fluorescent-labeled termina- normalities associated with tumors, suggesting that tor nucleotides optically imaged after each step. The the approach may be a viable option for the non- process is repeated until the fragment is analyzed invasive detection of cancer.63 Although NGS tech- by the required reaction cycles. After the sequencing nology can effectively identify key driver mutations, platform generates the sequencing images, the data such as activating mutations in EGFR in lung cancer, are analyzed in 5 steps: (1) image analysis, (2) base thus serving as a reliable guide for selecting tyrosine calling, (3) Bcl conversion, (4) sequence alignment, kinase inhibitor therapy,12 its use is still early for on- and (5) variant analysis and counting. cologists as the whole. For example, these advanced Another new technology is Ion Torrent (Life Tech- sequencing technologies bring new challenges to the nologies, South San Francisco, California). The Ion molecular diagnostic laboratory, including questions Personal Genome Machine (PGM; Life Technologies) about how to manage complex and large data sets,

146 Cancer Control April 2015, Vol. 22, No. 2 validating bioinformatics pipelines, and learning how therapy might have a beneficial effect on the disease to interpret the data to make clinically meaningful in that particular individual, thus creating significant treatment recommendations.64-66 challenges in how to apply these results. Ideally, pa- The College of American Pathologists has rec- tients with newly identified mutations would agree to ognized these challenges and is working to create participate in “basket” or “match” trials to ensure that proficiency testing for NGS clinical laboratories that the effects of treatment are recorded so as to increase focuses on determining laboratory performance in our knowledge base for future patients.59,73 both the “wet” and “dry” laboratory components of NGS clinical analysis.67 Tumor Heterogeneity In addition to addressing tissue requirements, NGS Making Sense of the Cancer Genome technology helps understand the phenomenon of tu- As we are becoming technologically proficient and mor heterogeneity. Intratumoral heterogeneity refers capable of generating complex analyses of genomic to biological differences between malignant cells orig- changes in cancerous tumors, a major challenge still inating within the same tumor. During tumorigene- exists to understand how such changes contribute to sis, the cells acquire necessary driver mutations and tumor biology or behavior.68 We must also learn how many somatic genetic alterations that do not seem to exploit these changes in the tumor to improve treat- to confer selective advantage (passengers). During ment strategies with targeted therapy agents. Existing tumor expansion, unstable tumor genomes give rise successes in personalized medicine capitalized on the to substantial diversified tumor cell populations.74 In- genomic knowledge accumulated from decades of tratumoral heterogeneity in driver mutations and in research68; however, to continue on the path of per- loci implicated in therapeutic resistance leads to clin- sonalized medicine, we must streamline the methods ical heterogeneity in tumor response to treatment.74-77 of clinical validation of newly discovered targets. A challenge facing us is the identification of these The scientific and medical communities must de- driver mutations for diagnostic purposes and targeted velop strategies to manage the torrent of emerging therapy development. This is complicated by the fact information regarding the genetic underpinnings of that classifying mutations into passengers and drivers cancer and the potentially targetable and actionable is context-specific and, as tumors change over time, mutations now routinely detected in the hospital lab- the selective value of a given mutation changes.74 In oratory. This has created a situation in which our addition, mutations in driver genes are often differ- knowledge of the biological makeup of cancer is far ent, thus altering different codons. For example, the ahead of high-quality evidence about how to use such KRAS G12D and KRAS G13D mutations do not appear knowledge to select therapy. Numerous efforts are to have the same clinical implications; in fact, most ongoing to collect the key evidence linking clinical activating mutations may have different effects on behavior to detectable genetic biomarkers observed cancer cells.9,48,78,79 in cancers.69,70 Genetic heterogeneity provides the substrate for Foundation Medicine realized the utility of NGS as fueling tumor evolution during tumor progression a clinical tool and in understanding the basic mecha- and therapeutic resistance.74 Target therapy based on nisms of the evolution of cancer. It brought to market founder mutations may eradicate the primary tumor, Foundation One (Foundation Medicine, Cambridge, but most patients with complete responses invariably Massachusetts), a Clinical Laboratory Improvement relapse. One example of this is the activating mutation Amendments–validated NGS for solid tumors. This of BCR-ABL in chronic myelogenous leukemias and test is a comprehensive genomic profile of 315 genes KIT in gastrointestinal stromal tumors, which tend to linked to the pathogenesis of solid tumors.71 The com- develop additional mutations, leading to loss of the pany also developed FoundationOne Heme (Foun- ability of the drug to bind to its target.80,81 dation Medicine) for hematological malignances and Tumors may also escape targeted therapy by ac- sarcomas. It is an expanded panel of 405 genes and tivating survival pathways such as those observed in translocations at 265 loci.72 Sarcomas and hematolog- BRAF V600E–mutant melanomas treated with vemu- ical malignancies are known for the presence of gene rafenib.82 Resistance often develops via activation of fusions that result in the formation of oncogenes that the survival pathway via the receptor tyrosine kinase are diagnostic, prognostic, and predictive.72 Thus, the or the RAS-mediated reactivation of the mitogen-acti- results of both tests can be used to inform the health vated protein kinase pathway.83 In lung cancer treated care team about clinically actionable mutations and with EGFR inhibitors, resistance is developed by ac- to help guide treatment options for patients. tivation of alternative prosurvival signaling pathways However, mutations are frequently detected that via MET amplification, as well as acquisition of the suggest a possible targeted therapy for a specific pa- EGFR T790M mutation, which affects the binding of tient, but specific evidence may be missing that the tyrosine kinase inhibitors.74,84

April 2015, Vol. 22, No. 2 Cancer Control 147 Cytotoxic therapies such as cisplatinum applied creasingly challenged in managing complex genetic to cancer may also result in the selective destruc- information. A typical large panel analysis can reveal tion of sensitive clones and the selection of resistant dozens of different mutations, including classical driv- clones, thus driving tumor evolution.85,86 Therefore, ers and other alterations of lesser-known consequence. the genetic heterogeneity of a tumor is an import- Many of these may be “passenger mutations” or inher- ant determinant of therapeutic outcome that requires ited polymorphisms of uncertain consequence. Some, comprehensive tumor characterization before, during, such as inherited germline missense mutations, will and after the initiation of therapy. result in defective tumor suppressors, such as BRCA or TP53, and may define cancer family syndromes, Classification whereas others may have less certain meaning. Some Historically, disease classification schemes in pathol- have suggested that large-panel NGS should be run ogy were based on the histological and morpholog- on both tumor and germline samples from patients ical appearance of the neoplasm and the tissue of to help identify which mutations are associated with origin. Advanced molecular analysis technologies somatic tumors and which ones are inherited.90 have created the opportunity to augment traditional classification systems with molecular features, such Clinical Implementation as the nature of the driver mutation (eg, presence of The development and rapid widespread adoption of mutant BRAF in malignant melanoma) that might as- low-cost NGS into the clinical laboratory have changed sociate a specific subtype of disease with a particular the way we think about cancer. The technology has etiology, risk factor, behavior, or treatment option. also challenged pathologists and other health care Similarly, molecular classification using messen- professionals to carefully think about how to interpret ger RNA expression patterns have altered the way data and incorporate them into meaningful treatment breast cancer is evaluated, resulting in a molecular planning.91-93 The main challenge facing many labo- subdivision of classification that defines new mo- ratories is how to validate the analytical and clinical lecular and biological subtypes, including luminal performance of these advanced tests.91,94-96 NGS tech- types A and B, human epidermal growth factor re- nology also poses major challenges for regulatory ceptor 2, and basal subtypes. Recognition of these agencies.97 The rapid adoption of these technologies types has led to the creation of new diagnostics and is outpacing the abilities of governmental agencies has altered the way clinical trials are considered to develop effective monitoring strategies.98,99 Can- and designed. The development of trials focusing cers are dynamic, heterogenous entities that harbor a on molecular features, rather than tissues of origin, diverse range of genetic alterations that drive tumor are highlighted by “basket” and “match” designs that progression and passenger mutations. In addition, tu- enroll participants based on the presence of specif- mors evolve biological subclones that possess various ic mutational driver mutations.87 However, caution phenotypes over time and in response to therapy. The should be exercised because some cancers with diversity of this biology is fundamentally challenging activating driver mutations may not equivalently the way we gather evidence, because it makes classical respond to a targeted treatment. An example of this clinical trial design difficult, if not impossible (eg, con- is the use of v-raf murine sarcoma viral oncogene ducting a randomized blinded trial for each mutation homolog B (BRAF) inhibitors in colorectal cancer is not feasible). Consequently, future approaches may displaying activating mutations of BRAF. Although rely on “basket” or “match” designs in which a large BRAF inhibitors appear to inhibit melanomas with set of therapeutic options can be utilized against a BRAF-activating mutations, disappointing results range of underlying driver molecular targets.59 were seen in study volunteers with BRAF-mutated colorectal cancer.88 This failure has been attribut- Investment Uncertainties ed to the bypass of the BRAF-signaling system by The implementation of personalized medicine testing the EGFR-signaling pathway in colorectal cancer, has also been hampered by the reimbursement land- a pathway not active in melanoma.88 Consequent- scape.98 For example, Current Procedural Terminology ly, health care professionals must use care when codes identify surgical, medical, and diagnostic ser- using molecular data across tumor types because vices used by insurance companies for reimbursement our knowledge of cancer biology is limited; using of services rendered. Genetic sequencing technology single markers may be inadequate to define specific is advancing faster than it can be incorporated into the treatment regimens.88,89 health care delivery system, and this is evidenced by the lack of specific Current Procedural Terminology Variants of Unknown Significance codes for genetic testing. With the vast amount of information produced from To recoup the capital investment in equipment NGS, molecular pathologists and oncologists are in- and procedural development, some laboratories use a

148 Cancer Control April 2015, Vol. 22, No. 2 Working Group: standardization for use in the clinical trial setting. J Clin Oncol. billing technique known as code “stacking” to recover 2008;26(6):983-994. costs.98 The problem with a cost model for reimburse- 13. Normanno N, Tejpar S, Morgillo F, et al. Implications for KRAS status and EGFR-targeted therapies in metastatic CRC. Nat Rev Clin Oncol. ment is that no premium is placed on innovation or 2009;6(9):519-527. clinical utility as would be apparent in a value-driven 14. Wang J, Ramakrishnan R, Tang Z, et al. Quantifying EGFR alterations in the lung cancer genome with nanofluidic digital PCR arrays. Clin Chem. system, a model used by the pharmaceutical industry 2010;56(4):623-632. to set drug prices.98,100,101 Reimbursement of diagnos- 15. Wang Z, Chen R, Wang S, et al. Quantification and dynamic monitoring of EGFR T790M in plasma cell-free DNA by digital PCR for prognosis of tics is set on a case-by-case basis by contractors of EGFR-TKI treatment in advanced NSCLC. PloS One. 2014;9(11):e110780. the Centers for Medicare & Medicaid Services. In ad- 16. Laurent-Puig P, Pekin D, Normand C, et al. Clinical relevance of KRAS-mutated subclones detected with picodroplet digital pcr in addition, some decisions regarding pricing have resulted vanced colorectal cancer treated with anti-EGFR therapy. Clin Cancer Res. in fees set at prices so inadequate that laboratories 2015;21(5):1087-1097. 17. Singer G, Oldt R 3rd, Cohen Y, et al. Mutations in BRAF and KRAS cannot cover the cost of such testing, a fact that has characterize the development of low-grade ovarian serous carcinoma. J Natl led to the bankruptcy and closure of some molecular Cancer Inst. 2003;95(6):484-486. 102 18. Diehl F, Schmidt K, Durkee KH, et al. Analysis of mutations in DNA laboratories in the United States. isolated from plasma and stool of colorectal cancer patients. Gastroenterology. It is clear that, in this era of advancing technology 2008;135(2):489-498. 19. Stranger BE, Forrest MS, Dunning M, et al. Relative impact of nucle- and novel treatment options, molecular pathologists otide and copy number variation on gene expression phenotypes. Science. and oncologists must closely work with patients and 2007;315(5813):848-853. 20. Oehler VG, Qin J, Ramakrishnan R, et al. Absolute quantitative detec- payers to develop a strategy to enable a sustainable tion of ABL tyrosine kinase domain point mutations in chronic myeloid leuke- provision of molecular testing technology, along with mia using a novel nanofluidic platform and mutation-specific PCR. Leukemia. 2009;23(2):396-399. a system to reward innovation and assess proper value 21. Yung TK, Chan KC, Mok TS, et al. Single-molecule detection of epider- to these analytical technologies. mal growth factor receptor mutations in plasma by microfluidics digital PCR in non-small cell lung cancer patients. Clin Cancer Res. 2009;15(6):2076-2084. 22. Li N, Ma J, Guarnera MA, et al. Digital PCR quantification of miR- Conclusions NAs in sputum for diagnosis of lung cancer. J Cancer Res Clin Oncol. 2014;140(1):145-150. We are living in a revolutionary time in the field 23. Heyries KA, Tropini C, Vaninsberghe M, et al. Megapixel digital PCR. of diagnostic pathology. The rapid introduction of Nat Methods. 2011;8(8):649-651. 24. Sanger F, Nicklen S, Coulson AR. DNA sequencing with chain-termi- next-generation sequencing and other high-complexi- nating inhibitors. Proc Natl Acad Sci U S A. 1977;74(12):5463-5467. ty testing technologies are altering our understanding 25. Applied Biosystems. DNA Sequencing by Capillary Electrophoresis. 2nd ed. Foster City, CA: Applied Biosystems; 2009. http://www.griffith.edu. of cancer and our capability for detecting, monitoring, au/__data/assets/pdf_file/0007/160369/DNA-Sequencing-Chemistry-Guide. and selecting treatment. This updated knowledge is pdf. Accessed April 20, 2015. 26. Jurinke C, van den Boom D, Cantor CR, et al. Automated genotyping likely to impact our classification systems and trans- using the DNA MassArray technology. Methods Mol Biol. 2001;170:103-116. form our approach to cancer treatment. Thus, it is 27. Jurinke C, van den Boom D, Cantor CR, et al. Automated genotyping using the DNA MassArray technology. Methods Mol Biol. 2002;187:179-192. vital that we learn how to effectively and sustain- 28. Karas M, Hillenkamp F. Laser desorption ionization of proteins with ably deploy and interpret the data from these new molecular masses exceeding 10,000 daltons. Anal Chem. 1988;60:2299-2301. 29. Tanaka K, Waki H, Ido Y, et al. Protein and polymer analyses up to m/z technologies to produce the most benefit for patients 100,000 by laser ionization time-of-flight mass spectrometry. Rapid Commun with cancer. Mass Spectrom. 1988;2(8):151-153. 30. Takeda N, Pomerantz SC, McCloskey JA. Detection of ribose-methylated nucleotides in enzymatic hydrolysates of RNA by thermospray liquid chromatography-mass spectrometry. J Chromatogr. 1991;562(1-2):225-235. References 31. Karas M, Bachmann D, Hillenkamp F. Influence of the wavelength 1. Saiki RK, Gelfand DH, Stoffel S, et al. Primer-directed enzymat- in high-irradiance ultraviolet laser desorption mass spectrometry of organic ic amplification of DNA with a thermostable DNA polymerase. Science. molecules. Analytical Chemistry. 1985;57(14):2935-2939 1988;239(4839):487-491. 32. Scheper T, Hoheisel JD, eds. Chip Technology. Volume 77. Berlin: 2. Green MR, Sambrook J. Molecular Cloning: A Laboratory Manual. 4th Springer-Verlag; 2002. ed. Cold Spring Harbor, NY: Cold Spring Harbor Laboratory Press; 2012. 33. Ahmad F, Babalola OO. Application of mass spectrometry as rapid 3. Vogelstein B, Kinzler KW. Digital PCR. Proc Natl Acad Sci U S A. detection tool in plant nematology. Appl Spectr Rev. 2014;19(1):1-10. 1999;96(16):9236-9241. 34. Nordhoff E, Ingendoh A, Cramer R, et al. Matrix-assisted laser desorp- 4. Sanders R, Huggett JF, Bushell CA, et al. Evaluation of digital PCR for tion/ionization mass spectrometry of nucleic acids with wavelengths in the absolute DNA quantification. Anal Chem. 2011;83(17):6474-6484. ultraviolet and infrared. Rapid Commun Mass Spectrom. 1992;6(12):771-776. 5. Day E, Dear PH, McCaughan F. Digital PCR strategies in the develop- 35. Braun A, Little DP, Köster H. Detecting CFTR gene mutations by ment and analysis of molecular biomarkers for personalized medicine. Meth- using primer oligo base extension and mass spectrometry. Clin Chem. ods. 2013;59(1):101-107. 1997;43(7):1151-1158. 6. McCaughan F, Dear PH. Single-molecule genomics. J Pathol. 36. Edenberg HJ, Liu Y. Laboratory methods for high-throughput genotyp- 2010;220(2):297-306. ing. Cold Spring Harb Protoc. 2009;4(11):1-9. 7. Hindson BJ, Ness KD, Masquelier DA, et al. High-throughput droplet 37. Farkas DH, Miltgen NE, Stoerker J, et al. The suitability of matrix assist- digital PCR system for absolute quantitation of DNA copy number. Anal Chem. ed laser desorption/ionization time of flight mass spectrometry in a laboratory 2011;83(22):8604-8610. developed test using cystic fibrosis carrier screening as a model. J Mol Diagn. 8. Zhong Q, Bhattacharya S, Kotsopoulos S, et al. Multiplex digital PCR: 2010;12(5):611-619. breaking the one target per color barrier of quantitative PCR. Lab Chip. 38. Su KY, Chen HY, Li KC, et al. Pretreatment epidermal growth factor 2011;11(13):2167-2174. receptor (EGFR) T790M mutation predicts shorter EGFR tyrosine kinase in- 9. Vogelstein B, Papadopoulos N, Velculescu VE, et al. Cancer genome hibitor response duration in patients with non-small-cell lung cancer. J Clin landscapes. Science. 2013;339(6127):1546-1558. Oncol. 2012;30(4):433-440. 10. Marin JJ, Sanchez de Medina F, Castaño B, et al. Chemoprevention, 39. Mullen MP, Creevey CJ, Berry DP, et al. Polymorphism discovery and chemotherapy, and chemoresistance in colorectal cancer. Drug Metab Rev. allele frequency estimation using high-throughput DNA sequencing of tar- 2012;44(2):148-172. get-enriched pooled DNA samples. BMC Genomics. 2012;13:16. 11. Navin N, Krasnitz A, Rodgers L, et al. Inferring tumor progression from 40. Agena Bioscience. Lungcarta panel. http://agenabio.com/lungcar- genomic heterogeneity. Genome Res. 2010;20(1):68-80. ta-panel. Accessed May 4, 2015. 12. Eberhard DA, Giaccone G, Johnson BE, et al. Biomarkers of response 41. Quinn AM, Hickson N, Adaway M, et al. Diagnostic mutation profiling to epidermal growth factor receptor inhibitors in Non-Small-Cell Lung Cancer and validation of non-small-cell lung cancer small biopsy samples using a

April 2015, Vol. 22, No. 2 Cancer Control 149 high throughput platform. J Thorac Oncol. 2015;10(5):784-792. 70. Weinstein JN, Collisson EA, Mills GB, et al; Cancer Genome Atlas 42. Agena Bioscience. High-throughput somatic mutation profiling for lung Research Network. The Cancer Genome Atlas Pan-Cancer analysis project. cancer research. http://agenabio.com/sites/default/files/51-20013R1.0-Lung- Nature Genet. 2013;45(10):1113-1120. Carta-Flyer_WEB.pdf. Accessed May 4, 2015. 71. Johnson DB, Dahlman KH, Knol J, et al. Enabling a genetically informed 43. Geiss GK, Bumgarner RE, Birditt B, et al. Direct multiplexed mea- approach to cancer medicine: a retrospective evaluation of the impact of com- surement of gene expression with color-coded probe pairs. Nat Biotechnol. prehensive tumor profiling using a targeted next-generation sequencing panel. 2008;26(3):317-325. Oncologist. 2014;19(6):616-622. 44. Textoris J, Loriod B, Benayoun L, et al. An evaluation of the role of gene 72. FoundationOne. Technical information and test overview. http://founda- expression in the prediction and diagnosis of ventilator-associated pneumonia. tionone.com/docs/FM_TechnicalOverview_HEM-I-001-20150105.pdf. Accessed Anesthesiology. 2011;115(2):344-352. May 7, 2015. 45. Brahmachary M, Guilmatre A, Quilez J, et al. Digital genotyping of 73. Lopez-Chavez A, Thomas A, Rajan A, et al. Molecular profiling macrosatellites and multicopy genes reveals novel biological functions as- and targeted therapy for advanced thoracic malignancies: a biomark- sociated with copy number variation of large tandem repeats. PLoS Genet. er-derived, multiarm, multihistology phase II basket trial. J Clin Oncol. 2014;10(6):e1004418. 2015;33(9):1000-1007. 46. Geiss GK, Bumgarner RE, Birditt B, et al. Direct multiplexed mea- 74. Marusyk A, Almendro V, Polyak K. Intra-tumour heterogeneity: a looking surement of gene expression with color-coded probe pairs. Nat Biotechnol. glass for cancer? Nat Rev. 2012;12(5):323-334. 2008;26(3):317-325. 75. Gerlinger M, Rowan AJ, Horswell S, et al. Intratumor heterogeneity 47. Nielsen TO, Parker JS, Leung S, et al. A comparison of PAM50 intrin- and branched evolution revealed by multiregion sequencing. N Engl J Med. sic subtyping with immunohistochemistry and clinical prognostic factors in 2012;366(10):883-892. tamoxifen-treated estrogen receptor-positive breast cancer. Clin Cancer Res. 76. Anderson K, Lutz C, van Delft FW, et al. Genetic variegation of clonal 2010;16(21):5222-5232. architecture and propagating cells in leukaemia. Nature. 2011;469(7330):356-361. 48. Chen J, Ye Y, Sun H, et al. Association between KRAS codon 13 muta- 77. Ruiz C, Lenkiewicz E, Evers L, et al. Advancing a clinically rele- tions and clinical response to anti-EGFR treatment in patients with metastatic vant perspective of the clonal nature of cancer. Proc Natl Acad Sci U S A. colorectal cancer: results from a meta-analysis. Cancer Chemother Pharmacol. 2011;108(29):12054-12059. 2013;71(1):265-272. 78. Stolze B, Reinhart S, Bulllinger L, et al. Comparative analysis of KRAS 49. Nielsen T, Wallden B, Schaper C, et al. Analytical validation of the codon 12, 13, 18, 61, and 117 mutations using human MCF10A isogenic cell PAM50-based Prosigna Breast Cancer Prognostic Gene Signature Assay lines. Sci Rep. 2015;5:8535. and nCounter Analysis System using formalin-fixed paraffin-embedded breast 79. Kodaz H, Hacibekiroglu I, Erdogan B, et al. Association between spe- tumor specimens. BMC Cancer. 2014;14:177. cific KRAS mutations and the clinicopathological characteristics of colorectal 50. NanoString Technologies. Prosigna Website overview. http://prosigna. tumors. Mol Clin Oncol. 2015;3(1):179-184. com/overview. Published 2015. Accessed April 20, 2015. 80. Deininger M. Resistance to imatinib: mechanisms and management. 51. Dowsett M, Sestak I, Lopez-Knowles E, et al. Comparison of PAM50 risk J Natl Compr Canc Netw. 2005;3(6):757-768. of recurrence score with oncotype DX and IHC4 for predicting risk of distant 81. Corless CL, Heinrich MC. Molecular pathobiology of gastrointestinal recurrence after endocrine therapy. J Clin Oncol. 2013;31(22):2783-2790. stromal sarcomas. Annu Rev Pathol. 2008;3:557-586. 52. NanoString Technologies. nCounter: leukemia fusion gene expres- 82. Spagnolo F, Ghiorzo P2, Orgiano L3, et al. BRAF-mutant melanoma: sion assay. Published July 18, 2011. http://www.nanostring.com/media/pdf treatment approaches, resistance mechanisms, and diagnostic strategies. /PDS_nCounter_Leukemia_Fusion.pdf. Accessed April 20, 2015. Onco Targets Ther. 2015;8:157-168. 53. Ullal AV, Peterson V, Agasti SS, et al. Cancer cell profiling by barcoding 83. Nazarian R, Shi H, Wang Q, et al. Melanomas acquire resis- allows multiplexed protein analysis in fine-needle aspirates. Science Transl tance to B-RAF(V600E) inhibition by RTK or N-RAS upregulation. Nature. Med. 2014;6(219):219ra9. 2010;468(7326):973-977. 54. Botella LM, Albiñana V, Ojeda-Fernandez L, et al. Research on potential 84. Engelman JA, Zejnullahu K, Mitsudomi T, et al. MET amplification leads biomarkers in hereditary hemorrhagic telangiectasia. Front Genet. 2015;6:115. to gefitinib resistance in lung cancer by activating ERBB3 signaling. Science. 55. Rehm HL, Bale SJ, Bayrak-Toydemir P, et al; Working Group of the 2007;316(5827):1039-1043. American College of Medical Genetics and Genomics Laboratory Quality 85. Diaz LA Jr, Williams RT, Wu J, et al. The molecular evolution of ac- Assurance Commitee. ACMG clinical laboratory standards for next-generation quired resistance to targeted EGFR blockade in colorectal cancers. Nature. sequencing. Genet Med. 2013;15(9):733-747. 2012;486(7404):537-540. 56. Grada A, Weinbrecht K. Next-generation sequencing: methodology 86. Wagle N, Emery C, Berger MF, et al. Dissecting therapeutic resis- and application. J Invest Dermatol. 2013;113(8):e11. tance to RAF inhibition in melanoma by tumor genomic profiling. J Clin Oncol. 57. Spisák S, Guttman A. From chemical genomics to chemical proteom- 2011;29(22):3085-3096. ics: the power of microarray technology. In: Darvas F, Guttman A, Dormán G. 87. Willyard C. ‘Basket studies’ will hold intricate data for cancer drug ap- Chemical Genomics and Proteomics. 2nd ed. Boca Raton, FL: CRC Press; provals. Nat Med. 2013;19(6):655. 2013:81-122. 88. Prahallad A, Sun C, Huang S, et al. Unresponsiveness of colon can- 58. Flusberg BA, Webster DR, Lee JH, et al. Direct detection of DNA cer to BRAF(V600E) inhibition through feedback activation of EGFR. Nature. methylation during single-molecule, real-time sequencing. Nat Methods. 2012;483(7387):100-103. 2010;7(6):461-465. 89. Stites EC. The response of cancers to BRAF inhibition underscores 59. Abrams J, Conley B, Mooney M, et al. National Cancer Institute’s Pre- the importance of cancer systems biology. Sci Signal. 2012;5(246):pe46. cision Medicine Initiatives for the new National Clinical Trials Network. Am 90. Jones S, Anagnostou V, Lytle K, et al. Personalized genomic anal- Soc Clin Oncol Educ Book. 2014:71-76. yses for cancer mutation discovery and interpretation. Sci Transl Med. 60. Craig DW, O’Shaughnessy JA, Kiefer JA, et al. Genome and tran- 2015;7(283):283ra53. scriptome sequencing in prospective metastatic triple-negative breast cancer 91. US Department of Health and Human Services; US Food and Drug uncovers therapeutic vulnerabilities. Mol Cancer Ther. 2013;12(1):104-116. Administration. Paving the Way for Personalized Medicine: FDA’s Role in a 61. Liang WS, Craig DW, Carpten J, et al. Genome-wide characterization New Era of Medical Product Development. Silver Spring, MD: US Depart- of pancreatic adenocarcinoma patients using next generation sequencing. ment of Health and Human Services. Published October 2013. http://www. PLoS One. 2012;7(10):e43192. fda.gov/downloads/ScienceResearch/SpecialTopics/PersonalizedMedicine/ 62. Frampton GM, Fichtenholtz A, Otto GA, et al. Development and vali- UCM372421.pdf. Accessed April 20, 2015. dation of a clinical cancer genomic profiling test based on massively parallel 92. Patel JN, Mandock K, McLeod HL. Clinically relevant cancer biomarkers DNA sequencing. Nat Biotechnol. 2013;31(11):1023-1031. and pharmacogenetic assays. J Oncol Pharm Pract. 2014;20(1):65-72. 63. Leary RJ, Sausen M, Kinde I, et al. Detection of chromosomal alter- 93. MacConaill LE, Van Hummelen P, Meyerson M, et al. Clinical imple- ations in the circulation of cancer patients with whole-genome sequencing. mentation of comprehensive strategies to characterize cancer genomes: op- Sci Transl Med. 2012;4(162):162ra154. portunities and challenges. Cancer Discov. 2011;1(4):297-311. 64. Deverka PA, Dreyfus JC. Clinical integration of next generation se- 94. Mao M, Tian F, Mariadason JM, et al. Resistance to BRAF inhibition in quencing: coverage and reimbursement challenges. J Law Med Ethics. BRAF-mutant colon cancer can be overcome with PI3K inhibition or demeth- 2014;42(suppl 1):22-41. ylating agents. Clin Cancer Res. 2013;19(3):657-667. 65. Javitt GH, Carner KS. Regulation of next generation sequencing. J Law 95. Mattocks CJ, Morris MA, Matthijs G, et al. A standardized framework Med Ethics. 2014;42(suppl 1):9-21. for the validation and verification of clinical molecular genetic tests. Eur J Hum 66. Yu B. Setting up next-generation sequencing in the medical laboratory. Genet. 2010;18(12):1276-1288. Methods Mol Biol. 2014;1168:195-206. 96. Botkin JR, Teutsch SM, Kaye CI, et al. Outcomes of interest in evi- 67. Gargis AS, Kalman L, Berry MW, et al. Assuring the quality of dence-based evaluations of genetic tests. Genet Med. 2010;12(4):228-235. next-generation sequencing in clinical laboratory practice. Nature Biotechnol. 97. US Food and Drug Administration. Framework for regulatory over- 2012;30(11):1033-1036. sight of laboratory developed tests; draft guidance for industry, Food and 68. Chin L, Andersen JN, Futreal PA. Cancer genomics: from discovery Drug Administration staff, and clinical laboratories; availability. Fed Regist. science to personalized medicine. Nat Med. 2011;17(3):297-303. 2014;79(192):59776-59779. 69. Fenstermacher DA, Wenham RM, Rollison DE, et al. Implementing 98. Caffrey MK. When science outpaces payers: molecular diagnostics personalized medicine in a cancer center. Cancer J. 2011;17(6):528-536. pose reimbursement dilemmas. OncLive. Published September 17, 2104.

150 Cancer Control April 2015, Vol. 22, No. 2 http://www.onclive.com/publications/oncology-live/2014/august-2014/when-science-outpaces-payers-molecular-diagnostics-pose-reimbursement-dilemmas. Accessed May 7, 2015. 99. California Healthcare Institute (CHI). A New Era for Diagnostics: Per- sonalized Healthcare and the Future of Medicine. La Jolla, CA: CHI. http://www. chi.org/uploadedFiles/CHI_Diagnostics_White_Paper_12-10.pdf. Accessed May 7, 2015. 100. Fugel HJ, Nuijten M, Faulkner E. The application of economics concepts to stratified medicine--use of health economics data to support market access for stratified medicine interventions. J Med Econ. 2014;17(5):305-311. 101. Fugel HJ, Nuijten M, Postma M. Stratified medicine and reimbursement issues. Front Pharmacol. 2012;3:181. 102. Burns J. Low 2013 molecular rates may bankrupt some labs. Dark Rep. 2013;1-6. http://www.darkreport.com/PDF/Low-2013-Molecular-Test-Rates- May-Bankrupt-Some-Pathology-Labs.pdf. Accessed May 4, 2015.

April 2015, Vol. 22, No. 2 Cancer Control 151 Molecular markers refine

the cytopathological diagnosis

of thyroid neoplasms.

Home Is Where the Heart Is. Photograph courtesy of Craig Damlo. www.soapboxrocket.com.

Molecular Assays in Cytopathology for Thyroid Cancer Pablo Valderrabano, MD, Victor E. Zota, MD, Bryan McIver, MD, PhD, Domenico Coppola, MD, and Marino E. Leon, MD

Background: Despite lack of adequate, validated, independently performed clinical studies, several molecular tests are commercially available on the market and are being used on indeterminate thyroid nodules to guide patient-care decisions. Methods: We summarize the current evidence on the role and limitations of molecular tests used in combination with thyroid cytopathology to refine the presurgical diagnosis of thyroid nodules. Results: The clinical performance of molecular tests depends on the pretest risk of malignancy within the specific cytological group being assessed. This risk is variable and should be assessed at each institution to optimize the selection of the molecular test and the interpretation of its results. Next-generation sequencing has increased the sensitivity of oncogene panels while maintaining high specificity. Tests assessing the gene expression pattern have shown promising results, with high sensitivity but low specificity. The impacts of molecular markers on clinical practice remains in flux and their effect on health care costs remains poorly understood. Conclusions: Further large, independent, confirmatory, clinical validation studies and real-world, cost-effectiveness studies are necessary before the widespread adoption of these tests can be endorsed as standard of care.

Introduction cytology provides valuable information in the presur- The incidences of benign thyroid nodules and thyroid gical evaluation of thyroid nodules, approximately cancer have increased in the last several decades.1 25% of biopsy samples do not render diagnostic infor- An accurate presurgical diagnosis of thyroid nodules mation and are classified as indeterminate.2 Repeating is important because of the implications in clinical FNA on an indeterminate specimen may be helpful at management. Although fine needle aspiration (FNA) times; however, for many patients, diagnostic surgery is eventually needed to clarify the diagnosis. One-third From the Endocrine Tumor Program (PV, BM), Departments of of these nodules resected for diagnostic purposes will Head and Neck (PV, BM), Endocrine Oncology (PV, BM), and An- prove to be malignant and, in many cases, will require atomic Pathology (VEZ, DC, MEL), H. Lee Moffitt Cancer Center & 3,4 Research Institute, Tampa, Florida. additional surgery to complete thyroidectomy. Con- Submitted September 4, 2014; accepted March 13, 2015. versely, two-thirds of such surgeries might have been 3 Address correspondence to Marino E. Leon, MD, Department of avoided with a more accurate presurgical diagnosis. Anatomic Pathology, Moffitt Cancer Center, 12902 Magnolia Drive, The Bethesda System for Reporting Thyroid Cy- Tampa, FL 33612. E-mail: [email protected] topathology (Bethesda) has standardized reporting Dr Valderrabano acknowledges financial support from the Alfonso terminology and diagnostic criteria, but it has not Martín Escudero Foundation. No other significant relationships 5 exist between the authors and the companies/organizations whose improved the performance of FNA. Indeterminate products or services may be referenced in this article. specimens are stratified by Bethesda in 3 diagnostic

152 Cancer Control April 2015, Vol. 22, No. 2 categories according to risk of malignancy (ROM): During recent years, our understanding and knowl- • Category III: Atypia/follicular lesion of unde- edge regarding driver mutations have been expanding. termined significance A recent publication by The Cancer Genome Atlas • Category IV: Follicular neoplasm/suspicious (TCGA) has significantly reduced the proportion of for follicular neoplasm and Hürthle cell neo- PTCs with unknown driver mutations from 25% to plasm/suspicious for Hürthle cell neoplasm 3.5%.17 Such mutations are almost always mutually • Category V: Suspicious for malignancy The estimated ROMs of Bethesda categories III to V Table. — Common Mutations and Rearrangements are 5% to 15%, 15% to 30%, and 60% to 75%, respective- in Common Thyroid Tumors 5 ly. However, the observed ROMs in the indeterminate Tumor Type Mutation and Rearrangement categories, particularly Bethesda categories III and IV, Papillary thyroid carcinoma BRAF V600E (classic, tall cell variant) are broader, ranging from 0% to 48% and 14% to 49%, TERT 2,3 respectively, in multiple large institutional studies. RET/PTC1 (RET/CCDC6) This diagnostic challenge reflects subtle mor- RET/PTC3 (RET/ELE1) phological differences and significant morphological RET/PTC6 (RET/HTIF1) overlap in the cytological and histological features Papillary thyroid carcinoma EIF1AX exhibited by thyroid follicular pattern lesions, in- (follicular variant) BRAF K601E cluding follicular hyperplasia, adenomatous nodule/ NRAS follicular thyroid adenoma, follicular thyroid carci- HRAS noma, and the follicular variant of papillary thyroid KRAS carcinoma (PTC).6,7 The subjectivity of the patholog- PAX8/PPARγ ical diagnosis further complicates this issue. Broad intraobserver and interobserver variabilities have Follicular thyroid carcinoma PAX8/PPARγ been reported for follicular thyroid lesions in cyto- NRAS logical and histological specimens.7-10 Several mo- HRAS lecular tests have been developed in an attempt to KRAS better and more reliably characterize these lesions BRAF K601E to avoid diagnostic surgery. These tests use different TSHR approaches to identify molecular signatures specif- Follicular thyroid carcinoma TP53 (oncocytic variant)/Hürthle cell HRAS ic for thyroid cancer. Some scrutinize the DNA to carcinoma detect mutations and others characterize the gene KRAS expression profiles at the transcriptional (messenger PTEN ribonucleic acids [RNAs]) or post-transcriptional level Poorly differentiated NRAS (microRNAs). All these methods have limitations and thyroid carcinoma PIK3CA no single test is 100% accurate. Furthermore, the GNAS predictive values of these tests — which represent RAF V600E their true clinical value — depend on the cytological Undifferentiated thyroid TP53 ROM, which represents the pretest probability of ma- carcinoma (anaplastic BRAF V600E carcinoma) lignancy; as the ROM increases, so does the positive NRAS predictive value (PPV) at the expense of a reduction in PIK3CA the negative predictive value (NPV), and vice versa.11 PTEN Therefore, until further standardization is achieved in CTNNB1 the classification of indeterminate cytology, the results Medullary thyroid carcinoma RET obtained at one center may not extrapolate to others. HRAS Marketing a test on the basis of a fixed NPV, without KRAS first knowing the true pretest ROM, is disingenuous. Follicular thyroid adenoma PAX8/PPARγ In addition, some of these molecular markers, such as NRAS the point mutation BRAF V600E, are also considered KRAS prognostic factors, and their study may be helpful for Benign thyroid TSHR individualizing clinical management.12-15 “hyperplastic nodule” NRAS HRAS Somatic Alterations KRAS Most thyroid cancers exhibit somatic point mutations PTEN or rearrangements that activate the mitogen-activat- GNAS ed protein kinase (MAPK) and phosphatidylinositol 3 kinase/protein kinase B pathways (Table16-21).22 Data from references 16 to 21.

April 2015, Vol. 22, No. 2 Cancer Control 153 exclusive and homogeneously present within the tu- tion sequencing is now commercially available as mor.17,18 Most well-differentiated thyroid cancers exhibit ThyGenX (Interpace Diagnostics, Parsippany, New a single point mutation or chromosomal rearrangement, Jersey), but the performance of this modified assay although 2 or more mutations may be found in more has not yet been reported.37 aggressive tumors.17 Some mutations are highly specific The advent of next-generation sequencing has for thyroid cancer such as the point mutation BRAF allowed the rapid translation of new data on addi- V600E; other mutations, like RAS mutations, are also tional, low-prevalence, somatic alterations into the found in benign lesions, thus limiting their diagnostic clinical field. An additional commercially available specificity. However, this finding could be explained panel is based on this technology and is marketed as by the evolution hypothesis, which posits that the ac- ThyroSeq V2 (University of Pittsburgh Medical Center, cumulation of mutations induces progression from a Pittsburgh, Pennsylvania). This panel now includes polyclonal proliferation (hyperplasia) to a benign mono- more than 400 point mutations in 14 genes,38 includ- clonal proliferation (follicular thyroid adenoma) to a ing AKT1, BRAF, CTNNB1, GNAS, HRAS, KRAS, NRAS, differentiated thyroid cancer (PTC [follicular variant] or PIK3CA, PTEN, RET, TP53, TSHR, TERT, and EIF1AX, follicular thyroid carcinoma) and potentially from there along with 42 gene rearrangements and can be per- to undifferentiated thyroid cancer. If this concept holds formed on as little as 10 ng of DNA.20,39 In the initial true, then a RAS mutation present in follicular thyroid publication on this panel, the test included mutations adenoma would represent an early neoplastic change in 13 genes (rather than the 14 genes now in the com- that might progress to malignancy (precancer) rather mercial offering) and 42 gene rearrangements, and it than represent a false-positive result. achieved sensitivity and specificity of 90% (80%–99%) The presurgical identification of the oncogenic and 93% (88%–98%), respectively, among cytological driver mutation in thyroid cancer has been explored specimens classified as Bethesda category IV.19 Such for several years as a method to segregate thyroid performance should improve and simplify the inter- nodules with indeterminate cytology. Initial studies pretation (NPVs and PPVs) of the results among dif- searched for 1 or just a few specific mutations.23-28 If re- ferent institutions with distinct pretest ROM. The NPV searchers must focus on a single marker, then the most and PPV achieved in this study were 96% and 83%, useful mutation is BRAF V600E for several reasons: respectively, with an overall ROM of 27%.19 Based on (1) It is the most prevalent mutation among PTC (45%) the reported sensitivity and specificity, we calculate and PTC represents 85% of all thyroid malignancies, that the NPV will remain above 95% at any ROM up to (2) its specificity for malignancy is nearly 100%, and 34%, whereas the PPV will remain above 75% at any (3) it can independently predict extrathyroidal exten- ROM above 15%. As more mutations are added to this sion and lymph node involvement, thus implying a panel, we anticipate an increase in the test sensitivity, higher risk for disease persistence or recurrence that but it is possible that the specificity could decrease, could influence the extent of the surgery.29-32 However, thus negatively impacting the predictive values. Con- the sensitivity of BRAF for detecting cancer improves sequently, although these early results are promising, when other frequently occurring mutations are includ- they still require external validation and additional ed in the screening panel. Such an observation led to information regarding the performance of the test the development of oncogene panels that simultane- on Bethesda category III. Data on this category were ously studied several mutations. presented at the annual meeting and exposition of The most well-studied panel assessed 14 point the Endocrine Society in early 2015, suggesting that mutations in BRAF and RAS and for RET/PTC1, sensitivity and specificity are well preserved in this RET/PTC3, and PAX8/PPARγ rearrangements. This cytological group.40 panel has been prospectively studied in the characterization of indeterminate thyroid nodules achieving Gene Expression Analysis estimated sensitivity and specificity for malignancy Somatic mutations directly impact the gene expression around 60% and 98%, respectively.33-35 Results from profile by acting on signaling pathways. For example, the commercially available variant of this panel mar- BRAF V600E is associated with high MAPK signaling, keted as mIRinform (Asuragen, Austin, Texas) were whereas RAS mutants have lower MAPK signaling. published in a validation study that indicated the This observation was used by TCGA to develop a score panel had a slightly worse performance than antici- that quantified the extent to which the gene expres- pated, perhaps in part because of differences in the sion profile of each somatic mutation resembled either technology used in the commercial version of the BRAF V600E–mutant or RAS-mutant PTCs.17 This score test.36 In the study, the panel achieved sensitivity and is associated with the degree of thyroid differentiation specificity of 48% (29%–68%) and 89% (72%–98%), of the tumor, histological grade, American Thyroid respectively, for indeterminate thyroid nodules.36 Association risk of recurrence stage,4 and the MACIS A modified version of this test using next-genera- score.41 RAS-like tumors were better differentiated and

154 Cancer Control April 2015, Vol. 22, No. 2 BRAF V600E–like tumors were less differentiated; in Therefore, further validation is needed before the addition, RAS-mutant tumors were homogeneous in widespread use of this test. terms of gene expression and degree of thyroid dif- As single-gene markers, HMGA2 and cancer-relat- ferentiation.17 Conversely, BRAF V600E–mutant tumors ed E2 ubiquitin-conjugating enzyme have also been had a heterogeneous gene expression profile and studied in the cytology specimens of indeterminate included at least 4 different groups, a finding that thyroid nodules.45,46 However, neither has demonstrat- could explain the uncertainty regarding the prog- ed sufficiently high sensitivity or specificity to justify nostic and predictive power of this mutation alone.17 widespread use. Nonetheless, their combined use, In light of these results, the authors suggested that along with other markers, continues to be explored.47 PTCs may be more appropriately classified according The detection of thyrotropin receptor mRNA in pe- to their genetic profiles rather than to their mor- ripheral blood achieved better results in a study on phological appearance.17 Although these data have thyroid nodules with cytological diagnosis of follicular the potential to transform histological classification neoplasm, reaching sensitivity and specificity of 97% and clinical practice, their utility in routine clinical and 88%, respectively, for detecting thyroid cancer.48 practice has yet to be established. However, prior However, further studies are necessary to validate studies have already used gene expression profiles these results. (at the transcriptional or post-transcriptional level) to differentiate thyroid nodules with indeterminate Post-Transcriptional Level cytology; these are discussed below. MicroRNAs are small, noncoding RNA molecules that bind to mRNA to promote or silence their transla- Transcriptional Level tion into proteins.49-51 These molecules can be ana- Several tools exploit the analysis of the gene expres- lyzed in cytological specimens and those affecting sion profile at the transcriptional level to differentiate oncogenes or tumor suppressor genes are also be- benign from malignant lesions among indeterminate ing studied. Several studies have reported attempts thyroid nodules. Although several RNA-based markers to identify the specific signatures of thyroid cancer have investigated single genes, a commercially avail- by developing predictive models on excised spec- able multigene expression panel has achieved prom- imens later investigated as single microRNAs or in ising results. The Afirma Gene Expression Classifier microRNA panels on FNA samples of thyroid nodules.52-59 (Veracyte, South San Francisco, California) assesses One meta-analysis found that panels had better per- the relative mRNA expression of 167 genes processed formance rates compared with single microRNAs.59 through a support vector machine trained on a group The overall sensitivity and specificity of 4 studies of histological and FNA material derived from a broad using microRNA panels in FNA specimens were spectrum of samples from benign and malignant 87% (77%–92%) and 85% (78%–90%), respectively.59 thyroid disease. Using a proprietary algorithm, the Performance was best in a study validating a panel classifier reports the nodule as benign or suspicious. of 4 microRNAs (222, 328, 197, and 21) in which In a prospective, double-blind, multicenter study, sensitivity and specificity of 100% and 86%, respec- 265 indeterminate thyroid nodules were analyzed.42 tively, were achieved.55 Larger prospective studies of The overall performance of the test showed sen- indeterminate thyroid nodules must be performed to sitivity and specificity of 92% (84%–97%) and 52% validate their diagnostic utility. (44%–59%), respectively.42 In this study, the test Some microRNAs also appear to have prognostic achieved an NPV of approximately 95% for Bethesda value. In the study by TCGA, 4 different expres- categories III and IV. Such an NPV justifies patient ob- sion patterns stratified the BRAF-like PTCs into servation in lieu of surgery per the recommendations groups with various histological grades of differen- of the National Comprehensive Cancer Network.43 tiation.17 The overexpression of microRNA-21 and However, an NPV of 95% can be assumed only if the microRNA-146b, as well as the down-regulation midpoint sensitivity and specificity achieved in this of microRNA-204 — which acts as a tumor sup- single study are used and the ROM of the cytology pressor — was found in the group with the most is below 24%.11,42 Because the ROM in the indeter- aggressive PTCs.17 Conversely, higher expressions of minate categories is higher in many institutions,3 its microRNA-182 and microRNA-183 were associated with use should be individualized according to the specific RAS-like PTCs that had a higher degree of thyroid ROM of each center. In an independent study, the sen- differentiation.17,60 sitivity and specificity rates achieved by Afirma were reported to be 83% and 36%, respectively, and were the Follicular Neoplasm (Oncocytic Variant) best rates of the case scenarios.44 If lower sensitivity /Hürthle Cell Neoplasm and specificity than anticipated are confirmed, then Most of the molecular markers previously discussed both predictive values would be negatively affected. are not helpful for Hürthle cell neoplasms. Hürthle

April 2015, Vol. 22, No. 2 Cancer Control 155 cell carcinomas rarely exhibit one of the commonly relation study of thyroid nodule cytopathology and histopathology. Thyroid. 2011;21(3):243-251. studied mutations.61 However, the use of next-gen- 4. Cooper DS, Doherty GM, Haugen BR, et al. Revised American Thyroid eration sequencing might improve the sensitivity Association management guidelines for patients with thyroid nodules and differentiated thyroid cancer. Thyroid. 2009;19(11):1167-1214. of previous oncogene panels. The initial version of 5. Cibas ES, Ali SZ. The Bethesda system for reporting thyroid cytopa- ThyroSeq detected a mutation in 39% of cases of thology. Thyroid. 2009;19(11):1159-1165. 6. Suster S. Thyroid tumors with a follicular growth pattern: problems in Hürthle cell carcinoma analyzed.20 Most of them had differential diagnosis. Arch Pathol Lab Med. 2006;130(7):984-988. a TP53 mutation, typically detected in undifferen- 7. Elsheikh TM, Asa SL, Chan JK, et al. Interobserver and intraobserver variation among experts in the diagnosis of thyroid follicular lesions with borderline tiated thyroid cancer but not in well-differentiated nuclear features of papillary carcinoma. Am J Clin Pathol. 2008;130(5):736-744. carcinomas.20 Moreover, Hürthle cell neoplasms have 8. Gerhard R, da Cunha Santos G. Inter- and intraobserver reproducibility of thyroid fine needle aspiration cytology: an analysis of discrepant cases. a distinct profile of genes and transcription factors Cytopathology. 2007;18(2):105-111. involved in tumorigenesis.62 Possibly because of this, 9. Lloyd RV, Erickson LA, Casey MB, et al. Observer variation in the diagnosis of follicular variant of papillary thyroid carcinoma. Am J Surg Pathol. Afirma classifies 90% of the Hürthle cell neoplasms 2004;28(10):1336-1340. as “suspicious,” even despite the fact that the ROM in 10. Hirokawa M, Carney JA, Goellner JR, et al. Observer variation of encapsulated follicular lesions of the thyroid gland. Am J Surg Pathol. these tumors is typically lower than that for the overall 2002;26(11):1508-1514. Bethesda category IV group, resulting in a low PPV 11. McIver B. Evaluation of the thyroid nodule. Oral Oncol. 2013;49(7):645-653. 12. Kim TH, Park YJ, Lim JA, et al. The association of the BRAF(V600E) in these nodules.42,43,63 In part, this likely reflects the mutation with prognostic factors and poor clinical outcome in papillary thyroid fact that the training of the support vector machine cancer: a meta-analysis. Cancer. 2012;118(7):1764-1773. 13. Elisei R, Viola D, Torregrossa L, et al. The BRAF V600E mutation is an that encodes the proprietary algorithm included a independent, poor prognostic factor for the outcome of patients with low-risk number of Hürthle cell carcinomas but almost no intrathyroid papillary thyroid carcinoma: single-institution results from a large cohort study. J Clin Endocrinol Metab. 2012;97(12):4390-4398. benign Hürthle cell adenomas; as a consequence, the 14. Howell GM, Carty SE, Armstrong MJ, et al. Both BRAF V600E muta- gene expression classifier determines all Hürthle cell tion and older age (≥ 65 years) are associated with recurrent papillary thyroid cancer. Ann Surg Oncol. 2011;18(13):3566-3571. lesions as suspicious. 15. Adeniran AJ, Zhu Z, Gandhi M, et al. Correlation between genetic alter- Similarly, microRNA panels may inadequately ations and microscopic features, clinical manifestations, and prognostic characteristics of thyroid papillary carcinomas. Am J Surg Pathol. 2006;30(2):216-222. classify this group of tumors. In a study, the accura- 16. Radkay LA, Chiosea SI, Seethala RR, et al. Thyroid nodules with KRAS cy of the panel rose from 90% to 97% when Hürthle mutations are different from nodules with NRAS and HRAS mutations with regard to cytopathologic and histopathologic outcome characteristics. Cancer cell neoplasms were excluded.55 Thus, in our opin- Cytopathol. 2014;122(12):873-882. ion, Hürthle cell neoplasms should not be routinely 17. The Cancer Genome Atlas Research Network. Integrated genomic characterization of papillary thyroid carcinoma. Cell. 2014;159(3):676-690. evaluated with the currently available molecular tests, 18. Soares P, Trovisco V, Rocha AS, et al. BRAF mutations and RET/PTC and, if used, the test results should be interpreted rearrangements are alternative events in the etiopathogenesis of PTC. On- cogene. 2003;22(29):4578-4580. with caution. 19. Nikiforov YE, Carty SE, Chiosea SI, et al. Highly accurate diagnosis of cancer in thyroid nodules with follicular neoplasm/suspicious for a follicular neoplasm cytology by ThyroSeq v2 next-generation sequencing assay. Cancer. Conclusions 2014;120(23):3627-3634. The rapid translation into the clinical field of new 20. Nikiforova MN, Wald AI, Roy S, et al. Targeted next-generation sequencing panel (ThyroSeq) for detection of mutations in thyroid cancer. J Clin discoveries in the molecular basis of thyroid cancer Endocrinol Metab. 2013;98(11):E1852-E1860. has led to the development of several molecular tests 21. Kimura ET, Nikiforova MN, Zhu Z, et al. High prevalence of BRAF mutations in thyroid cancer: genetic evidence for constitutive activation of that address the deficiencies of thyroid cytopathology. the RET/PTC-RAS-BRAF signaling pathway in papillary thyroid carcinoma. The initial results of such tests are promising, and Cancer Res. 2003;63(7):1454-1457. 22. Xing M. Molecular pathogenesis and mechanisms of thyroid cancer. the rapid expansion of our knowledge in this regard Nat Rev Cancer. 2013;13(3):184-199. promises further advances in the near future. Addi- 23. Xing M, Tufano RP, Tufaro AP, et al. Detection of BRAF mutation on fine needle aspiration biopsy specimens: a new diagnostic tool for papillary tional studies that independently validate these results thyroid cancer. J Clin Endocrinol Metab. 2004;89(6):2867-2872. and define the role of these tests in routine clinical 24. Salvatore G, Giannini R, Faviana P, et al. Analysis of BRAF point mutation and RET/PTC rearrangement refines the fine-needle aspiration diagnosis of practice are needed. Until the cytological classifica- papillary thyroid carcinoma. J Clin Endocrinol Metab. 2004;89(10):5175-5180. tion has been further standardized, each institution 25. Cheung CC, Carydis B, Ezzat S, et al. Analysis of ret/PTC gene rearrangements refines the fine needle aspiration diagnosis of thyroid cancer. interested in using these tests should determine its J Clin Endocrinol Metab. 2001;86(5):2187-2190. own risk of malignancy for each of the indetermi- 26. Pizzolanti G, Russo L, Richiusa P, et al. Fine-needle aspiration molecular analysis for the diagnosis of papillary thyroid carcinoma through BRAF nate categories to select the most appropriate test V600E mutation and RET/PTC rearrangement. Thyroid. 2007;17(11):1109-1115. and to adequately interpret the results. However, in 27. Chung KW, Yang SK, Lee GK, et al. Detection of BRAFV600E mutation on fine needle aspiration specimens of thyroid nodule refines cyto-pathology the future, molecular markers will be instrumental in diagnosis, especially in BRAF600E mutation-prevalent area. Clin Endocrinol refining the diagnostic accuracy of cytology and to in- (Oxf). 2006;65(5):660-666. 28. Jin L, Sebo TJ, Nakamura N, et al. BRAF mutation analysis in fine needle dividualize thyroid cancer management and treatment. aspiration (FNA) cytology of the thyroid. Diagn Mol Pathol. 2006;15(3):136-143. 29. Joo JY, Park JY, Yoon YH, et al. Prediction of occult central lymph node metastasis in papillary thyroid carcinoma by preoperative BRAF analysis using References fine-needle aspiration biopsy: a prospective study. J Clin Endocrinol Metab. 1. Chen AY, Jemal A, Ward EM. Increasing incidence of differentiated thyroid 2012;97(11):3996-4003. cancer in the United States, 1988-2005. Cancer. 2009;115(16):3801-3807. 30. Xing M, Clark D, Guan H, et al. BRAF mutation testing of thyroid 2. Bongiovanni M, Spitale A, Faquin WC, et al. The Bethesda System for fine-needle aspiration biopsy specimens for preoperative risk stratification in reporting thyroid cytopathology: a meta-analysis. Acta Cytol. 2012;56(4):333-339. papillary thyroid cancer. J Clin Oncol. 2009;27(18):2977-2982. 3. Wang CC, Friedman L, Kennedy GC, et al. A large multicenter cor- 31. Howell GM, Nikiforova MN, Carty SE, et al. BRAF V600E mutation independently predicts central compartment lymph node metastasis in patients

156 Cancer Control April 2015, Vol. 22, No. 2 with papillary thyroid cancer. Ann Surg Oncol. 2013;20(1):47-52. 59. Zhang Y, Zhong Q, Chen X, et al. Diagnostic value of microRNAs in 32. Lin KL, Wang OC, Zhang XH, et al. The BRAF mutation is predictive discriminating malignant thyroid nodules from benign ones on fine-needle of aggressive clinicopathological characteristics in papillary thyroid microcar- aspiration samples. Tumour Biol. 2014;35(9):9343-9353. cinoma. Ann Surg Oncol. 2010;17(12):3294-3300. 60. Reddi HV, Driscoll CB, Madde P, et al. Redifferentiation and induction 33. Nikiforov YE, Steward DL, Robinson-Smith TM, et al. Molecular test- of tumor suppressors miR-122 and miR-375 by the PAX8/PPARγ fusion protein ing for mutations in improving the fine-needle aspiration diagnosis of thyroid inhibits anaplastic thyroid cancer: a novel therapeutic strategy. Cancer Gene nodules. J Clin Endocrinol Metab. 2009;94(6):2092-2098. Ther. 2013;20(5):267-275. 34. Ohori NP, Nikiforova MN, Schoedel KE, et al. Contribution of molecular 61. Filicori F, Keutgen XM, Buitrago D, et al. Risk stratification of indeter- testing to thyroid fine-needle aspiration cytology of “follicular lesion of undeter- minate thyroid fine-needle aspiration biopsy specimens based on mutation mined significance/atypia of undetermined significance”. Cancer Cytopathol. analysis. Surgery. 2011;150(6):1085-1091. 2010;118(1):17-23. 62. Máximo V, Lima J, Prazeres H, et al. The biology and the genetics of 35. Nikiforov YE, Ohori NP, Hodak SP, et al. Impact of mutational testing Hurthle cell tumors of the thyroid. Endoc Relat Cancer. 2012;19(4):R131-R147. on the diagnosis and management of patients with cytologically indeterminate 63. Harrell RM, Bimston DN. Surgical utility of afirma: effects of high cancer thyroid nodules: a prospective analysis of 1056 FNA samples. J Clin Endocrinol prevalence and oncocytic cell types in patients with indeterminate thyroid Metab. 2011;96(11):3390-3397. cytology. Endocr Pract. 2014;20(4):364-369. 36. Beaudenon-Huibregtse S, Alexander EK, Guttler RB, et al. Centralized molecular testing for oncogenic gene mutations complements the local cytopathologic diagnosis of thyroid nodules. Thyroid. 2014;24(10):1479-1487. 37. Interpace Diagnostics. Thyroid cancer diagnosis. http://www.interpa cediagnostics.com/hcpw.php. Accessed April 1, 2015. 38. CBLPath, Inc. ThyroSeq v.2 next generation sequencing. http://www. cblpath.com/products-and-services/test-menu-cblpath/item/1628-thyroseq-v2. Accessed April 28, 2015. 39. University of Pittsburgh Medical Center. ThyroSeq - thyroid cancer next-generation sequencing panel. http://mgp.upmc.com/Applications/mgp /Home/Test/ThyroSeq_Details. Accessed April 1, 2015. 40. Nikiforova MN. Next-gen sequencing for diagnosis, prognostication and treatment of thyroid cancer. Presented at: Endocrine Society 97th Annual Meeting & Expo; March 5–8, 2015; San Diego, CA. https://endo.confex.com/ endo/2015endo/webprogram/Paper18101.html. Accessed April 28, 2015. 41. Hay ID, Bergstralh EJ, Goellner JR, et al. Predicting outcome in papillary thyroid carcinoma: development of a reliable prognostic scoring system in a cohort of 1779 patients surgically treated at one institution during 1940 through 1989. Surgery. 1993;114(6):1050-1058. 42. Alexander EK, Kennedy GC, Baloch ZW, et al. Preoperative diagnosis of benign thyroid nodules with indeterminate cytology. N Engl J Med. 2012;367(8):705-715. 43. National Comprehensive Cancer Network. NCCN Clinical Practice Guidelines in Oncology. v2.2014. Published August 12, 2014. http://www.nccn.org/professionals/physician_gls/pdf/thyroid.pdf. Accessed April 13, 2015. 44. McIver B, Castro MR, Morris JC, et al. An independent study of a gene expression classifier (Afirma) in the evaluation of cytologically indeterminate thyroid nodules. J Clin Endocrinol Metab. 2014:jc20133584. 45. Jin L, Lloyd RV, Nassar A, et al. HMGA2 expression analysis in cytological and paraffin-embedded tissue specimens of thyroid tumors by relative quantitative RT-PCR. Diagn Mol Pathol. 2011;20(2):71-80. 46. Guerriero E, Ferraro A, Desiderio D, et al. UbcH10 expression on thyroid fine-needle aspirates. Cancer Cytopathol. 2010;118(3):157-165. 47. Prasad NB, Kowalski J, Tsai HL, et al. Three-gene molecular diagnostic model for thyroid cancer. Thyroid. 2012;22(3):275-284. 48. Milas M, Shin J, Gupta M, et al. Circulating thyrotropin receptor mRNA as a novel marker of thyroid cancer: clinical applications learned from 1758 samples. Ann Surg. 2010;252(4):643-651. 49. Calin GA, Dumitru CD, Shimizu M, et al. Frequent deletions and down-regulation of micro- RNA genes miR15 and miR16 at 13q14 in chronic lymphocytic leukemia. Proceed Natl Acad Sci U S A. 2002;99(24):15524-15529. 50. Calin GA, Liu CG, Sevignani C, et al. MicroRNA profiling reveals distinct signatures in B cell chronic lymphocytic leukemias. Proceed Natl Acad Sci U S A. 2004;101(32):11755-11760. 51. Ambros V. MicroRNA pathways in flies and worms: growth, death, fat, stress, and timing. Cell. 2003;113(6):673-676. 52. Kitano M, Rahbari R, Patterson EE, et al. Expression profiling of difficult-to-diagnose thyroid histologic subtypes shows distinct expression profiles and identify candidate diagnostic microRNAs. Ann Surg Oncol. 2011;18(12):3443-3452. 53. Mazeh H, Mizrahi I, Halle D, et al. Development of a microRNA-based molecular assay for the detection of papillary thyroid carcinoma in aspiration biopsy samples. Thyroid. 2011;21(2):111-118. 54. Kitano M, Rahbari R, Patterson EE, et al. Evaluation of candidate diagnostic microRNAs in thyroid fine-needle aspiration biopsy samples. Thyroid. 2012;22(3):285-291. 55. Keutgen XM, Filicori F, Crowley MJ, et al. A panel of four miRNAs accurately differentiates malignant from benign indeterminate thyroid lesions on fine needle aspiration. Clin Cancer Res. 2012;18(7):2032-2038. 56. Vriens MR, Weng J, Suh I, et al. MicroRNA expression profiling is a potential diagnostic tool for thyroid cancer. Cancer. 2012;118(13):3426-3432. 57. Shen R, Liyanarachchi S, Li W, et al. MicroRNA signature in thyroid fine needle aspiration cytology applied to “atypia of undetermined significance” cases. Thyroid. 2012;22(1):9-16. 58. Mazeh H, Levy Y, Mizrahi I, et al. Differentiating benign from malignant thyroid nodules using micro ribonucleic acid amplification in residual cells obtained by fine needle aspiration biopsy. J Surg Res. 2013;180(2): 216-221.

April 2015, Vol. 22, No. 2 Cancer Control 157 Molecular testing is entrenched

in the workup and management

of hematological malignancies.

Sam’s. Photograph courtesy of Craig Damlo. www.soapboxrocket.com.

Biomarkers in Hematological Malignancies: A Review of Molecular Testing in Hematopathology Mohammad Hussaini, MD

Background: Molecular interrogation of genetic information has transformed our understanding of disease and is now routinely integrated into the workup and monitoring of hematological malignancies. In this article, a brief but comprehensive review is presented of state-of-the-art testing in hematological disease. Methods: The primary medical literature and standard textbooks in the field were queried and reviewed to assess current practices and trends for molecular testing in hematopathology by disease. Results: Pertinent materials were summarized under appropriate disease categories. Conclusion: Molecular testing is well entrenched in the diagnostic and therapeutic pathways for hematological malignancies, with rapid growth and insights emerging following the integration of next-generation sequencing into the clinical workflow.

Introduction leukemia and lymphoma.1,2 Molecular testing has wide Perhaps in no other field of oncology is the routine applicability in hematopathology, guiding diagnosis use of molecular markers more integrated into the (eg, TCR gene rearrangement to establish T-cell clonal- diagnostic, prognostic, and therapeutic workup of ity), subclassification (eg, recurrent cytogenetic trans- disease as in the realm of hematological malignan- locations in acute myeloid leukemia [AML]), prognosis cies. Molecular diagnostics is a burgeoning field in (eg, Philadelphia chromosome–positive [Ph+] in acute the era of personalized medicine, with high-volume lymphoblastic leukemia [ALL]), and minimal residual laboratories running 10,000 molecular tests or more disease testing (eg, BCR-ABL transcripts in chronic every year, many of which are for the workup of myelogenous leukemia [CML]).

Myeloid Neoplasms From the Department of Hematopathology and Laboratory Myeloproliferative Neoplasms Medicine, H. Lee Moffitt Cancer Center & Research Institute, Tampa, Florida. Chronic Myelogenous Leukemia: The Ph chro- 3,4 Submitted July 15, 2014; accepted February 9, 2015. mosome in CML was discovered in 1960. t(9;22) Address correspondence to Mohammad Hussaini, MD, Department (q34;q11) juxtaposes most of ABL1 to 5' regions of of Hematopathology and Laboratory Medicine, Moffitt Cancer BCR, resulting in constitutively increased kinase ac- Center, 12902 Magnolia Drive, Tampa, FL 33612. tivity and neoplastic transformation.5 Although the E-mail: [email protected] breakpoint for ABL1 is mostly conserved, occurring No significant relationship exists between the author and the companies/organizations whose products or services may be in the intron preceding exon 2, the breakpoints in referenced in this article. BCR are more variable and typically occur in either

158 Cancer Control April 2015, Vol. 22, No. 2 the major or minor breakpoint regions (M- or m-bpr). and atypical CML in 59% of patients,16 and these findings M-bpr fusions result in a p210 fusion protein, which is were subsequently documented in all World Health Or- the form typically found in Ph+ CML. A sizeable num- ganization (WHO)–defined CNL cases, possibly prompt- ber of patients with Ph+ B-cell acute lymphoblastic ing a future revision of the WHO diagnostic criteria.17 leukemia (B-ALL; 40% of adults and 10% of children6) Polycythemia Vera, Essential Thrombocytosis, also harbor the p210 product. Conversely, m-bpr trans- and Primary Myelofibrosis: JAK2 codes for an in- locations result in a p190 fusion found in most Ph+ tracellular tyrosine kinase and provides signaling for B-ALL cases but rarely in Ph+ CML.7 Uncommonly, BCR growth factor receptors, including the erythropoie- breakpoints fall in the microregion (µ-bcr), resulting tin receptor. The JAK2 V617F mutation was discov- in the p230 fusion product associated with chronic ered in 2005 and was shown to be present in 95% neutrophilic leukemia (CNL).6,8 Detection of t(9;22) of polycythemia vera cases and approximately 50% is most commonly performed by either cytogenetics, to 65% of cases of essential thrombocytosis (ET) and fluorescence in situ hybridization (FISH), or reverse primary myelofibrosis (PMF).18 In addition, the JAK2 transcription–polymerase chain reaction (RT-PCR; am- V617F mutation can also be seen in nearly one-half plification of the transcript product); the latter is used of cases of refractory anemia with ring sideroblasts for minimal residual disease testing. associated with marked thrombocytosis.19,20 In cases At diagnosis, conventional cytogenetics can detect of polycythemia vera in which the JAK2 V617F muta- t(9;22) in 95% of cases of CML; however, an additional tion is not detected, the remaining 5% of patients may 2.5% of cases with submicroscopic translocations can harbor mutations in exon 12 of JAK2.20 Similarly, in be recovered by applying molecular methods.6 If the ET and PMF cases lacking the JAK2 V617F mutation, results of both cytogenetics and FISH are negative, an assessment of MPL is indicated, given that 5% or then an alternative diagnosis should be considered. more of patients with PMF and even fewer patients Assaying for t(9;22) can be used to monitor the with ET (1%) will show an aberration in this gene therapeutic response of imatinib mesylate and for (W515K/L).20-22 Exon 10 c-MPL mutations have also been relapse surveillance (using quantitative polymerase reported in ET or PMF (5%).23 The JAK2 V617F mutation chain reaction [qPCR] methods, particularly screen- can be detected via targeted PCR followed by sequencing for positive results 6–12 months after transplan- ing of the amplicon. Other methods include restriction tation).6 However, the importance of the BCR-ABL1 digest of PCR-amplified products followed by separation fusion gene in CML is in its role as a paradigm for by capillary electrophoresis, allele-specific PCR using targeted cancer therapy; patients with CML may re- probe-based gene expression analysis, real-time PCR, ceive imatinib as first-line therapy as established by pyrosequencing, and melting curve analysis.24,25 data from the International Randomized Study of In- Despite the discovery of JAK2 and MPL mutations, terferon and STI571 trial.9 Baseline values from this until recently many ET and PMF cases did not have a trial are also used as the basis of the international unique genetic basis (ie, JAK2, MPL) until 2 indepen- reporting scale (IS). The IS allows for the standard- dent groups identified CALR mutations in this patient ization and comparison between laboratories with subset. CALR mutation, which comprises insertions regard to BCR-ABL levels.10 (ins) and deletions (del) leading to a frameshift, are Response to therapy can be classified as com- found in 20% to 25% of ET and PMF cases and tend plete hematological response, complete cytogenetic to cluster in exon 9. 26-28 Commercial testing utilizes response, and molecular response based on levels of sequencing and fragment length analysis. fusion transcripts by RT-qPCR. A major molecular re- Recently, the Dynamic International Prognostic sponse is defined as a 3-log reduction compared with Scoring System Plus listed unfavorable karyotype as baseline (or ≤ 0.1% IS), and a complete molecular re- a risk factor for predicting survival in primary my- sponse is defined as a 4.5-log reduction or more from elofibrosis.29 baseline.9,11 A lack of response may indicate acquired Mastocytosis: Activating point mutations in KIT resistance. In such cases, Bcr-Abl kinase domain mu- are highly associated with mastocytosis and can be tations (> 100 types documented) can be found in detected in more than 95% of cases of systemic masto- one-half of refractory cases and are an indication to cytosis using real-time qPCR, allele-specific oligonu- adjust therapy by integrating a second-generation ty- cleotide PCR, or direct sequencing.30,31 KIT mutations rosine kinase inhibitor into treatment.12 result in the ligand-independent activation of the c-kit Atypical Chronic Myelogenous Leukemia and tyrosine kinase. The most common mutation in sys- Chronic Neutrophilic Leukemia: The clonality of temic mastocytosis is the D816V variant seen in 68% CNL has been well established given the prior detec- of cases of mastocytosis; however, in certain subsets tions of 20q-, 11q-, and JAK2 V617F mutation in this (eg, aggressive systemic mastocytosis), its incidence disease, but they are not disease specific.13-15 Deep may exceed 80%.32 The presence of this variant con- sequencing has identified CSF3R mutations in CNL stitutes a minor criterion for the diagnosis of systemic

April 2015, Vol. 22, No. 2 Cancer Control 159 mastocytosis. Other KIT variants have been described -Y, del[20q], isochromosome [i][17q], -13/del[13q], (< 5%) and are more likely to be detected in the con- del[11q], del[12p], del[9q], isodicentric [idic][Xq13], text of cutaneous mastocytosis rather than systemic and certain balanced translocations involving chro- mastocytosis.20 Patients with the D816V variant are mosomes 1, 2, 3, 9, 11, 16, and 21).20 Various prog- resistant to imatinib.20,33 nostic models are available for MDS. The most widely adopted is the International Prognostic Scoring Sys- Myeloid and Lymphoid Neoplasms With tem (IPSS) and its revised version (IPSS-R), both of Eosinophilia and Abnormalities of PDGFRA, which integrate the percentage of blasts in the bone PDGFRB, or FGFR1 marrow, cytogenetic abnormalities, and number of A unique group of myeloid and lymphoid neoplasms cytopenias.39 In the latter scheme, cytogenetics are are defined by aberrant tyrosine kinase activity due to placed in 5 tiers: very good (-Y, del[11q]), good (nor- translocations involving PDGFRA, PDGFRB, or FGFR1, mal, del[5q], del[12p], del[20q], and del[5q] + 1 more), all of which are characteristically associated with eo- intermediate (del[7q], +8, +19, i[17q], and others), poor sinophilia. A workup for abnormalities in these genes (-7, inversion [inv][3/t3q/del{3q}], -7/del[7q] + 1 more, should be considered in cases of eosinophilia with and 3 cytogenetic aberrations), and very poor (> 3 ab- end-organ damage or in which secondary reactive normalities).40 Although gene-expression profiling and eosinophilia has been excluded. single nucleotide polymorphism arrays are powerful The cellular ontogeny of these disorders may orig- tools, they are not routinely employed in the clinical inate from a pluripotent (lymphoid–myeloid) stem cell. setting. However, somatic mutation in 40 genes has PDGFRB or FGFR1 can be detected with convention- been found in MDS and analysis for these genes can al cytogenetic analysis (ie, karyotype); however, the add prognostic value.41 By contrast to cytogenetic FIP1LI-PDGFRA results in an 800-kb cryptic del(4q12) abnormalities, which are seen in one-half of cases, that houses CHIC2. Typically, it is detected using FISH at least 1 of these “driver” mutations can be found with a probe spanning CHIC2 or break-apart assay for in most cases of MDS.42,43 For example, patients with either of the translocation partners. The translocation 1 or more mutations in TP53, EZH2, ETV6, RUNX1, can also be detected using RT-PCR.20 FIP1LI–PDGFRA or ASXL1 show survival patterns analogous to those disease commonly manifests as chronic eosinophilic in the next higher tier by subgrouping in the IPSS.44 leukemia, and FIP1LI–PDGFRA is detected in 10% to Various epigenetic modifiers (DNA methylation 20% of those with idiopathic hypereosinophilia.34 Pa- regulators, spliceosome mutations, and histone mod- tients have a response to imatinib more than 100 times ifiers TET2, IDH1/2, DNMT3A, EZH2, ASXL1, SF3B1, greater than that seen in BCR-ABL rearrangement.20,35 U2AF1, SRSF2, and ZRSR2), transcription factor genes, Neoplasms associated with PDGFRB commonly and kinase signaling genes have been implicated in present as chronic myelomonocytic leukemia. ETV6 is MDS, providing a basis for approved therapies and the most common translocation partner, but more than those in development.45 However, these aberrations 13 others have been described; patients will be respon- have limitations because their clinical significance sive to imatinib.34 is not always clear given their association with poor Neoplasms associated with FGFR1 can manifest as prognostic clinical features, our lack of knowledge acute leukemias (myeloid or lymphoid) or as chronic of their interactions with other markers often con- eosinophilic leukemia. Translocation partners include currently detected, intratumoral clonal heterogeneity, ZNF198, CEP110, FGFR10P1, BCR, TRIM24, MYO18A, and the wide gamut of mutations in any given gene.42 HERVK, and FGFR10P2. By contrast to PDGFRA- and Next-generation sequencing (NGS) technologies, PDGFRB-associated neoplasms, these disorders are which garner the power of massively parallel sequence unresponsive to tyrosine kinase inhibitors.20,34 generation, enable laboratories to clinically sequence many genes simultaneously, which was previously Myelodysplastic Syndrome untenable by traditional sequencing technologies. Myelodysplastic syndrome (MDS) is a clonal disor- Commercial testing is available for activated signal- der of myeloid cells characterized by morphological ing genes (KIT, JAK2, NRAS, CBL, MPL), transcription dysplasia and ineffective hematopoiesis that mani- factors (RUNX1, ETV6), epigenetic genes (IDH1/2, fests as peripheral cytopenia.36 Cytogenetic abnor- TET2, DNMT3A, EZH2, ASXL1, SETBP1), ribonucleic malities are seen in one-half of MDS cases, and they acid splicing genes (SF3B1, U2AF1, ZRSF2, SRSF2), most commonly involve del(5q/7q) or monosomies and tumor suppressors (TP53, NPM1, PHF6), among of the same.20,37 TP53 mutations are associated with others. In cases of MDS or MDS/myeloproliferative therapy-related MDS and have a poor prognosis.38 neoplasms in which the diagnoses are unclear or dys- Various cytogenetic abnormalities can be considered plasia has yet to emerge, detecting a mutation in one presumptive evidence of MDS even in the absence of these key genes may be helpful in establishing the of sufficient dysplasia (ie, -5/del[5q], -7/del[7q], +8, diagnosis of a clonal myeloid neoplasm. Although

160 Cancer Control April 2015, Vol. 22, No. 2 they are not formally incorporated into prognostic NPM1 mutations involve 4 to 11 break-point insertions stratification schemas, certain mutations may also car- in exon 12 that lead to the mislocalization of normal ry prognostic implications.46 In addition, robust my- nucleophosmin to the cytoplasm via dimerization.59,60 eloid testing commercially available (FoundationOne This can be detected by PCR followed by sizing via Heme, Cambridge, Massachusetts) can interrogate 405 capillary electrophoresis. NPM1-mutated AML has cancer-related genes, allowing — in theory — for been designated as a provisional entity in the 2008 the identification of targetable mutations and patient WHO classification and is associated with unique mor- enrollment in clinical trials.47 phological (blasts with “cup-like” nuclei) features and a favorable prognosis in normal karyotype AML.20 Acute Myeloid Leukemia CCAAT/enhancer-binding protein α is a 42 kDa AML is the most common type of acute leukemia oc- transcription factor whose loss is associated with the curring in adults.48 In 2015, an estimated 20,830 new oncogenic transformation of myeloid cells due to a cases of AML will occur in the United States, along loss of differentiation.61 Patients with mutated CEBPA with 10,460 deaths.48 AML is a lethal disease and has show outcomes similar to those in the favorable cy- a 5-year relative survival rate of 24.2%.49 However, togenetic subgroup of AML (eg, t[8;21]+ AML).62 The outcomes are heterogenous and overall survival rates prognostic value of CEBPA is in the double-mutated range from approximately 5% to 70%.50 Thus, a need subset of patients lacking FLT3 and NPM1 mutations.63 exists for prognostic markers to predict outcomes and Other single gene alterations that may carry im- guide therapeutic decision-making. Prognostic mark- portant prognostic implications have been identified ers can be clinical, disease related, and molecular, in AML — many were identified during whole genome although the strongest prognostic factor for predict- sequencing studies — and include DNMT3A, IDH1/2, ing therapeutic response and survival is cytogenetic TET2, WT1, ERG expression, BAALC expression, and subgrouping. MN1 expression, among others. IDH2 is associated The results of numerous clinical trials across with a good prognosis and TET2, ASXL1, and PHF6 several decades have indicated that overall survival confer poor prognoses. However, in the absence of rates can be as long as 11.5 years in favorable pa- prospective trials and the present controversy regard- tient groups or shorter than 1 year in patients with ing them, many of these single genes have not been adverse risk.51-53 Those with favorable risk (5-year sur- formally integrated into accepted risk-stratification vival rate of 50%–80%) include those with t(15;17), models.64 In general, investigating the mutation status t(8;21), inv(16/t[16;16]), a normal karyotype and mu- of these genes is simultaneously obtained using NGS tated NPM, or a normal karyotype with biallelic CEBPA technologies. mutation.54 These patients may receive consolidation Biomarkers are important for subclassifying AML therapy following induction 7 + 3 with chemotherapy types, and several categories of AML are defined based (eg, high-dose cytarabine). Those with adverse risk on the presence or absence of recurrent genetic abnor- (overall survival rate of 5%–20%), including those with malities alone, in particular t(8;21)(q22;q22), inv(16), MLL aberrations, inv(3), t(6;9), -7/del(7q), -5/del(5q), t(15;17)(q22;q12), t(9;11)(p22;q23), t(6;9)(p23;q34), TP53 deletions, and a complex karyotype, may under- inv(3), and t(1;22)(p13;q13).13 In some cases, the de- go transplantation after standard induction.55,56 tection of 1 of these aberrations alone is enough to Determining whether consolidation therapy is diagnose AML, even in the absence of the convention- appropriate in those with intermediate risk (overall al criteria of 20% blasts in the marrow or peripheral survival rate of 20%–40%55) is not as clear. In this co- blood, such as in the case of t(15;17) and core-binding hort, molecular testing for FLT3, NPM, and CEBPA is factor-related leukemias (eg, t[8;21][q22;q22], inv[16]) informative and has therapeutic implications. For ex- and possibly inv(3)/t(3;3).65-67 These translocations can ample, detecting FLT3 internal tandem duplication by be detected by conventional cytogenetics, FISH, and PCR in patients with normal karyotype AML may lead more novel technologies, including single molecule to consolidation therapy with hematopoietic stem cell imaging and NGS. One such type of sequencing uses transplantation, after which patients may have a 30% color-coded barcodes directly hybridized to individual likelihood of cure.55 FLT3 codes for a transmembrane target molecules and then digitally detects them in a signal-transducing protein of the tyrosine receptor ki- multiplexed manner.68 nase family and reveals 2 major abnormalities in AML, A comprehensively targeted clinical panel cur- ie, internal tandem duplication in the juxtamembrane rently on the market uses NGS to interrogate the portion resulting in constitutive activation and a point exons of 405 genes and examines the intronic regions mutation in Asp835 (the activity loop portion of proof 31 genes involved in rearrangements as well as tein) resulting in dysregulation. FLT3 aberrations are complementary DNA (ribonucleic acid) to sequence seen in 5% to 10% of AMLs.57,58 265 genes to detect translocations.69 NPM1 encodes nucleophosmin, a 37 kDa protein. In some cases, the detection of a translocation

April 2015, Vol. 22, No. 2 Cancer Control 161 carries both diagnostic and therapeutic importance. eosinophilia due to the overexpression of IL3.77 Namely, t(15;17)(q22;q21), which is diagnostic for t(1;19)+ B-ALL is historically associated with a acute promyelocytic leukemia (AML M3), juxtaposes poor prognosis, but this has changed through the use the 17(q21) retinoic acid receptor α (the receptor for of intensive chemotherapy regimens.78 Immunophe- vitamin A involved in cell proliferation and differen- notypically, the blasts lack CD34 but have aberrant tiation) to the 15(q22) promyelocytic leukemia zinc CD9 positivity.79 Based on gene-expression profiling, finger protein (involved in transcriptional regulation BCR-ABL1–like B-ALL has been identified as being and apoptosis). The chimeric protein blocks differ- associated with deletions of IKZF1, CRLF2 rearrange- entiation beyond the promyelocytic stage, resulting ments, and poor outcomes.80 JAK1/2-activating mu- in acute leukemia. However, treatment with all tran- tations are present in a subset of these patients and sretinoic acid allows for differentiation and, in com- may benefit from Janus kinase inhibitor therapy.80 bination with cytotoxic chemotherapy, can result in In all children with ALL, cytogenetic testing or flow complete remission. Variant translocations involving cytometric analysis of ploidy should be undertaken. RARA are seen in fewer than 2% of cases, but they are Hyperdiploidy (> 50 chromosomes) is associated with important given that some patients may not respond a better prognosis, whereas hypodiploidy (< 44 chro- to all transretinoic acid therapy.20,70 mosomes) is associated with a poor prognosis.80,81 Another example of molecular testing informing Translocations in T-cell ALL (T-ALL) common- therapeutic management involves KIT testing in AML. ly involve 1 of the TCR loci (A, B, G, D). The most C-kit mutations are associated with core-binding fac- common translocation partner includes HOX11 on tor AMLs and may abrogate the favorable prognosis chromosome 10 (occurring in 10%–30% of cases) or generally associated with this group.71 various other transcription factors dysregulated by juxtaposition to 1 of the TCR genes.82 PICALM-MLLT10 Lymphoid Neoplasms and MLL rearrangements are seen in approximately Lymphoblastic Leukemia/Lymphoma 10% of cases.83 Similar to AML, the detection of certain characteristic Both B-ALL and mature B-cell non-Hodgkin lym- translocations further subclassifies cases of B-ALL, in- phomas (NHLs) show clonal immunoglobulin (Ig) cluding t(12;21)(p12,q22) TEL/AML-1, t(1;19)(q23;p13) gene rearrangements, which are helpful in residual PBX/E2A, t(9;22)(q34;q11) ABL/BCR, (5;14)(q31;q32) disease testing as well as in establishing a malignant IL3-IGH, and (V;11)(V;q23) V/MLL.20 diagnosis. For follow-up specimens, screening for t(9;22)(q34;q11) is seen in 25% of adult cases clonal peaks identical to those identified at diagno- and 2% to 4% pediatric cases.72 Patients with this Ph+ sis can be performed to assess for residual/relapsed translocation carry the worst prognosis of all types minimal residual disease. of lymphoblastic leukemia73; however, these patients B-cell antigen receptors are encoded by IGH can be treated with adjuvant imatinib, which improves (14q32), IGLK (2p11), and IGLL (20q11), coding for complete remission rates.74 the Ig heavy chain, κ light chain, and the λ light chain, The presence of the translocation also serves as a respectively. Each contains variable (V), joining (J), useful marker for minimum residual disease testing. and constant regions; IGH contains an additional di- Molecular methods for detecting the translocation are versity (D) region. Multiplex PCR that uses primers to similar to those used for CML. Notably, when qRT-PCR target highly conserved framework regions within the testing is undertaken, the p190 protein product is V segment are used to generate PCR products, which typically associated with B-ALL, not the p210 protein can then be separated using capillary electrophoresis, typical of CML; in children with ALL, the break point which is preferred to Southern blot analysis.84,85 occurs in m-bcr, which generates the p190 protein Monoclonal peaks have heights 2 to 3 times that product in 90% of cases.6 If a p210 protein product of the background and can be seen in clonal B-cell is detected, then consideration should be given to a neoplasms. Repeat peaks in duplicate wells raise con- lymphoid blast crisis arising in CML. fidence that clonal peaks do not represent a PCR By contrast to t(9;22), the reverse demographic artifact. Care must be taken because false-positive appears to be true for t(12;21)(p12,q22): It occurs results can occur in cases of benign lymphoid hy- in 25% of pediatric cases but is rare in adults and perplasia, which may be present in the setting of associated with a favorable prognosis and curative immunodeficiency and autoimmune disease.86,87 Fur- rates of higher than 90% in children.75 MLL can have thermore, lineage infidelity is present with BCR gene various translocation partners, the most common of rearrangement and is similar to that seen in T-cell which is AF4 on chromosome 4. MLL translocations lymphoma. Most precursor and 5% to 10% of ma- carry a poor prognosis, and this is particularly true in ture B-cell neoplasms will harbor clonal T-cell gene infants.76 The unique characteristic of t(5;14)(q31;q32) rearrangements.6 Other various factors may result in IL3/IGH B-ALL, which is rare, is its association with false-negative results, including primer failure due to

162 Cancer Control April 2015, Vol. 22, No. 2 somatic hypermutation (which can occur at a rate of are worthy of mention (t[14;18][q32;q21], t[11;18] > 50% in certain lymphoid neoplasms [eg, follicular [q21;q21], t[1;14][p22;q32]) because they represent lymphoma]), complex IGH rearrangements, or DNA mucosa-associated lymphoid tissue that usually does of poor quality.6,88 not respond to Helicobacter pylori eradication.100-102 Although multitudinous, single nucleotide vari- Mature B-Cell Neoplasms ants and copy number changes have been found in Various translocations are associated with B-cell NHLs B-cell NHL, sometimes even with reported prognostic and their detection helps to establish a diagnosis in significance (eg, NOTCH1 mutations in chronic lym- these entities. t(14;18) involves BCL2 on chromosome phocytic leukemia [CLL]), in clinical practice testing 18 and IGH on chromosome 14. BCL2 is juxtaposed for these in B-cell NHL has a limited role.103 A limited to the J region of the heavy chain. Given that the 7-gene CLL panel with targets that carry prognostic IGH enhancer element is highly active, bcl2 can be- implications has been launched by Cancer Genetics come overexpressed. Because bcl2 has antiapoptotic (Rutherford, New Jersey). properties, its overexpression will result in neoplasia. Commonly, when molecular testing is indicated in This translocation is found in 85% to 90% of cases of B-NHL, the genetic aberrations are usually of diagnos- follicular lymphoma (a lower percentage occurs in tic importance. BRAF V600E mutation was originally cases of high-grade follicular lymphoma) and 25% found in 100% patients with hairy cell leukemia com- of cases of diffuse large B-cell lymphoma (DLBCL).20 pared with none of the 195 other peripheral B-cell Because follicular lymphomas may lack demonstrable lymphoma/leukemias.104-106 The results of subsequent Ig clonality due to ongoing somatic hypermutation, studies have confirmed that the mutation is present the use of FISH or PCR for the translocation offers in all cases of hairy cell leukemia and is rare in other alternative markers to assess for clonality, establish a chronic lymphoproliferative disorders.104-106 In lymph- diagnosis, or both; however, FISH is preferred to PCR oplasmacytic lymphoma, MYD88 mutation has been as it is more sensitive and specific.89 detected with high frequency (> 90%),107 and detecting DLBCL, in addition to BCL2, can have translo- the mutation may be diagnostically useful given the cations of BCL6 and MYC (10% of cases).90 When a overlap with lymphoplasmacytic lymphoma and other MYC translocation is detected along with other spe- low-grade B-cell lymphomas that may be associated cific translocations (usually BCL2 and BCL6) in an with plasmacytic differentiation, including marginal intermediate to large B-cell lymphoma, its presence zone lymphoma, multiple myeloma, and CLL. In these qualifies as a “double hit” lymphoma, which may be other conditions, the prevalence of the mutation is categorized under the rubric of large B-cell lympho- 3% to 9%.108 Of note, nearly one-third of activated ma with features intermediate between DLBCL and B-cell-like DLBLC harbors the mutation, and its pres- Burkitt lymphoma.91 ence is not useful in the differential with IgM mono- Typically, MYC gene rearrangements are associ- clonal gammopathy of undetermined significance.108 ated with Burkitt lymphoma, but they can also be Thus, a correlation with morphology and other an- present in plasmablastic lymphomas (50% of the time) cillary studies is needed. and, rarely, in follicular lymphoma and primary central In CLL, hypermutation status is assessed by com- nervous system DLBCL.92,93 Burkitt lymphoma is char- paring each IGH clonally rearranged gene sequence acterized by t(8;14) involving MYC and IGH and will with a database of germline V-region sequences to less commonly show translocations involving light determine the expressed V-region gene and the extent chain loci (κ or λ).94-96 BCL6 translocations can be seen and position of somatic mutations. If a difference in follicular lymphoma, DLBCL, and are frequently exists of more than 2%, then the tumor is considered identified in primary cutaneous leg-type DLBCL.97 hypermutated and confers a better prognosis.109 Nearly all cases of mantle cell lymphoma carry t(11;14)(q13;q32) CCND1-IGH, which can be assessed Mature T-Cell Lymphoproliferative Disorders by FISH, and is preferred over PCR-based method- A total of 95% of T cells express the a-b receptor ologies that demonstrate lower sensitivity rates and a smaller proportion express the γ-δ receptor; (50%–60%); this is because of the large number of both of these receptors contain heterodimer proteins dispersed break points at 11q13.98 encoded by TCR genes located on chromosomes 7 and Translocation of CCND1 with light chain has also 14.6,110 Early in development, the TCR genes undergo been reported; rarely, cyclin D2 may be translocated, somatic rearrangement involving V, D, and J regions which should be a consideration in cyclin D1– tumors (TCRB and TCRD) or V–J rearrangements alone (TCRA otherwise characteristic of mantle cell lymphoma.99 and TCRG). Various translocations have also been described in Unlike in B cells, in which Ig light chains (κ and lymphoma involving the mucosa-associated lymphoid λ) can be assessed for clonality by flow cytometry or tissue. Of these, MALT1 and BCL10 translocations immunohistochemistry, establishing the clonal na-

April 2015, Vol. 22, No. 2 Cancer Control 163 ture of T cells using these techniques is difficult, thus second chromosome 7 may be present. i(7)(q10) can making TCR gene rearrangement studies valuable.111 be detected using FISH.118 Each T cell bears a unique, rearranged sequence. Un- In adult T-cell leukemia, clonal integration of the der normal circumstances, a range of gene products human T-lymphotropic virus type 1 viral DNA can be can be seen given the gamut of polyclonal T cells seen. Although it is conceivable to perform testing present. However, if a clonal process is present, then via Sanger sequencing, it is typically easier to per- a particular gene rearrangement product should pre- form serum studies for human T-lymphotropic virus dominate,112 and it can be detected using Southern type 1.119 In enteropathy-associated T-cell lymphoma, blot analysis as a single clonal band. amplification of 19q31.3, del(16q12.1), or both have Although Southern blot analysis is considered the been reported.20 gold standard, it is inefficient and seldom used in modern clinical laboratories for T-cell clonality detec- Conclusion tion. Drawbacks of Southern blot analysis include its Molecular testing is well entrenched in the workup high cost, increased time, large sample requirements, and management of hematological malignancies. As and low sensitivity rates compared with PCR (5%–10% sequencing technologies become both more powerful vs 1%).112,113 and affordable, they will take on an even larger role PCR amplification of TCRG and TCRB gene prod- in the molecular diagnostics of hematopathology and ucts followed by gel separation or capillary electro- in the era of precision medicine. phoresis is employed in the clinical laboratory. PCR testing demonstrates a clonal peak 2 to 3 times larger References than the background peaks in T-cell lymphomas. In 1. Memorial Sloan Kettering Cancer Center. Molecular pathology fellow- certain cases, false-negative results may occur if the ship. http://www.mskcc.org/education/fellowships/fellowship/molecular-diagnostics-fellowship. Accessed February 18, 2015. rearrangement involves the primer site or too few T 2. University of Texas MD Anderson Cancer Center. Molecular genetic cells are present for analysis. Positive cases of gene pathology fellowship. http://www.mdanderson.org/education-and-research/ education-and-training/schools-and-programs/graduate-medical-education/ arrangements should not be taken to mean that T-cell residency-and-fellowship-programs/molecular-genetic-pathology-fellowship. lymphoma is present. Such positivity can be seen in html. Accessed February 18, 2015. 3. Gonon-Demoulian R, Goldman JM, Nicolini FE. History of chronic cases of B-cell lymphoblastic leukemia (approximately myeloid leukemia: a paradigm in the treatment of cancer [in French]. Bull 50% of cases), mature B-cell lymphomas (5%–10%), Cancer. 2014;101(1):56-67. 4. Gahrton G. Historical note on the discovery of the Philadelphia chro- AML (10%), and non-neoplastic conditions such as mosome. Cancer Genet. 2012;205(6):338-339. autoimmune disorders, certain infectious diseases 5. Zhen C, Wang YL. Molecular monitoring of chronic myeloid leukemia: international standardization of BCR-ABL1 quantitation. J Mol Diagn. (Epstein–Barr virus–induced oligoclonal processes), 2013;15(5):556-564. and certain cutaneous lesions (eg, lymphomatoid 6. Leonard DG, ed. Molecular Pathology in Clinical Practice: Genetics. 6 New York: Springer; 2008. papulosis). 7. Nashed AL, Rao KW, Gulley ML. Clinical applications of BCR-ABL In ALK-positive anaplastic large cell lymphoma, molecular testing in acute leukemia. J Mol Diagn. 2003;5(2):63-72. 8. Pane F, Frigeri F, Sindona M, et al. Neutrophilic-chronic myeloid leu- t(2;5)(p23;q35) juxtaposing ALK and NPM, respec- kemia: a distinct disease with a specific molecular marker (BCR/ABL with C3/ tively, is the most frequent genetic translocation A2 junction). Blood. 1996;88(7):2410-2414. 9. O’Brien SG, Guilhot F, Larson RA, et al; IRIS Investigators. Imatinib (83% of pediatric and 31% of adult cases) present; compared with interferon and low-dose cytarabine for newly diagnosed chron- however, various, less frequently seen partners ic-phase chronic myeloid leukemia. N Engl J Med. 2003;348(11):994-1004. 10. Branford S, Fletcher L, Cross NC, et al. Desirable performance char- have also been described, including TPM3 (13%), acteristics for BCR-ABL measurement on an international reporting scale to ATIC, TGS, CLTC, MSN, TPM4, MYH9, and ALO17 allow consistent interpretation of individual patient response and comparison 20,114 of response rates between clinical trials. Blood. 2008;112(8):3330-3338. (all < 1%). The translocation can be assayed using 11. Press RD. Major molecular response in CML patients treated with RT-PCR or break-apart FISH probes.6,20,114,115 tyrosine kinase inhibitors: the paradigm for monitoring targeted cancer therapy. Oncologist. 2010;15(7):744-749. In T-cell prolymphocytic leukemia, the most com- 12. Frankfurt O, Licht JD. Ponatinib--a step forward in overcoming resis- mon genetic aberration (80%) involves inv(14) jux- tance in chronic myeloid leukemia. Clin Cancer Res. 2013;19(21):5828-5834. 13. Gotlib J, Maxson JE, George TI, et al. The new genetics of chronic taposing the TRA locus at 14q11 to the TCL1A and neutrophilic leukemia and atypical CML: implications for diagnosis and treat- TCL1B oncogenes.20 In a subset of cases, a reciprocal ment. Blood. 2013;122(10):1707-1711. 14. Steensma DP, Dewald GW, Lasho TL, et al. The JAK2 V617F activating tandem t(14;14) is present; t(X;14)(q28;q11) has also tyrosine kinase mutation is an infrequent event in both “atypical” myeloprolifer- been described but is less common. Both can be as- ative disorders and myelodysplastic syndromes. Blood. 2005;106(4):1207-1209. 15. Elliott MA, Hanson CA, Dewald GW, et al. WHO-defined chronic neu- sayed using FISH. Cytogenetics can be used to detect trophilic leukemia: a long-term analysis of 12 cases and a critical review of chromosome 8 abnormalities (70%–80%), ATM dele- the literature. Leukemia. 2005;19(2):313-317. 16. Maxson JE, Gotlib J, Pollyea DA, et al. Oncogenic CSF3R mu- tions, as well as del(12p13), all of which can be seen tations in chronic neutrophilic leukemia and atypical CML. N Engl J Med. in the setting of T-cell prolymphocytic leukemia.116,117 2013;368(19):1781-1790. 17. Elliott MA, Tefferi A. Chronic neutrophilic leukemia 2014: update on diag- Hepatosplenic T-cell lymphoma is associated with nosis, molecular genetics, and management. Am J Hematol. 2014;89(6):651-658. numerical abnormalities of chromosome 7, and most 18. Baxter EJ, Scott LM, Campbell PJ, et al; Cancer Genome Project. Acquired mutation of the tyrosine kinase JAK2 in human myeloproliferative cases will demonstrate i(7q). As the disease progress- disorders. Lancet. 2005;365(9464):1054-1061. es, 2 to 5 copies of i(7)(q10) or derangements in the 19. Szpurka H, Jankowska AM, Makishima H, et al. Spectrum of muta-

164 Cancer Control April 2015, Vol. 22, No. 2 tions in RARS-T patients includes TET2 and ASXL1 mutations. Leuk Res. dation of a clinical cancer genomic profiling test based on massively parallel 2010;34(8):969-973. DNA sequencing. Nat Biotechnol. 2013;31(11):1023-1031. 20. Swerdlow EA. WHO Classification of Tumours of the Heamatopoietic 48. American Cancer Society. Cancer Facts & Figures 2015. Atlanta, GA: and Lymphoid Tissues. 4th ed. Lyon: IARC; 2008. American Cancer Society; 2015. http://www.cancer.org/acs/groups/content/@ 21. Pancrazzi A, Guglielmelli P, Ponziani V, et al. A sensitive detection editorial/documents/document/acspc-044552.pdf. Accessed February 17, 2015. method for MPLW515L or MPLW515K mutation in chronic myeloproliferative 49. Nestal de Moraes G, Castro CP, Salustiano EJ, et al. The pterocarpan- disorders with locked nucleic acid-modified probes and real-time polymerase quinone LQB-118 induces apoptosis in acute myeloid leukemia cells of distinct chain reaction. J Mol Diagn. 2008;10(5):435-441. molecular subtypes and targets FoxO3a and FoxM1 transcription factors. Int 22. Pikman Y, Lee BH, Mercher T, et al. MPLW515L is a novel somat- J Oncol. 2014;45(5):1949-1958. ic activating mutation in myelofibrosis with myeloid metaplasia. PLoS Med. 50. Grimwade D, Walker H, Oliver F, et al; The Medical Research Council 2006;3(7):e270. Adult and Children’s Leukaemia Working Parties. The importance of diagnostic 23. Ding J, Komatsu H, Wakita A, et al. Familial essential thrombocythemia cytogenetics on outcome in AML: analysis of 1,612 patients entered into the associated with a dominant-positive activating mutation of the c-MPL gene, which MRC AML 10 trial. Blood. 1998;92(7):2322-2333. encodes for the receptor for thrombopoietin. Blood. 2004;103(11):4198-4200. 51. Slovak ML, Kopecky KJ, Cassileth PA, et al. Karyotypic analysis predicts 24. Cankovic M, Whiteley L, Hawley RC, et al. Clinical performance of outcome of preremission and postremission therapy in adult acute myeloid JAK2 V617F mutation detection assays in a molecular diagnostics labora- leukemia: a Southwest Oncology Group/Eastern Cooperative Oncology Group tory: evaluation of screening and quantitation methods. Am J Clin Pathol. study. Blood. 2000;96(13):4075-4083. 2009;132(5):713-721. 52. Mrózek K, Marcucci G, Nicolet D, et al. Prognostic significance of the 25. Steensma DP. JAK2 V617F in myeloid disorders: molecular diagnostic European LeukemiaNet standardized system for reporting cytogenetic and techniques and their clinical utility: a paper from the 2005 William Beaumont molecular alterations in adults with acute myeloid leukemia. J Clin Oncol. Hospital Symposium on Molecular Pathology. J Mol Diagn. 2006;8(4):397-411. 2012;30(36):4515-4523. 26. Cazzola M, Kralovics R. From Janus kinase 2 to calreticulin: the clin- 53. Döhner H, Estey EH, Amadori S, et al; European LeukemiaNet. Diag- ically relevant genomic landscape of myeloproliferative neoplasms. Blood. nosis and management of acute myeloid leukemia in adults: recommendations 2014;123(24):3714-3719. from an international expert panel, on behalf of the European LeukemiaNet. 27. Klampfl T, Gisslinger H, Harutyunyan AS, et al. Somatic mutations of cal- Blood. 2010;115(3):453-474. reticulin in myeloproliferative neoplasms. N Engl J Med. 2013;369(25):2379-2390. 54. Dufour A, Schneider F, Metzeler KH, et al. Acute myeloid leukemia 28. Nangalia J, Massie CE, Baxter EJ, et al. Somatic CALR mutations with biallelic CEBPA gene mutations and normal karyotype represents a dis- in myeloproliferative neoplasms with nonmutated JAK2. N Engl J Med. tinct genetic entity associated with a favorable clinical outcome. J Clin Oncol. 2013;369(25):2391-2405. 2010;28(4):570-577. 29. Gangat N, Caramazza D, Vaidya R, et al. DIPSS plus: a refined Dynamic 55. Zeisig BB, Kulasekararaj AG, Mufti GJ, et al. SnapShot: acute myeloid International Prognostic Scoring System for primary myelofibrosis that incor- leukemia. Cancer Cell. 2012;22(5):698-698.e1. porates prognostic information from karyotype, platelet count, and transfusion 56. Estey EH. Acute myeloid leukemia: 2013 update on risk-stratification status. J Clin Oncol. 2011;29(4):392-397. and management. Am J Hematol. 2013;88(4):318-327. 30. Kristensen T, Vestergaard H, Møller MB. Improved detection of the KIT 57. Bullinger L, Döhner K, Kranz R, et al. An FLT3 gene-expres- D816V mutation in patients with systemic mastocytosis using a quantitative sion signature predicts clinical outcome in normal karyotype AML. Blood. and highly sensitive real-time qPCR assay. J Mol Diagn. 2011;13(2):180-188. 2008;111(9):4490-4495. 31. Tefferi A, Verstovsek S, Pardanani A. How we diagnose and treat 58. Stirewalt DL, Radich JP. The role of FLT3 in haematopoietic malignan- WHO-defined systemic mastocytosis in adults. Haematologica. 2008;93(1):6-9. cies. Nat Rev Cancer. 2003;3(9):650-665. 32. Lim KH, Tefferi A, Lasho TL, et al. Systemic mastocytosis in 342 consecu- 59. Tan AY, Westerman DA, Carney DA, et al. Detection of NPM1 exon tive adults: survival studies and prognostic factors. Blood. 2009;113(23):5727-5736. 12 mutations and FLT3 - internal tandem duplications by high resolution 33. Pardanani A. Systemic mastocytosis in adults: 2013 update on diagno- melting analysis in normal karyotype acute myeloid leukemia. J Hematol sis, risk stratification, and management. Am J Hematol. 2013;88(7):612-624. Oncol. 2008;1:10. 34. Gotlib J. World Health Organization-defined eosinophilic disorders: 60. Thiede C, Creutzig E, Reinhardt D, et al. Different types of NPM1 2014 update on diagnosis, risk stratification, and management. Am J Hematol. mutations in children and adults: evidence for an effect of patient age on the 2014;89(3):325-337. prevalence of the TCTG-tandem duplication in NPM1-exon 12. Leukemia. 35. Cools J, DeAngelo DJ, Gotlib J, et al. A tyrosine kinase created by fusion 2007;21(2):366-367. of the PDGFRA and FIP1L1 genes as a therapeutic target of imatinib in idiopathic 61. Lin LI, Chen CY, Lin DT, et al. Characterization of CEBPA mutations in hypereosinophilic syndrome. N Engl J Med. 2003;348(13):1201-1214. acute myeloid leukemia: most patients with CEBPA mutations have biallelic 36. Bejar R, Tiu RV, Sekeres MA, et al. Myelodysplastic syndromes: recent mutations and show a distinct immunophenotype of the leukemic cells. Clin advancements in risk stratification and unmet therapeutic challenges. Am Cancer Res. 2005;11(4):1372-1379. Soc Clin Oncol Educ Book. 2013:e256-e270. http://meetinglibrary.asco.org/ 62. Leroy H, Roumier C, Huyghe P, et al. CEBPA point mutations in hema- sites/meetinglibrary.asco.org/files/Educational%20Book/PDF%20Files/2013/ tological malignancies. Leukemia. 2005;19(3):329-334. EdBookAM201333e256.pdf. Accessed February 18, 2015. 63. Green CL, Koo KK, Hills RK, et al. Prognostic significance of CEBPA 37. Jacoby MA, Walter MJ. Detection of copy number alterations in acute mutations in a large cohort of younger adult patients with acute myeloid leu- myeloid leukemia and myelodysplastic syndromes. Expert Rev Mol Diagn. kemia: impact of double CEBPA mutations and the interaction with FLT3 and 2012;12(3):253-264. NPM1 mutations. J Clin Oncol. 2010;28(16):2739-2747. 38. Lai JL, Preudhomme C, Zandecki M, et al. Myelodysplastic syndromes 64. Naoe T, Kiyoi H. Gene mutations of acute myeloid leukemia in the and acute myeloid leukemia with 17p deletion. An entity characterized by genome era. Int J Hematol. 2013;97(2):165-174. specific dysgranulopoïesis and a high incidence of P53 mutations. Leukemia. 65. Sangle NA, Perkins SL. Core-binding factor acute myeloid leukemia. 1995;9(3):370-381. Arch Pathol Lab Med. 2011;135(11):1504-1509. 39. Greenberg P, Cox C, LeBeau MM, et al. International scoring system for 66. Haferlach C, Bacher U, Haferlach T, et al. The inv(3)(q21q26)/t(3;3) evaluating prognosis in myelodysplastic syndromes. Blood. 1997;89(6):2079-2088. (q21;q26) is frequently accompanied by alterations of the RUNX1, KRAS and 40. Schanz J, Tüchler H, Solé F, et al. New comprehensive cytogenetic NRAS and NF1 genes and mediates adverse prognosis both in MDS and in scoring system for primary myelodysplastic syndromes (MDS) and oligoblastic AML: a study in 39 cases of MDS or AML. Leukemia. 2011;25(5):874-877. acute myeloid leukemia after MDS derived from an international database 67. Rogers HJ, Vardiman JW, Anastasi J, et al. Complex or monoso- merge. J Clin Oncol. 2012;30(8):820-829. mal karyotype and not blast percentage is associated with poor survival in 41. Malcovati L, Hellström-Lindberg E, Bowen D, et al; European Leuke- acute myeloid leukemia and myelodysplastic syndrome patients with inv(3) mia Net. Diagnosis and treatment of primary myelodysplastic syndromes (q21q26.2)/t(3;3)(q21;q26.2): a Bone Marrow Pathology Group study. Hae- in adults: recommendations from the European LeukemiaNet. Blood. matologica. 2014;99(5):821-829. 2013;122(17):2943-2964. 68. Reis PP, Waldron L, Goswami RS, et al. mRNA transcript quantification 42. Bejar R. Prognostic models in myelodysplastic syndromes. Hematology in archival samples using multiplexed, color-coded probes. BMC Biotechnol. Am Soc Hematol Educ Program. 2013;2013:504-510. 2011;11:46. 43. Papaemmanuil E, Gerstung M, Malcovati L, et al; Chronic Myeloid 69. Foundation One. Technical information and test overview. http://foun- Disorders Working Group of the International Cancer Genome Consortium. dationone.com/ONE-I-001-20140804_(nobleed)TechnicalSpecs.pdf. Accessed Clinical and biological implications of driver mutations in myelodysplastic syn- February 25, 2015. dromes. Blood. 2013;122(22):3616-3627. 70. Wang ZY, Chen Z. Acute promyelocytic leukemia: from highly fatal to 44. Bejar R, Stevenson K, Abdel-Wahab O, et al. Clinical effect of point mu- highly curable. Blood. 2008;111(5):2505-2515. tations in myelodysplastic syndromes. N Engl J Med. 2011;364(26):2496-2506. 71. Cairoli R, Beghini A, Grillo G, et al. Prognostic impact of c-KIT muta- 45. Bravo GM, Lee E, Merchan B, et al. Integrating genetics and epigenetics tions in core binding factor leukemias: an Italian retrospective study. Blood. in myelodysplastic syndromes: advances in pathogenesis and disease evolu- 2006;107(9):3463-3468. tion. Br J Haematol. 2014;166(5):646-659. 72. Moorman AV, Chilton L, Wilkinson J, et al. A population-based cyto- 46. Haferlach T, Nagata Y, Grossmann V, et al. Landscape of genetic lesions genetic study of adults with acute lymphoblastic leukemia [Erratum appears in 944 patients with myelodysplastic syndromes. Leukemia. 2014;28(2):241-247. in Blood. 2010;116(5):1017]. Blood. 2010;115(2):206-214. 47. Frampton GM, Fichtenholtz A, Otto GA, et al. Development and vali- 73. Hoelzer D, Thiel E, Löffler H, et al. Prognostic factors in a multi-

April 2015, Vol. 22, No. 2 Cancer Control 165 center study for treatment of acute lymphoblastic leukemia in adults. Blood. 101. Streubel B, Simonitsch-Klupp I, Müllauer L, et al. Variable frequencies 1988;71(1):123-131. of MALT lymphoma-associated genetic aberrations in MALT lymphomas of 74. Fielding AK. How I treat Philadelphia chromosome-positive acute lym- different sites. Leukemia. 2004;18(10):1722-1726. phoblastic leukemia. Blood. 2010;116(18):3409-3417. 102. Hamoudi RA, Appert A, Ye H, et al. Differential expression of NF-kappaB 75. Hunger SP, Lu X, Devidas M, et al. Improved survival for children and target genes in MALT lymphoma with and without chromosome translocation: adolescents with acute lymphoblastic leukemia between 1990 and 2005: a report insights into molecular mechanism. Leukemia. 2010;24(8):1487-1497. from the Children’s Oncology Group. J Clin Oncol. 2012;30(14):1663-1669. 103. Fabbri G, Rasi S, Rossi D, et al. Analysis of the chronic lymphocytic 76. Hilden JM, Dinndorf PA, Meerbaum SO, et al; Children’s Oncology leukemia coding genome: role of NOTCH1 mutational activation. J Exp Med. Group. Analysis of prognostic factors of acute lymphoblastic leukemia in 2011;208(7):1389-1401. infants: report on CCG 1953 from the Children’s Oncology Group. Blood. 104. Arcaini L, Zibellini S, Boveri E, et al. The BRAF V600E mutation in hairy 2006;108(2):441-451. cell leukemia and other mature B-cell neoplasms. Blood. 2012;119(1):188-191. 77. Grimaldi JC, Meeker TC. The t(5;14) chromosomal translocation in a 105. Tiacci E, Trifonov V, Schiavoni G, et al. BRAF mutations in hairy-cell case of acute lymphocytic leukemia joins the interleukin-3 gene to the immu- leukemia. N Engl J Med. 2011;364(24):2305-2315. noglobulin heavy chain gene. Blood. 1989;73(8):2081-2085. 106. Tiacci E, Schiavoni G, Forconi F, et al. Simple genetic diagnosis of 78. Pui CH, Relling MV, Downing JR. Acute lymphoblastic leukemia. N Engl hairy cell leukemia by sensitive detection of the BRAF-V600E mutation. Blood. J Med. 2004;350(15):1535-1548. 2012;119(1):192-195. 79. Borowitz MJ, Hunger SP, Carroll AJ, et al. Predictability of the t(1;19) 107. Treon SP, Xu L, Yang G, et al. MYD88 L265P somatic mutation in (q23;p13) from surface antigen phenotype: implications for screening cases Waldenstrom’s macroglobulinemia. N Engl J Med. 2012;367(9):826-833. of childhood acute lymphoblastic leukemia for molecular analysis: a Pediatric 108. Treon SP, Hunter ZR. A new era for Waldenstrom macroglobulinemia: Oncology Group study. Blood. 1993;82(4):1086-1091. MYD88 L265P. Blood. 2013;121(22):4434-4436. 80. Mullighan CG. Genome sequencing of lymphoid malignancies. Blood. 109. Murray F, Darzentas N, Hadzidimitriou A, et al. Stereotyped patterns 2013;122(24):3899-3907. of somatic hypermutation in subsets of patients with chronic lymphocytic leu- 81. Nachman JB, Heerema NA, Sather H, et al. Outcome of treat- kemia: implications for the role of antigen selection in leukemogenesis. Blood. ment in children with hypodiploid acute lymphoblastic leukemia. Blood. 2008;111(3):1524-1533. 2007;110(4):1112-1115. 110. Rådestad E, Wikell H, Engström M, et al. Alpha/beta T-cell depleted 82. Ferrando AA, Neuberg DS, Dodge RK, et al. Prognostic importance of grafts as an immunological booster to treat graft failure after hematopoietic TLX1 (HOX11) oncogene expression in adults with T-cell acute lymphoblastic stem cell transplantation with HLA-matched related and unrelated donors. J leukaemia. Lancet. 2004;363(9408):535-536. Immunol Res. 2014;2014:578741. 83. Turkmen S, Timmermann B, Bartels G, et al. Involvement of the 111. British Committee for Standards in Haematology; Royal College of MLL gene in adult T-lymphoblastic leukemia. Genes Chromosomes Cancer. Pathologists. Best Practice in Lymphoma Diagnosis and Reporting. London: 2012;51(12):1114-1124. British Society for Haemotology; 2010. http://www.bcshguidelines.com/docu- 84. Bagg A. Immunoglobulin and T-cell receptor gene rearrangements: ments/Lymphoma_diagnosis_bcsh_042010.pdf. Accessed February 18, 2015. minding your B’s and T’s in assessing lineage and clonality in neoplastic 112. Chitgopeker P, Sahni D. T-cell receptor gene rearrangement detec- lymphoproliferative disorders. J Mol Diagn. 2006;8(4):426-429. tion in suspected cases of cutaneous T-cell lymphoma. J Invest Dermatol. 85. Sandberg Y, van Gastel-Mol EJ, Verhaaf B, et al. BIOMED-2 multiplex 2014;134(4):e19. immunoglobulin/T-cell receptor polymerase chain reaction protocols can re- 113. Ioachim HL, Medeiros LJ, eds. Ioachim’s Lymph Node Pathology. 4th liably replace Southern blot analysis in routine clonality diagnostics. J Mol ed. Philadelphia: Wolters Kluwer/Lippincott Williams & Wilkins; 2009. Diagn. 2005;7(4):495-503. 114. Liang X, Meech SJ, Odom LF, et al. Assessment of t(2;5)(p23;q35) 86. Elenitoba-Johnson KS, Bohling SD, Mitchell RS, et al. PCR analy- translocation and variants in pediatric ALK+ anaplastic large cell lymphoma. sis of the immunoglobulin heavy chain gene in polyclonal processes can Am J Clin Pathol. 2004;121(4):496-506. yield pseudoclonal bands as an artifact of low B cell number. J Mol Diagn. 115. Rudolph C, Bittner C, Feller AC, et al. Cytogenetic characteristics of a 2000;2(2):92-96. murine in vitro model for the human anaplastic large cell lymphoma (ALCL). 87. Nihal M, Mikkola D, Wood GS. Detection of clonally restricted immu- Cytogenet Genome Res. 2006;114(3-4):292-295. noglobulin heavy chain gene rearrangements in normal and lesional skin: 116. Delgado P, Starshak P, Rao N, et al. A comprehensive update on mo- analysis of the B cell component of the skin-associated lymphoid tissue and lecular and cytogenetic abnormalities in T-cell prolymphocytic leukemia (T-pll). implications for the molecular diagnosis of cutaneous B cell lymphomas. J J Assoc Genet Technol. 2012;38(4):193-198. Mol Diagn. 2000;2(1):5-10. 117. Dearden C. How I treat prolymphocytic leukemia. Blood. 2012;120(3): 88. Grody WW, Nakamura RM, Kiechle FL, et al, eds. Molecular Diagnos- 538-551. tics: Techniques and Applications for the Clinical Laboratory. San Diego, CA: 118. Mandava S, Sonar R, Ahmad F, et al. Cytogenetic and molecular char- Academic Press; 2010. acterization of a hepatosplenic T-cell lymphoma: report of a novel chromosomal 89. Wang SA, Wang L, Hochberg EP, et al. Low histologic grade follicular aberration. Cancer Genet. 2011;204(2):103-107. lymphoma with high proliferation index: morphologic and clinical features. Am 119. Tsukasaki K, Tsushima H, Yamamura M, et al. Integration patterns of J Surg Pathol. 2005;29(11):1490-1496. HTLV-I provirus in relation to the clinical course of ATL: frequent clonal change 90. Tomita N. BCL2 and MYC dual-hit lymphoma/leukemia. J Clin Exp at crisis from indolent disease. Blood. 1997;89(3):948-956. Hematop. 2011;51(1):7-12. 91. Swerdlow SH. Diagnosis of ‘double hit’ diffuse large B-cell lymphoma and B-cell lymphoma, unclassifiable, with features intermediate between DL- BCL and Burkitt lymphoma: when and how, FISH versus IHC. ASH Education Handbook. 2014;(1):90-99. 92. Valera A, Balague O, Colomo L, et al. IG/MYC rearrangements are the main cytogenetic alteration in plasmablastic lymphomas. Am J Surg Pathol. 2010;34(11):1686-1694. 93. Aukema SM, Siebert R, Schuuring E, et al. Double-hit B-cell lymphomas. Blood. 2011;117(8):2319-2331. 94. Hummel M, Bentink S, Berger H, et al; Molecular Mechanisms in Ma- lignant Lymphomas Network Project of the Deutsche Krebshilfe. A biologic definition of Burkitt’s lymphoma from transcriptional and genomic profiling. N Engl J Med. 2006;354(23):2419-2430. 95. Haralambieva E, Boerma EJ, van Imhoff GW, et al. Clinical, immuno- phenotypic, and genetic analysis of adult lymphomas with morphologic features of Burkitt lymphoma. Am J Surg Pathol. 2005;29(8):1086-1094. 96. Ferry JA. Burkitt’s lymphoma: clinicopathologic features and differential diagnosis. Oncologist. 2006;11(4):375-383. 97. Crisan D, ed. Hematopathology: Genomic Mechanisms of Neoplastic Diseases. New York: Humana Press; 2010. 98. Sun T, Nordberg ML, Cotelingam JD, et al. Fluorescence in situ hybridization: method of choice for a definitive diagnosis of mantle cell lymphoma. Am J Hematol. 2003;74(1):78-84. 99. Seto M. Cyclin D1-negative mantle cell lymphoma. Blood. 2013;121(8):1249-1250. 100. Dierlamm J, Baens M, Wlodarska I, et al. The apoptosis inhibitor gene API2 and a novel 18q gene, MLT, are recurrently rearranged in the t(11;18) (q21;q21) associated with mucosa-associated lymphoid tissue lymphomas. Blood. 1999;93(11):3601-3609.

166 Cancer Control April 2015, Vol. 22, No. 2 The ability to isolate and molecularly

analyze circulating tumor cells

is becoming a reality.

The Closest Exit May Be Behind You. Photograph courtesy of Craig Damlo. www.soapboxrocket.com.

Circulating Tumor Cells: A Window Into Tumor Development and Therapeutic Effectiveness Gisela Cáceres, PhD, John A. Puskas, PhD, and Anthony M. Magliocco, MD

Background: Circulating tumor cells (CTCs) are an important diagnostic tool for understanding the metastatic process and the development of cancer. Methods: This review covers the background, relevance, and potential limitations of CTCs as a measurement of cancer progression and how information derived from CTCs may affect treatment efficacy. It also highlights the difficulties of characterizing these rare cells due to the limited cell surface molecules unique to CTCs and each particular type of cancer. Results: The analysis of cancer in real time, through the measure of the number of CTCs in a “liquid” biopsy specimen, gives us the ability to monitor the therapeutic efficacy of treatments and possibly the metastatic potential of a tumor. Conclusions: Through novel and innovative techniques yielding encouraging results, including microfluidic techniques, isolating and molecularly analyzing CTCs are becoming a reality. CTCs hold promise for understanding how tumors work and potentially aiding in their demise.

Introduction States is attributed to cancer.2 The search for biomark- Cancer is a comprehensive term that includes a group ers that allow early detection and may therefore affect of diseases characterized by host cells growing with- the outcome of the disease is critical. The ability of out control. It has a multifactorial origin and, even cancer cells to spread in the body, producing metas- though many cancers are directly associated with risk tasis, is one of the most relevant characteristics of the factors in modern life, cancer is an ancient disease, disease and the cause of most cancer deaths. Circulat- with the oldest confirmation of metastatic carcinoma ing tumor cells (CTCs) are thought to originate from dating to 1200 bc.1 Cancer is a major health problem primary tumors and, due to evolutionary pressures, worldwide, and 1 of every 4 deaths in the United acquire a genetic heterogeneity that gives them the potential to reach the circulation and colonize distant From the Department of Anatomic Pathology, H. Lee Moffitt Cancer organs, thus gaining access to better nutrients and Center & Research Institute, Tampa, Florida. biological niches to enhance their survival.3,4 Concep- Submitted October 14, 2014; accepted March 12, 2015. tually, this idea is more than 140 years old; in 1874, De Address correspondence to Anthony M. Magliocco, MD, Department Morgan5 proposed that cells from a primary tumor can of Anatomic Pathology, Moffitt Cancer Center, 12902 Magnolia escape and travel through surrounding tissue along Drive, Tampa, FL 33612. E-mail: [email protected] the lymphatics, or blood vessels, to invade new areas. No significant relationships exist between the authors and the 6 companies/organizations whose products or services may be Ashworth was the first person to observe cells in referenced in this article. the peripheral blood that he characterized as “exactly

April 2015, Vol. 22, No. 2 Cancer Control 167 in shape, size, and appearance like to those of the sparse (diluted by billions of other cells) and tumours,” but the first systematic study on smears of possibly unstable in a typical blood sample, they blood from individuals with cancer showing CTCs in pose technological challenges for their detection, 43% of cases was not performed until 1934.7 isolation, and characterization. Therapeutically, CTCs are primarily thought of as Moving beyond the simple isolation and quanti- a “liquid” biopsy specimen — a snapshot of tumor fication of CTCs in blood samples from patients with cells in the circulation at a specific point in time — cancer is the possibility of analyzing the functional, and their presence has been linked to poor survival molecular, and genetic alterations in these cells, and rates in metastatic breast,8-12 colon,13-15 and prostate applying this information to improve cancer treatment cancers.16-20 A major advantage of using CTCs is that in “real time.” If realized, this possibility will revo- a simple blood draw is faster and less invasive than lutionize the analysis, monitoring, and treatment of collecting a tumor sample via a surgical procedure; cancer. Can essential information about a tumor be furthermore, some tumors are in locations that prohib- gathered and understood before it seeds and before it biopsies. In addition, individual CTCs may contain the heterogeneity morphs it into another form that unique genetic information essential to understanding will lend current therapies ineffective? Many reviews the tumor biology and the metastatic process of an cover the different aspects of CTCs, illustrating the individual patient’s tumor. This, of course, is a ma- continued interest in this field of study.15,34-42 The cur- jor oversimplification of the heterogeneous nature of rent review will briefly describe CTCs, and then focus tumors, CTCs, and the complexity of the metastatic on developments to further characterize them and process, particularly when considering that the gen- the utilization of CTCs as biomarkers for cancer. We eral consensus is that CTCs are heterogeneous and will address the potential problems of phenotypical- can differ between different cancers as well as within ly identifying CTCs, isolating them via microfluidic an individual patient.21 techniques, and using next-generation sequencing to CTCs have been garnering attention from re- molecularly characterize CTCs. searchers during the previous decade, and attempts to apply discoveries about individual CTCs to the Relevance cancer clinic for individual patients is a major goal Currently, CTCs are used as a biomarker associated of many researchers and health care professionals. with the status of metastatic cancer. One test alone The presence of CTCs is one of many predictors of has been approved by the US Food and Drug Ad- overall survival rates in patients with early stage22 ministration for enumerating CTCs in metastatic and metastatic breast cancer,8,22-25 and their presence breast, colorectal, and prostate cancers, and that is has shown to have a superior overall survival prog- the CellSearch System (Janssen Diagnostics, Raritan, nostic value compared with serum tumor markers, New Jersey).43 The test enumerates CTCs and is com- such as carcinoembryonic antigen and cancer an- monly used before and after a given treatment. If tigen 15-3.26 CTCs provide equivalent or superior the number of CTCs remains the same or increases, prognostic information as radiographic methods but then the treatment may be ineffective; however, if the without the possible health risks associated with number of CTCs decreases, then the treatment may some of them.27-29 be effective. This concept is important because many Disseminated tumor cells are a subtype of CTCs studies have shown that as few as 3 to 5 CTCs in localized in bone marrow that may also provide prog- 7.5 mL of blood can lead to a poor prognosis in terms nostic information in many types of cancer.15,30-32 A of progression-free and overall survival rates.10,13,16 retrospective study of metastatic breast cancer showed CellSearch identifies CTCs as captured circulating a significantly higher number of tumor cells in bone cells expressing the cell-surface marker epithelial marrow compared with the corresponding peripher- cell adhesion molecule (EpCAM; CD326), cytokeratin al blood stem cells,33 but comparative studies have (CK; 8, 18, and 19), and having a positive result on the showed variable concordance rates.34 However, bone 4’6-diamidino-2-phenylindole (DAPI) nuclear stain. marrow biopsy is invasive and requires special tech- In addition, CTCs must be negative for the common niques; in addition, oftentimes patients experience leukocyte marker CD45, and the morphology (intact significant discomfort during the procedure. Thus, the cell membrane) and size (> 4 × 4 µm2) of the CTCs study of CTCs is more favorable because CTCs can must also be taken into account.13,16,28 Studies with the be obtained through a simple peripheral blood draw. CellSearch System and others have shown that high Currently, the major advantage of liquid bi- numbers of CTCs are associated with shorter dis- opsy is to measure treatment effectiveness. CTCs ease-free and overall survival rates.10,44-51 The presence of have been extended to many types of cancer as 1 or 2 CTCs in patients with nonmalignant disease or biomarkers for malignancies, and this trend is in healthy individuals has been detected, but the fre- likely to continue. However, because CTCs are quency of appearance in these conditions is rare.47,52,53

168 Cancer Control April 2015, Vol. 22, No. 2 Pliability for metastatic dominance.94,95 Currently, Paget’s “seed Most human solid tumors contain an heterogeneous and soil” hypothesis is still valid, but metastasis is population of cancer cells that originate from epi- now recognized as a dynamic process whose out- thelial cell types; many of these are shed on a daily come depends on the interfaces with the surrounding basis into the blood, where they survive a couple of homeostatic mechanisms that will differentiate the hours.54 Some are apoptotic,55,56 whereas others are cell from the original tumor.96,97 Even with metastasis, destroyed by the immune system,57 even though the which is normally assumed as a late-stage process immune system of patients with cancer is significantly after the tumor reaches considerable size, evidence diminished.58,59 But the most relevant patient char- exists that invasion could occur in the early stages of acteristic is that few cells have the potential to form disease and then stay clinically dormant.98-101 new metastases, and these cells are sometimes known as cancer stem cells.38,60-64 However, as documented Novel Isolation Techniques by Plaks et al,65 more studies are necessary to clarify CTCs are difficult to isolate because so few are pres- the relationship between cancer stem cells and CTCs. ent in the general circulation; in addition, blood is a Epithelial tumor cells acquire characteristics that complex liquid tissue composed of billions of red cor- allow them to leave the primary tumor in a process puscles and hundreds of millions of white blood cells known as the epithelial–mesenchymal transition per milliliter, thus presenting technical challenges for (EMT), which is normally part of embryonic devel- the identification and isolation of CTCs. Many com- opment but, in cancer, is associated with an increase mon methods for the enrichment of these rare cells in invasion, metastasis, and drug resistance. The EMT have been previously reviewed, and a summary of process has been extensively reviewed39,66-71 and will these select techniques is shown in the Table.102,103 Be- be touched on here to illustrate the heterogeneity of cause many of these techniques have been reviewed these cells, exemplifying the challenges when iden- elsewhere,103-108 we will focus on novel CTC isolation tifying and isolating CTCs. techniques primarily using microfluidics that decrease The initiation of EMT is directed by signals in the CTC cell loss, increase recovery rates, use small vol- tumor cell and the surrounding non-neoplastic matrix, umes of blood, and enable the molecular analysis but the central characteristic of the EMT is associated of individual CTCs.109 Their current advantages and with a decrease in members of the cadherin fami- disadvantages are also reviewed. ly, particularly E-cadherin expression, which allows Two recent studies evaluated the ability to cap- tumor cells to detach and move to the surrounding ture and genetically analyze CTCs through gravi- areas, lymph nodes, and enter the blood vessel system ty-fed microfluidic devices.110,111 The first used photo- as CTCs.72-77 The overexpression of N-cadherin has lithography and deep-reactive ion etching to create also been observed in prostate cancer and is associ- a micronetwork of cell-trapping chambers of 20 × 25 ated with tumor cell motility and cancer progression × 30 microns containing 8 to 10 micron pore chan- in breast cancer.78,79 Moreover, evidence reveals het- nels.110 From this master silicon template, a soft, elas- erogeneous cadherin switching depending on the site tomeric negative mold was forged by pouring and of origin of the carcinoma.76 Complex communication curing against the silicon master. The final microsub- between different cellular events control the EMT pro- strate was created by hot embossing a plastic plate cess, including transforming growth factor β family made of cyclic olefin polymer against the elastomer- members associated with the plasticity of epithelial ic negative mold. A thin plastic laminate was then cells,80-82 the overexpression of transcription factors, laminated against the cyclic olefin polymer to create such as snail/slug,83-86 zinc finger E-box binding ho- the final microchip that allowed blood cells to flow meobox 2,87 forkhead box C2,88 goosecoid,89 twist through, leaving the larger CTCs in the chamber.110 Using family basic helix-loop-helix transcription factor 1,58 the breast cancer cell lines MCF7, MDA-MB-231, and and activation of members of the Src tyrosine kinase SK-BR-3 spiked into phosphate-buffered solution and family.74 EMT cells in breast cancer show resistance to human blood, the researchers captured more than 80% conventional therapies and surface markers associated of the CTCs.110 They were also able to show the dif- with stem cells CD44+/CD24–/low.90 The surviving CTCs ferential expression of the cancer biomarkers PanCK, travel through the circulatory system to distant organs ERBB2 (formally known as HER2), ER, PI3K, E-cadherin, where the reciprocal process of the mesenchymal–ep- and vimentin in the cell lines tested by immunostaining ithelial transition allows them to invade the tissue.91,92 antibodies and fluorescence microscopy.110 Some tissues may be predisposed to metastasis The second study was conducted by a group from to specific cancers, which Paget93 stated as early as Japan who developed a novel, rapid CTC isolation 1889: “In cases of cancer of the breast, it is strange device using a 3-dimensional palladium filter.111 The how often the liver is the seat of secondary cancer.” filter was produced by microfabrication technology Indeed, the bone, liver, lungs, and brain are all sites and consisted of lithography and electroforming pro-

April 2015, Vol. 22, No. 2 Cancer Control 169 Table. — CTC Isolation Methods Isolation Method Mode of Action Advantages Disadvantages Examples Cell-density separation Separates cells based on Separates cells into discrete CTC size and density not Ficoll different densities after layers uniform Lymphoprep (Stem Cell centrifugation CTCs may be lost in plasma Technologies, Vancouver, or by formation of CTC Canada) aggregates OncoQuick (VWR, Radnor, PA) Cell size by filtration Filtration to remove smaller Simple, 1-step process Not validated for clinical Polycarbonate membrane cells in the blood applications ScreenCell (ScreenCell, Will exclude small-size CTCs Westford, MA) Microfiltration Photolithography using Bottom membrane supports Experimental 3-dimensional membranes separated captured CTCs, minimizing cell Not validated for clinical by a precisely defined gap stress applications Immunomagnetic Antibodies specific to Can be performed alone or with Variable rates of false- MACS epithelial antigens (eg, CK, other methods (eg, qRT-PCR, positive and false-negative RosetteSep (Stem Cell PSMA, EpCAM) FISH, flow cytometry) staining, depending on Technologies) antibodies used MagSweeper (Jeffrey Lab, Stanford, CA) CellSearch (Janssen Diagnostics, Raritan, NJ) AdnaTest (Qiagen, Hannover, Germany) Direct analysis High-throughput assay Uses whole blood Not validated for clinical Fiberoptic array screening applications Micro-Hall sensor Leukapheresis Separates leukocytes by Captures many more CTCs Risky procedure and not passing blood through due to the volume of blood likely to be used more than external filters CTCs amenable to further once characterization and molecular Not validated for clinical analysis applications High-definition CTCs All nucleated cells retained Enrichment-free system resulting Not validated for clinical and stained with fluorescent in high sensitivity and specificity applications antibodies to CK, CD45, rates and DAPI Adds high-definition CTC cytomorphology, allowing multiple analysis parameter characterization Fluorescence-activated Flow cytometry–based Allows for semiautomated Not validated for clinical cell-sorting analysis approach of sorting a mixed isolation of single cells or cell applications population of cells using pools with less operator inter- fluorescent anti-EpCAM vention and higher throughput antibodies Micromanipulation and Manual microscopy-based Can isolate individual cells Can be harsh on fragile, laser microdissection techniques used to identify sticky cells during transfer and isolate antibody-labeled Not validated for clinical cells using a micromanipu- applications lator or micropipette CK = cytokeratin, CTC = circulating tumor cell, DAPI = 4’6-diamidino-2-phenylindole, EpCAM = epithelial cell adhesion molecule, FISH = fluorescence in situ hybridization, MACS = magnetic-activated cell sorting, PSMA = prostate-specific membrane antigen, qRT-PCR = quantitative reverse transcription–polymerase chain reaction. cesses forming 8 micron pores that allowed normal suggesting that the filter may have clinical utility.111 blood cells and leukocytes to flow through while larg- Using COLM-5 cells, the researchers also showed that er CTCs remained in the pocket. Cell-spiking experi- further analysis could be performed on the filter us- ments showed a 90% recovery rate across a range of ing immunohistochemistry for ERBB2, ER, and PGR, 50 to 5,000 green fluorescent protein-tagged COLM-5 and genetic analysis for EGFR/KRAS mutations by tumor cells.111 The filter detected and enumerated polymerase chain reaction (PCR), whole genome am- CTCs from 19 individuals with metastatic breast can- plification, and direct sequencing.111 cer, 12 individuals with nonmetastatic breast cancer, The major advantages of these techniques are the and 12 healthy volunteers.111 The average numbers capture of EpCAM-negative CTCs and the capability of of CTCs detected were 3.37, 0.23, and 0, respectively, enabling single CTC genomic analysis. Two major dis-

170 Cancer Control April 2015, Vol. 22, No. 2 advantages of these methods are that the US Food and whole blood.116 This method of CTC visual microscop- Drug Administration has not yet approved them for ic enumeration was also amenable to isolation via use and they lack the validation required to be used trypsinization and high flow rate washing, recovering in a clinical setting. In addition, size filtration–based more than 60% of the CTCs with a high viability for approaches similar to the 2 reviewed here can have either CTC culture or further characterization.116 How- significant red and white blood cell contamination if ever, a drawback to this method was that it showed the pores are plugged with CTCs.109 This could hinder high levels of contaminating leukocytes, especially downstream genetic analysis by increasing the DNA from patients with pancreatic cancer (> 24,000/mL) background contamination. compared with healthy donors (~ 3,500/mL).116 Because EpCAM expression may be lost during An improvement of existing technology uses mi- EMT, microfluidic devices are being developed to cap- croscale vortices (Vortex Chip [University of Califor- ture CTCs — in particular, EpCAM-negative CTCs — by nia, Los Angeles]) and inertial focusing (which uses other means. Galletti et al112 developed a geometrically fluid inertia through shaped microchannels to align enhanced, differential immunocapture microfluidic cells at high flow rates) for the high-purity isolation device using anti-ERBB2 antibody-coated microposts of the relatively larger CTCs from whole blood. By to isolate CTCs (DAPI+, CK+, CD45–) from patients with changing the flow rates, channel heights, and length breast and gastric cancers. Notably, ERBB2 expression of the Vortex Chip, researchers optimized the system is correlated with poor prognosis and plays an im- with 19 micron diameter particles.117 Using 7.5 mL portant role in the metastasis and invasion of breast of blood from 8 individuals with breast and lung cancer to surrounding tissues during EMT.113 The re- cancers, they isolated CTCs (23–317 per 7.5 mL) and searchers optimized their system using breast cancer reduced leukocyte contamination (57%–94% purity).117 cell lines and tested it in individuals both positive and The major advantages of this improved technology negative for ERBB2 and found CTCs.112 This may be over current technologies are reduced processing time an alternative method for evaluating EpCAM-negative (20 minutes), the applicability of the system to other CTCs. The same system has also been developed for types of cancer, and maintaining the integrity of CTCs prostate cancer using a monoclonal, anti–prostate-spe- for downstream molecular analysis.117 cific membrane antigen antibody.114 Which one of these isolation methods, if any, will A novel, high throughput spiral microfluidic de- be appropriate for health care professionals evaluating vice with a trapezoidal cross-section that relies on patients with cancer remains an open question. What is the microchannel dimensions and fluidic forces for clear is that the isolation of these rare cells is impera- the size-based separation of CTCs was developed and tive for the molecular analysis of CTCs and, ultimately, tested in patients with metastatic breast and lung can- the personalized patient care for all cancer types. cers.115 The device has a low cost and can process 7.5 mL of peripheral blood in 8 minutes, and its Limitations capture efficiency rate is approximately 80% using The main limitation of the CellSearch System is that spiked MCF7, T24, and MDA-MB-231 cell lines.115 the isolation method utilizes EpCAM expression on Analysis of patient samples revealed CD24+CD44– and the cell surface of the tumor,118,119 which is expressed CD24–CD44+ cancer stem cell subpopulations and Ep- in 75% of cancer types alone (and may be lost in un- CAM–pan-CK+ cells, which are not currently identified differentiated tumors or subclones).120 In renal cell in the traditional EpCAM-positive CTC capture meth- carcinoma subgroups, EpCAM is expressed in chro- ods.115 The system is also amenable to fluorescence mophobe, papillary, and oncocytoma types but is sig- in situ hybridization and other molecular techniques nificantly absent in clear cell carcinomas,121 whereas following CTC isolation.115 others report decreased EpCAM expression in me- Sheng et al116 developed a geometrically enhanced, tastases.122 Similarly, anaplastic thyroid carcinomas mixing, high-performance microchip for the efficient lack EpCAM expression, whereas differentiated and capture of pure CTCs. After testing with anti-EpCAM poorly differentiated thyroid carcinomas show over- antibody–coated surfaces within the specialized stag- expression.123 Brain metastasis from breast cancer has gered herringbone grooves in the microchip, they been found without the presence of EpCAM-positive found an optimal flow rate of 1 µL/second, a flow CTCs.124,125 Similarly, in different human carcinomas, velocity of 0.75 mm/second, and a maximum sheer an approximately 10-fold lower rate of EpCAM ex- stress of 0.38 dyn/cm.116 The researchers achieved a pression has been found in CTCs than in primary and capture efficiency rate of 90% of the EpCAM-posi- metastatic tissues.126 A significant association exists tive human pancreatic cancer cell line L3.6pl.116 After between positive EpCAM expression and survival rate spiking between 50 and 50,000 labeled L3.6pl cells in esophageal cancer,127 clear-cell renal cell carcino- into lysed or whole blood, they achieved a 92% cap- ma,122,128 moderately differentiated colon cancer,129 ture rate from lysed and a 89% capture rate from and thyroid carcinoma123; however, EpCAM expres-

April 2015, Vol. 22, No. 2 Cancer Control 171 sion has also been associated with a worse prognosis to take shape and become available. The technology in different types of breast cancer,130-132 pancreatic now exists to look into the genome of individual cancer,133 gallbladder carcinoma,134 esophageal squa- CTCs, potentially discovering the genetic drivers that mous cell carcinoma,135 and urothelial carcinoma of allow tumors to progress and evade therapy. Learning the bladder.136 For epithelial ovarian cancers,137 lung more about the genetic makeup of a tumor may allow cancer,129,30 prostate cancer, and well and poorly dif- researchers to discover new and potentially import- ferentiated colon cancers, no clear correlation with ant driver mutations, give biotechnology investigators survival rates has been observed (Fig).129 These con- the information they need to develop novel drugs tradictory data highlight the need to discover other that target those mutations, and inform health care phenotypic markers to more precisely identify CTCs. professionals on how to better treat and cure more The CellSearch method could use alternative capture patients with cancer. approaches beyond antigens to EpCAM, but these However, this is a formable task; simply knowing methods would be considered experimental ap- that a DNA mutation exists does not necessarily mean proaches and not readily available to patients. it is important for tumor development or that it is the only mutation within the tumor. Temporal, spatial, and Molecular Characterization biological considerations, among many others, may Many methods are used to molecularly characterize affect the way an individual tumor is developing, but CTCs, including those that are protein- or nucleic ac- even this view may be too narrow. There are cave- id-based (eg, reverse transcription–PCR), as well as mi- ats. For example, given that primary and metastatic croarray analysis and sequencing. These methods and tumors as well as CTCs are generally heterogeneous, the variety of iterations within each general method how many individual CTCs should be characterized to have been reviewed in detail by Lowes et al.138 Thus, achieve a reasonably confident estimate of the make- this review focuses instead on massively parallel or up of CTCs in the circulation? Which genetic profile — next-generation sequencing and the practical appli- if there are multiple variations as there undoubtedly cations pertaining to patient care and personalized will be — represents the most meaningful mutational medicine. “snapshot” of the tumor? Which driver mutations will The method of molecular characterization is gen- give individual CTCs the selective advantage to seed erating interest because of the multiplexing capability and create new metastases? These and many other and the generation of clonal evolution information questions must first be answered. that may point to a key player in the progression In a molecular analysis of comprehensive genom- of cancer.139-142 After enumeration and classification, ic profiling of CTCs from patients with stage 4 col- delving into the genetic makeup of CTCs may be orectal carcinoma using array comparative genomic important because such data may provide additional hybridization (CGH) and next-generation sequencing, information on the mutational status of the tumor. researchers discovered that the primary tumor differs Although the technologies used to isolate individual from a liver metastatic tumor, which also differs from CTCs are technically demanding, they are beginning CTCs.143 When deep sequencing the primary tumor, they determined that mutations thought to be unique to CTCs instead originated from small subpopulations within the primary tumor.143 GOOD UNCLEAR WORSE As proof of concept, Lohr et al144 established a Prognosis Correlation Prognosis standardized process for isolating, qualifying, and retrieving pure CTCs in a study of metastatic prostate cancer; they also generated and quantified independent libraries for whole exome sequencing from RCC Ovarian Breast magnetically isolated CTCs. This approach looked to Esophagus Lung Pancreas find mutations in early tumor evolution (early trunk) Thyroid Prostate Gallbladder or metastatic precursor (metastatic trunk) mutations but did not emphasize private, somatic, single-nu- High EpCAM Expression cleotide variant mutations from single CTCs. Rather, the researchers used consensus-based variant calling to obtain a combined 19-CTC composite of the CTC 144 Fig. — High levels of EpCAM expression in tumor cells are controversial. population. The process worked well for individuals EpCAM expression is associated with a good prognosis in RCC, esophageal with cancer who had more than 5 CTCs per 3.75 mL of cancer, and thyroid cancer, but with a worse prognosis in breast, pancreatic, blood and generated high-quality libraries for whole and gallbladder cancers. EpCAM expression has no clear correlation with 144 prognosis in epithelial ovarian, lung, and prostate cancers. EpCAM = epithelial exome sequencing. The researchers were also able cell adhesion molecule, RCC = renal cell carcinoma. to identify 10 early trunk mutations, including TP53,

172 Cancer Control April 2015, Vol. 22, No. 2 found in the primary tumor and metastasis; 90% of could suggest the possibility that CTC CNV analysis these were found in the CTCs analyzed.144 This finding may be a biomarker for lung adenocarcinoma. implies that the cancer came from a single ancestor and It is likely that the genetic analysis of CTCs will then diverged. Lohr et al144 also found 56 metastatic be an important area in the development of biomark- trunk mutations present in the primary tumor and me- ers for cancer. However, questions still remain about tastasis, and 73% of these were also found in the CTCs. whether circulating tumor DNA (ctDNA) may be a Magbanua et al145 developed a new CTC isolation better or more useful biomarker than CTC analysis. method based on immunomagnetic enrichment via the Evidence is accumulating, and analysis of 62 normal CellSearch System and fluorescent-activated cell sorting and clinically matched samples showed that CTC DNA followed by genomic analysis, including profiled copy returned actionable mutations 3 times more often than number variation (CNV) by CGH and whole genome ctDNA, and ctDNA produced 7 times more false-posi- amplification. CTCs from 9 individuals with castrate-re- tive results and 50 times more variation in sensitivity sistant prostate cancer were successfully profiled and rates than CTC DNA.151 These are preliminary data; had multiple copy number aberrations, including gains more studies are needed before such a question can in 8q and losses in 8p, as previously reported.146,147 The be fully answered. androgen receptor locus had high CNV in 78% of the However, it is likely that ctDNA will be a comple- cases.145 Similarly, in metastatic breast cancers, CTCs in mentary assay to CTCs, providing information on the 102 of 181 individuals were profiled and revealed a global tumor load in a patient and the indicators of wide range in CNV; serial testing of the CTCs was also the genomic conditions of active and dormant cancer performed that confirmed reproducibility and indicat- cells undergoing apoptosis. In this setting, CTCs more ed genomic changes over time.148 In addition, an array likely represent a real-time glimpse of the biological CGH analysis confirmed that primary tumors and nodal and functional condition of viable malignant cells and metastasis were highly correlated with each other but may provide opportunities for a pharmacodynamic less so with CTCs, suggesting that a CTC copy number evaluation and rapid treatment switching to optimize analysis might provide information about the progres- therapy for patients with cancer. sion of the tumor in real time.148 This can be advanta- geous for health care professionals who may need to Conclusions change therapeutic interventions based on the genetic Circulating tumor cells (CTCs) play a key role in the signature of the CTCs. metastatic process. Ever since the CellSearch System In a study of individuals with colorectal cancer, was first approved by the US Food and Drug Admin- a group of researchers isolated single CTCs using a istration for the enumeration of these rare cells and micromanipulator and showed by array CGH, muta- was shown to have prognostic value, emphasis has tional profiling, and microsatellite instability analysis been placed on using these cells to further evaluate that heterogeneity exists between tissue and CTCs tumors. The ability to have a noninvasive, therapeu- and single CTCs.149 In addition, transcriptional analy- tically relevant diagnostic test has great appeal for sis revealed a down regulation of epithelial markers health care professionals and patients alike. Discover- (EpCAM, CK19, and carcinoembryonic antigen), CTC ing the genetic potential of these cells will no doubt dormancy (reduced Ki67 and c-Myc expression), and lead to a better understanding of the evolution of an immune escape phenotype (overexpression of tumors in patients with cancer. CD47, an antiphagocytic signal expressed on cancer The future of CTC research is bright. As research- cells that signals to macrophages and dendritic cells) ers develop more precise CTC phenotypic markers, compared with the corresponding tumor tissue.149 This such as ERBB2, EGFR, heparanase, and notch1, a analysis highlighted the survival mechanisms of CTCs signature found in CTCs of metastatic breast tumors and their ability to escape the immune response against that colonize the brain,124 the ability to further de- tumors, a signature event in the metastatic process. lineate CTC subsets will become a reality. Although In patients with lung cancer, reproducible CNV the concept of CTCs is not a novel one, this field of patterns can be found from single CTCs using multiple study is still in its infancy. With the development of annealing and looping-based amplification cycles.150 better isolation methods, coupled with improvements Study findings suggest that gains and losses of copy in technologies that allow for the genetic character- numbers at certain chromosome regions during me- ization of CTCs, new discoveries about how tumors tastasis are selected for and supply the tumor cells progress, their metastatic potential, and the time-de- with a survival advantage in the circulatory system.150 pendent nature of tumor progression will most cer- The study researchers also found that a small number tainly occur. However, to achieve reliable and compa- of individuals with lung adenocarcinoma (N = 4) ex- rable results, researchers must standardize the types hibited nearly identical global CNV patterns (78% of of isolation methods they use. It is likely that different the gain and loss regions were shared).150 This finding isolations methods, with unique limitations, could

April 2015, Vol. 22, No. 2 Cancer Control 173 12. Rack B, Schindlbeck C, Jückstock J, et al; SUCCESS Study Group. give different results. Circulating tumor cells predict survival in early average-to-high risk breast Many more questions still need answers to realize cancer patients [Erratum appears in J Natl Cancer Inst. 2014;106(8):dju273]. J Natl Cancer Inst. 2014;106(5):dju066. the potential of CTCs. Because CTCs are a heteroge- 13. Cohen SJ, Punt CJ, Iannotti N, et al. Relationship of circulating tumor neous population of cells, single-cell genetic analysis cells to tumor response, progression-free survival, and overall survival in patients with metastatic colorectal cancer. J Clin Oncol. 2008;26(19):3213-3221. will be required, as will searching the profile(s), de- 14. Tol J, Koopman M, Miller MC, et al. Circulating tumour cells early predict termining the metastatic potential of the tumor cell, progression-free and overall survival in advanced colorectal cancer patients treated with chemotherapy and targeted agents. Ann Oncol. 2010;21(5):1006-1012. and understanding the copy number variations of 15. Steinert G, Schölch S, Koch M, et al. Biology and significance of circu- specific genes known to be critical in CTCs. In their lating and disseminated tumour cells in colorectal cancer. Langenbecks Arch 150 Surg. 2012;397(4):535-542. study of lung cancer, Ni et al illustrated that all 16. de Bono JS, Scher HI, Montgomery RB, et al. Circulating tumor cells CTCs from a given individual, irrespective of cancer predict survival benefit from treatment in metastatic castration-resistant prostate cancer. Clin Cancer Res. 2008;14(19):6302-6309. subtype, showed reproducible copy number varia- 17. Scher HI, Jia X, de Bono JS, et al. Circulating tumour cells as prognostic tion patterns similar to the metastatic tumor in the markers in progressive, castration-resistant prostate cancer: a reanalysis of IMMC38 trial data. Lancet Oncol. 2009;10(3):233-239. same individual. 18. Olmos D, Arkenau HT, Ang JE, et al. Circulating tumour cell (CTC) An intriguing advantage of the liquid biopsy and counts as intermediate end points in castration-resistant prostate cancer (CRPC): a single-centre experience. Ann Oncol. 2009;20(1):27-33. CTC analysis is the ability to serially sample and mon- 19. Danila DC, Heller G, Gignac GA, et al. Circulating tumor cell number itor in real time. Until now, health care profession- and prognosis in progressive castration-resistant prostate cancer. Clin Cancer Res. 2007;13(23):7053-7058. als have seen snapshots of the tumor, instead of the 20. Goldkorn A, Ely B, Quinn DI, et al. Circulating tumor cell counts are full picture, and the “average” profile. Monitoring the prognostic of overall survival in SWOG S0421: a phase III trial of docetaxel with or without atrasentan for metastatic castration-resistant prostate cancer. progression of cancer is a significant advantage for J Clin Oncol. 2014;32(11):1136-1142. health care professionals who may change therapeu- 21. van de Stolpe A, Pantel K, Sleijfer S, et al. Circulating tumor cell isolation and diagnostics: toward routine clinical use. Cancer Res. 2011;71(18):5955-5960. tic interventions in their patients based on real-time 22. Zhang L, Riethdorf S, Wu G, et al. Meta-analysis of the prognostic value of information. If a treatment course has failed (based circulating tumor cells in breast cancer. Clin Cancer Res. 2012;18(20):5701-5710. 23. Bidard FC, Hajage D, Bachelot T, et al. Assessment of circulating tumor on the elevated CTC level), then the patient could be cells and serum markers for progression-free survival prediction in meta- switched to another therapy earlier. static breast cancer: a prospective observational study. Breast Cancer Res. 2012;14(1):R29. In addition, detailing the genetic makeup of CTCs 24. Nakamura S, Yagata H, Ohno S, et al. Multi-center study evaluating over time may allow personalized medicine to be- circulating tumor cells as a surrogate for response to treatment and overall survival in metastatic breast cancer. Breast Cancer. 2010;17(3):199-204. come a reality by addressing specific mutations rather 25. Smerage JB, Barlow WE, Hortobagyi GN, et al. Circulating tumor cells than using chemotherapeutic approaches that do not and response to chemotherapy in metastatic breast cancer: SWOG S0500. J Clin Oncol. 2014;32(31):3483-3489. selectively kill all tumor cells. It is unlikely that an 26. Pierga JY, Hajage D, Bachelot T, et al. High independent prognostic analysis of CTCs will be the answer to all of the bio- and predictive value of circulating tumor cells compared with serum tumor markers in a large prospective trial in first-line chemotherapy for metastatic logical questions related to tumors, but it is possible breast cancer patients. Ann Oncol. 2012;23(3):618-624. that analyzing CTCs may unlock enough information 27. De Giorgi U, Valero V, Rohren E, et al. Circulating tumor cells and [18F]fluorodeoxyglucose positron emission tomography/computed tomog- about them to arm health care professionals with raphy for outcome prediction in metastatic breast cancer. J Clin Oncol. the data needed to improve the quality, efficacy, and 2009;27(20):3303-3311. 28. Budd GT, Cristofanilli M, Ellis MJ, et al. Circulating tumor cells versus effectiveness of therapies aimed at eradicating cancer. imaging--predicting overall survival in metastatic breast cancer. Clin Cancer Res. 2006;12(21):6403-6409. 29. Liu MC, Shields PG, Warren RD, et al. Circulating tumor cells: a use- References ful predictor of treatment efficacy in metastatic breast cancer. J Clin Oncol. 1. Binder M, Roberts C, Spencer N, et al. On the antiquity of cancer: 2009;27(31):5153-5159. evidence for metastatic carcinoma in a young man from ancient Nubia 30. Chiang AC, Massagué J. Molecular basis of metastasis. N Engl J Med. (c. 1200 BC). PLoS One. 2014;9(3):e90924. 2008;359(26):2814-2823. 2. American Cancer Society. Cancer Facts & Figures 2015. Atlanta, GA: 31. Vashist YK, Effenberger KE, Vettorazzi E, et al. Disseminated tumor American Cancer Society; 2015. http://www.cancer.org/acs/groups/content cells in bone marrow and the natural course of resected esophageal cancer. /@editorial/documents/document/acspc-044552.pdf. Accessed April 7, 2015. Ann Surg. 2012;255(6):1105-1112. 3. Gupta GP, Massagué J. Cancer metastasis: building a framework. Cell. 32. Kim MY, Oskarsson T, Acharyya S, et al. Tumor self-seeding by circu- 2006;127(4):679-695. lating cancer cells. Cell. 2009;139(7):1315-1326. 4. Hanahan D, Weinberg RA. The hallmarks of cancer. Cell. 2000;100(1):57-70. 33. Ross AA, Cooper BW, Lazarus HM, et al. Detection and viability of 5. De Morgan C. Observations on cancer: its pathology, and its relations tumor cells in peripheral blood stem cell collections from breast cancer pa- to the organism and to other morbid growths. Lancet. 1874;103(2636):325-329. tients using immunocytochemical and clonogenic assay techniques. Blood. 6. Ashworth TR. A case of cancer in which cells similar to those in the 1993;82(9):2605-2610. tumors were seen in the blood after death. Aust Med J. 1869;14:146-147. 34. Bednarz-Knoll N, Alix-Panabières C, Pantel K. Clinical relevance and 7. Pool EH, Dunlop GR. Cancer cells in the blood stream. Am J Cancer. biology of circulating tumor cells. Breast Cancer Res. 2011;13(6):228. 1934;21:99-102. 35. Hogan BV, Peter MB, Shenoy H, et al. Circulating tumour cells in breast 8. Cristofanilli M, Budd GT, Ellis MJ, et al. Circulating tumor cells, dis- cancer: prognostic indicators, metastatic intermediates, or irrelevant bystand- ease progression, and survival in metastatic breast cancer. N Engl J Med. ers? Mol Med Rep. 2008;1(6):775-779. 2004;351(8):781-791. 36. Hou JM, Krebs M, Ward T, et al. Circulating tumor cells, enumeration 9. Cristofanilli M, Hayes DF, Budd GT, et al. Circulating tumor cells: a novel and beyond. Cancers (Basel). 2010;2(2):1236-1250. prognostic factor for newly diagnosed metastatic breast cancer [Erratum appears 37. Paterlini-Brechot P, Benali NL. Circulating tumor cells (CTC) detection: in J Clin Oncol. 2005;23(21):4808]. J Clin Oncol. 2005;23(7):1420-1430. clinical impact and future directions. Cancer Lett. 2007;253(2):180-204. 10. Hayes DF, Cristofanilli M, Budd GT, et al. Circulating tumor cells at 38. Krebs MG, Hou JM, Ward TH, et al. Circulating tumour cells: their each follow-up time point during therapy of metastatic breast cancer patients utility in cancer management and predicting outcomes. Ther Adv Med Oncol. predict progression-free and overall survival. Clin Cancer Res. 2006;12(14 pt 2010;2(6):351-365. 1):4218-4224. 39. Bednarz-Knoll N, Alix-Panabières C, Pantel K. Plasticity of disseminat- 11. Nolé F, Munzone E, Zorzino L, et al. Variation of circulating tumor cell ing cancer cells in patients with epithelial malignancies. Cancer Metastasis levels during treatment of metastatic breast cancer: prognostic and therapeutic Rev. 2012;31(3-4):673-687. implications. Ann Oncol. 2008;19(5):891-897. 40. Danila DC, Pantel K, Fleisher M, et al. Circulating tumors cells as biomark-

174 Cancer Control April 2015, Vol. 22, No. 2 ers: progress toward biomarker qualification. Cancer J. 2011;17(6):438-450. 73. Vincent-Salomon A, Thiery JP. Host microenvironment in breast cancer 41. Barradas AM, Terstappen LW. Towards the biological understanding development: epithelial-mesenchymal transition in breast cancer development. of CTC: capture technologies, definitions and potential to create metastasis. Breast Cancer Res. 2003;5(2):101-106. Cancers (Basel). 2013;5(4):1619-1642. 74. Fujita Y, Krause G, Scheffner M, et al. Hakai, a c-Cbl-like protein, ubiq- 42. O’Flaherty JD, Gray S, Richard D, et al. Circulating tumour cells, uitinates and induces endocytosis of the E-cadherin complex. Nat Cell Biol. their role in metastasis and their clinical utility in lung cancer. Lung Cancer. 2002;4(3):222-231. 2012;76(1):19-25. 75. Pece S, Gutkind JS. E-cadherin and Hakai: signalling, remodeling or 43. Food and Drug Administration; HHS. Medical devices; immunology destruction? Nat Cell Biol. 2002;4(4):E72-74. and microbiology devices; classification of the immunomagnetic circula- 76. Peinado H, Portillo F, Cano A. Transcriptional regulation of cadherins tin cancer cell selection and enumeration system. Final rule. Fed Regist. during development and carcinogenesis. Int J Dev Biol. 2004;48(5-6):365-375. 2004;69(91):26036-26038. 77. Christofori G, Semb H. The role of the cell-adhesion molecule E-cad- 44. Riethdorf S, Fritsche H, Müller V, et al. Detection of circulating tumor herin as a tumour-suppressor gene. Trends Biochem Sci. 1999;24(2):73-76. cells in peripheral blood of patients with metastatic breast cancer: a validation 78. Tomita K, van Bokhoven A, van Leenders GJ, et al. Cadherin switching study of the CellSearch system. Clin Cancer Res. 2007;13(3):920-928. in human prostate cancer progression. Cancer Res. 2000;60(13):3650-3654. 45. Lloyd JM, McIver CM, Stephenson SA, et al. Identification of early-stage 79. Nieman MT, Prudoff RS, Johnson KR, et al. N-cadherin promotes mo- colorectal cancer patients at risk of relapse post-resection by immunobead tility in human breast cancer cells regardless of their E-cadherin expression. reverse transcription-PCR analysis of peritoneal lavage fluid for malignant J Cell Biol. 1999;147(3):631-644. cells. Clin Cancer Res. 2006;12(2):417-423. 80. Zavadil J, Cermak L, Soto-Nieves N, et al. Integration of TGF-beta 46. Hardingham JE, Kotasek D, Sage RE, et al. Detection of circulating /Smad and Jagged1/Notch signalling in epithelial-to-mesenchymal transition. tumor cells in colorectal cancer by immunobead-PCR is a sensitive prognostic EMBO J. 2004;23(5):1155-1165. marker for relapse of disease. Mol Med. 1995;1(7):789-794. 81. Medici D, Hay ED, Goodenough DA. Cooperation between snail and 47. Hardingham JE, Hewett PJ, Sage RE, et al. Molecular detection of LEF-1 transcription factors is essential for TGF-beta1-induced epithelial-mes- blood-borne epithelial cells in colorectal cancer patients and in patients with enchymal transition. Mol Biol Cell. 2006;17(4):1871-1879. benign bowel disease. Int J Cancer. 2000;89(1):8-13. 82. Zavadil J, Böttinger EP. TGF-beta and epithelial-to-mesenchymal tran- 48. Cohen SJ, Punt CJ, Iannotti N, et al. Prognostic significance of circu- sitions. Oncogene. 2005;24(37):5764-5774. lating tumor cells in patients with metastatic colorectal cancer. Ann Oncol. 83. Cano A, Pérez-Moreno MA, Rodrigo I, et al. The transcription factor 2009;20(7):1223-1229. snail controls epithelial-mesenchymal transitions by repressing E-cadherin 49. Krebs MG, Sloane R, Priest L, et al. Evaluation and prognostic signif- expression. Nat Cell Biol. 2000;2(2):76-83. icance of circulating tumor cells in patients with non-small-cell lung cancer. 84. Hajra KM, Chen DY, Fearon ER. The SLUG zinc-finger protein represses J Clin Oncol. 2011;29(12):1556-1563. E-cadherin in breast cancer. Cancer Res. 2002;62(6):1613-1618. 50. Hou JM, Krebs MG, Lancashire L, et al. Clinical significance and molec- 85. Rosivatz E, Becker I, Specht K, et al. Differential expression of the ular characteristics of circulating tumor cells and circulating tumor microemboli epithelial-mesenchymal transition regulators snail, SIP1, and twist in gastric in patients with small-cell lung cancer. J Clin Oncol. 2012;30(5):525-532. cancer. Am J Pathol. 2002;161(5):1881-1891. 51. Miller MC, Doyle GV, Terstappen LW. Significance of circulating tumor 86. Bolós V, Peinado H, Pérez-Moreno MA, et al. The transcription factor cells detected by the CellSearch System in patients with metastatic breast Slug represses E-cadherin expression and induces epithelial to mesenchymal colorectal and prostate cancer. J Oncol. 2010;2010:617421. transitions: a comparison with Snail and E47 repressors. J Cell Sci. 2003;116(pt 52. Allard WJ, Matera J, Miller MC, et al. Tumor cells circulate in the pe- 3):499-511. ripheral blood of all major carcinomas but not in healthy subjects or patients 87. Comijn J, Berx G, Vermassen P, et al. The two-handed E box binding with nonmalignant diseases. Clin Cancer Res. 2004;10(20):6897-6904. zinc finger protein SIP1 downregulates E-cadherin and induces invasion. Mol 53. Pantel K, Denève E, Nocca D, et al. Circulating epithelial cells in patients Cell. 2001;7(6):1267-1278. with benign colon diseases. Clin Chem. 2012;58(5):936-940. 88. Mani SA, Yang J, Brooks M, et al. Mesenchyme Forkhead 1 (FOXC2) 54. Meng S, Tripathy D, Frenkel EP, et al. Circulating tumor cells in patients plays a key role in metastasis and is associated with aggressive basal-like with breast cancer dormancy. Clin Cancer Res. 2004;10(24):8152-8162. breast cancers. Proc Natl Acad Sci U S A. 2007;104(24):10069-10074. 55. Méhes G, Witt A, Kubista E, et al. Circulating breast cancer cells are 89. Hartwell KA, Muir B, Reinhardt F, et al. The Spemann organizer frequently apoptotic. Am J Pathol. 2001;159(1):17-20. gene, Goosecoid, promotes tumor metastasis. Proc Natl Acad Sci U S A. 56. Larson CJ, Moreno JG, Pienta KJ, et al. Apoptosis of circulating tumor 2006;103(50):18969-18974. cells in prostate cancer patients. Cytometry A. 2004;62(1):46-53. 90. May CD, Sphyris N, Evans KW, et al. Epithelial-mesenchymal transition 57. Hanna N. Inhibition of experimental tumor metastasis by selective and cancer stem cells: a dangerously dynamic duo in breast cancer progres- activation of natural killer cells. Cancer Res. 1982;42(4):1337-1342. sion. Breast Cancer Res. 2011;13(1):202. 58. Hoek K, Rimm DL, Williams KR, et al. Expression profiling reveals 91. Ocaña OH, Córcoles R, Fabra A, et al. Metastatic colonization requires novel pathways in the transformation of melanocytes to melanomas. Cancer the repression of the epithelial-mesenchymal transition inducer Prrx1. Cancer Res. 2004;64(15):5270-5282. Cell. 2012;22(6):709-724. 59. Green TL, Cruse JM, Lewis RE, et al. Circulating tumor cells (CTCs) 92. Brabletz T. EMT and MET in metastasis: where are the cancer stem from metastatic breast cancer patients linked to decreased immune function cells? Cancer Cell. 2012;22(6):699-701. and response to treatment. Exp Mol Pathol. 2013;95(2):174-179. 93. Paget S. Distribution of secondary growth in cancer of the breast. 60. Al-Hajj M, Wicha MS, Benito-Hernandez A, et al. Prospective iden- Lancet. 1889;1:99-101. tification of tumorigenic breast cancer cells. Proc Natl Acad Sci U S A. 94. Pantel K, Brakenhoff RH. Dissecting the metastatic cascade. Nat Rev 2003;100(7):3983-3988. Cancer. 2004;4(6):448-456. 61. Southam CM, Brunschwig A. Quantitative studies of autotransplantation 95. Hess KR, Varadhachary GR, Taylor SH, et al. Metastatic patterns in of human cancer. Preliminary report. Cancer. 1961;14(5):971-978. adenocarcinoma. Cancer. 2006;106(7):1624-1633. 62. Al-Hajj M, Clarke MF. Self-renewal and solid tumor stem cells. Onco- 96. Fidler IJ. The pathogenesis of cancer metastasis: the ‘seed and soil’ gene. 2004;23(43):7274-7282. hypothesis revisited. Nat Rev Cancer. 2003;3(6):453-458. 63. Fidler IJ. Metastasis: quantitative analysis of distribution and fate of 97. Kang Y, Siegel PM, Shu W, et al. A multigenic program mediating breast tumor embolilabeled with 125 I-5-iodo-2’-deoxyuridine. J Natl Cancer Inst. cancer metastasis to bone. Cancer Cell. 2003;3(6):537-549. 1970;45(4):773-782. 98. Wittekind C, Neid M. Cancer invasion and metastasis. Oncology. 64. Chaffer CL, Weinberg RA. A perspective on cancer cell metastasis. 2005;69(suppl 1):14-16. Science. 2011;331(6024):1559-1564. 99. Schmidt-Kittler O, Ragg T, Daskalakis A, et al. From latent dissemi- 65. Plaks V, Koopman CD, Werb Z. Cancer. Circulating tumor cells. Science. nated cells to overt metastasis: genetic analysis of systemic breast cancer 2013;341(6151):1186-1188. progression. Proc Natl Acad Sci U S A. 2003;100(13):7737-7742. 66. Thiery JP, Sleeman JP. Complex networks orchestrate epithelial-mes- 100. Bernards R, Weinberg RA. A progression puzzle. Nature. enchymal transitions. Nat Rev Mol Cell Biol. 2006;7(2):131-142. 2002;418(6900):823. 67. Yang J, Weinberg RA. Epithelial-mesenchymal transition: at the cross- 101. Uhr JW, Pantel K. Controversies in clinical cancer dormancy. Proc Natl roads of development and tumor metastasis. Dev Cell. 2008;14(6):818-829. Acad Sci U S A. 2011;108(30):12396-12400. 68. Thiery JP. Epithelial-mesenchymal transitions in tumour progression. 102. Danova M, Torchio M, Mazzini G. Isolation of rare circulating tumor Nat Rev Cancer. 2002;2(6):442-454. cells in cancer patients: technical aspects and clinical implications. Expert 69. Pantel K, Brakenhoff RH, Brandt B. Detection, clinical relevance and Rev Mol Diagn. 2011;11(5):473-485. specific biological properties of disseminating tumour cells. Nat Rev Cancer. 103. Harouaka R, Kang Z, Zheng SY, et al. Circulating tumor cells: advances 2008;8(5):329-340. in isolation and analysis, and challenges for clinical applications. Pharmacol 70. Joosse SA, Pantel K. Biologic challenges in the detection of circulating Ther. 2014;141(2):209-221. tumor cells. Cancer Res. 2013;73(1):8-11. 104. van de Stolpe A, den Toonder JM. Circulating tumor cells: what is in it for 71. Weinberg RA. Twisted epithelial-mesenchymal transition blocks senes- the patient? A vision towards the future. Cancers (Basel). 2014;6(2):1195-1207. cence. Nat Cell Biol. 2008;10(9):1021-1023. 105. Marrinucci D, Bethel K, Kolatkar A, et al. Fluid biopsy in patients with 72. Guilford P, Hopkins J, Harraway J, et al. E-cadherin germline mutations metastatic prostate, pancreatic and breast cancers. Phys Biol. 2012;9(1):016003. in familial gastric cancer. Nature. 1998;392(6674):402-405. 106. Greene BT, Hughes AD, King MR. Circulating tumor cells: the substrate

April 2015, Vol. 22, No. 2 Cancer Control 175 of personalized medicine? Front Oncol. 2012;2:69. J Clin Pathol. 2008;61(3):307-310. 107. Miyamoto DT, Sequist LV, Lee RJ. Circulating tumour cells-monitoring 137. Heinzelmann-Schwarz VA, Gardiner-Garden M, Henshall SM, et al. treatment response in prostate cancer. Nat Rev Clin Oncol. 2014;11(7):401-412. Overexpression of the cell adhesion molecules DDR1, Claudin 3, and Ep- 108. Magbanua MJ, Park JW. Advances in genomic characterization of CAM in metaplastic ovarian epithelium and ovarian cancer. Clin Cancer Res. circulating tumor cells. Cancer Metastasis Rev. 2014;33(2-3):757-769. 2004;10(13):4427-4436. 109. Hyun KA, Jung HI. Advances and critical concerns with the microfluidic 138. Lowes LE, Allan AL. Recent advances in the molecular characterization enrichments of circulating tumor cells. Lab Chip. 2014;14(1):45-56. of circulating tumor cells. Cancers (Basel). 2014;6(1):595-624. 110. Riahi R, Gogoi P, Sepehri S, et al. A novel microchannel-based device 139. Roychowdhury S, Iyer MK, Robinson DR, et al. Personalized oncology to capture and analyze circulating tumor cells (CTCs) of breast cancer. Int J through integrative high-throughput sequencing: a pilot study. Sci Transl Med. Oncol. 2014;44(6):1870-1878. 2011;3(111):111ra121. 111. Yusa A, Toneri M, Masuda T, et al. Development of a new rapid isola- 140. Liang WS, Craig DW, Carpten J, et al. Genome-wide characterization tion device for circulating tumor cells (CTCs) using 3D palladium filter and its of pancreatic adenocarcinoma patients using next generation sequencing. application for genetic analysis. PLoS One. 2014;9(2):e88821. PLoS One. 2012;7(10):e43192. 112. Galletti G, Sung MS, Vahdat LT, et al. Isolation of breast cancer and 141. Craig DW, O’Shaughnessy JA, Kiefer JA, et al. Genome and tran- gastric cancer circulating tumor cells by use of an anti HER2-based microfluidic scriptome sequencing in prospective metastatic triple-negative breast cancer device. Lab Chip. 2014;14(1):147-156. uncovers therapeutic vulnerabilities. Mol Cancer Ther. 2013;12(1):104-116. 113. Ai M, Liang K, Lu Y, et al. Brk/PTK6 cooperates with HER2 and Src in 142. Landau DA, Carter SL, Getz G, et al. Clonal evolution in hematological regulating breast cancer cell survival and epithelial-to-mesenchymal transition. malignancies and therapeutic implications. Leukemia. 2014;28(1):34-43. Cancer Biol Ther. 2013;14(3):237-245. 143. Heitzer E, Auer M, Gasch C, et al. Complex tumor genomes inferred 114. Kirby BJ, Jodari M, Loftus MS, et al. Functional characterization of from single circulating tumor cells by array-CGH and next-generation sequenc- circulating tumor cells with a prostate-cancer-specific microfluidic device. PLoS ing. Cancer Res. 2013;73(10):2965-2975. One. 2012;7(4):e35976. 144. Lohr JG, Adalsteinsson VA, Cibulskis K, et al. Whole-exome sequencing 115. Warkiani ME, Guan G, Luan KB, et al. Slanted spiral microfluidics of circulating tumor cells provides a window into metastatic prostate cancer. for the ultra-fast, label-free isolation of circulating tumor cells. Lab Chip. Nat Biotechnol. 2014;32(5):479-484. 2014;14(1):128-137. 145. Magbanua MJ, Sosa EV, Scott JH, et al. Isolation and genomic analysis 116. Sheng W, Ogunwobi OO, Chen T, et al. Capture, release and culture of circulating tumor cells from castration resistant metastatic prostate cancer. of circulating tumor cells from pancreatic cancer patients using an enhanced BMC Cancer. 2012;12:78. mixing chip. Lab Chip. 2014;14(1):89-98. 146. Sun J, Liu W, Adams TS, et al. DNA copy number alterations in 117. Sollier E, Go DE, Che J, et al. Size-selective collection of circulating prostate cancers: a combined analysis of published CGH studies. Prostate. tumor cells using Vortex technology. Lab Chip. 2014;14(1):63-77. 2007;67(7):692-700. 118. Sieuwerts AM, Kraan J, Bolt J, et al. Anti-epithelial cell adhesion mole- 147. Bubendorf L, Kononen J, Koivisto P, et al. Survey of gene amplifications cule antibodies and the detection of circulating normal-like breast tumor cells. during prostate cancer progression by high-throughout fluorescence in situ J Natl Cancer Inst. 2009;101(1):61-66. hybridization on tissue microarrays. Cancer Res. 1999;59(4):803-806. 119. Grover PK, Cummins AG, Price TJ, et al. Circulating tumour cells: 148. Magbanua MJ, Sosa EV, Roy R, et al. Genomic profiling of isolated the evolving concept and the inadequacy of their enrichment by Ep- circulating tumor cells from metastatic breast cancer patients. Cancer Res. CAM-based methodology for basic and clinical cancer research. Ann Oncol. 2013;73(1):30-40. 2014;25(8):1506-1516. 149. Steinert G, Schölch S, Niemietz T, et al. Immune escape and surviv- 120. Went PT, Lugli A, Meier S, et al. Frequent EpCam protein expression al mechanisms in circulating tumor cells of colorectal cancer. Cancer Res. in human carcinomas. Hum Pathol. 2004;35(1):122-128. 2014;74(6):1694-1704. 121. Eichelberg C, Chun FK, Bedke J, et al. Epithelial cell adhesion molecule 150. Ni X, Zhuo M, Su Z, et al. Reproducible copy number variation patterns is an independent prognostic marker in clear cell renal carcinoma. Int J Cancer. among single circulating tumor cells of lung cancer patients. Proc Natl Acad 2013;132(12):2948-2955. Sci U S A. 2013;110(52):21083-21088. 122. Went P, Dirnhofer S, Salvisberg T, et al. Expression of epithelial cell 151. Strauss WM, Carter C, Klem E, et al. Molecular platforms for mutation adhesion molecule (EpCam) in renal epithelial tumors. Am J Surg Pathol. analysis from whole blood derived clinical samples by NextGen sequencing. 2005;29(1):83-88. Poster presented at: Proceedings of the 105th Annual Meeting of the Ameri- 123. Ensinger C, Kremser R, Prommegger R, et al. EpCAM overexpression can Association for Cancer Research; April 5–9 2014; San Diego, CA. https:// in thyroid carcinomas: a histopathological study of 121 cases. J Immunother. static.squarespace.com/static/526eafafe4b0c46729f78d96/t/5348165ce4b- 2006;29(5):569-573. 02c0c8e9d5ce3/1397233244253/2014_AACR_CYN_Final.pdf. Accessed 124. Zhang L, Ridgway LD, Wetzel MD, et al. The identification and char- April 7, 2015. acterization of breast cancer CTCs competent for brain metastasis. Sci Transl Med. 2013;5(180):180ra148. 125. Mego M, De Giorgi U, Dawood S, et al. Characterization of metastatic breast cancer patients with nondetectable circulating tumor cells. Int J Cancer. 2011;129(2):417-423. 126. Rao CG, Chianese D, Doyle GV, et al. Expression of epithelial cell adhesion molecule in carcinoma cells present in blood and primary and metastatic tumors. Int J Oncol. 2005;27(1):49-57. 127. Kimura H, Kato H, Faried A, et al. Prognostic significance of EpCAM expression in human esophageal cancer. Int J Oncol. 2007;30(1):171-179. 128. Seligson DB, Pantuck AJ, Liu X, et al. Epithelial cell adhesion molecule (KSA) expression: pathobiology and its role as an independent predictor of survival in renal cell carcinoma. Clin Cancer Res. 2004;10(8):2659-2669. 129. Went P, Vasei M, Bubendorf L, et al. Frequent high-level expression of the immunotherapeutic target Ep-CAM in colon, stomach, prostate and lung cancers. Br J Cancer. 2006;94(1):128-135. 130. Gastl G, Spizzo G, Obrist P, et al. Ep-CAM overexpression in breast cancer as a predictor of survival. Lancet. 2000;356(9246):1981-1982. 131. Spizzo G, Went P, Dirnhofer S, et al. High Ep-CAM expression is associated with poor prognosis in node-positive breast cancer. Breast Cancer Res Treat. 2004;86(3):207-213. 132. Schmidt M, Hasenclever D, Schaeffer M, et al. Prognostic effect of epithelial cell adhesion molecule overexpression in untreated node-negative breast cancer. Clin Cancer Res. 2008;14(18):5849-5855. 133. Fong D, Steurer M, Obrist P, et al. Ep-CAM expression in pancreatic and ampullary carcinomas: frequency and prognostic relevance. J Clin Pathol. 2008;61(1):31-35. 134. Varga M, Obrist P, Schneeberger S, et al. Overexpression of epithelial cell adhesion molecule antigen in gallbladder carcinoma is an independent marker for poor survival. Clin Cancer Res. 2004;10(9):3131-3136. 135. Stoecklein NH, Siegmund A, Scheunemann P, et al. Ep-CAM expression in squamous cell carcinoma of the esophagus: a potential therapeutic target and prognostic marker. BMC Cancer. 2006;6:165. 136. Brunner A, Prelog M, Verdorfer I, et al. EpCAM is predominantly expressed in high grade and advanced stage urothelial carcinoma of the bladder.

176 Cancer Control April 2015, Vol. 22, No. 2 Grading is essential for treatment plans,

follow-up visits, and therapeutic

interventions.

Three Layers of Paint. Photograph courtesy of Craig Damlo. www.soapboxrocket.com.

Surgical and Molecular Pathology of Barrett Esophagus Sherma Zibadi, MD, PhD, and Domenico Coppola, MD

Background: Patients with Barrett esophagus (BE) are predisposed to developing dysplasia and cancer. Adenocarcinoma, which is associated with BE, is the most common type of esophageal tumor and, typically, it has an aggressive clinical course and a high rate of mortality. Methods: The English-language literature relating to tumor epidemiology, etiology, and the pathogenesis of BE was reviewed and summarized. Results: The role of pathologists in the diagnosis and pitfalls associated with grading Barrett dysplasia is addressed. Current molecular testing for Barrett neoplasia, as well as testing methods currently in development, is discussed, focusing on relevant tests for diagnosing tumor types, determining prognosis, and assessing therapeutic response. Conclusions: Grading is essential for developing appropriate treatment plans, follow-up visits, and therapeutic interventions for each patient. Familiarity with current molecular testing methods will help physicians correctly diagnose the disease and select the most appropriate therapy for each of their patients.

Introduction tinal metaplasia are also defined as Barrett mucosa.1 Barrett mucosa refers to a metaplastic process in- Barrett esophagus (BE) is more common in men duced by the acid-peptic content of the stomach and has a male:female ratio of 3:1 to 7:1.2 Although that then erodes the esophageal squamous mucosa. BE can present at any age, its peak incidence oc- During endoscopy, changes of any length recognized curs in the sixth decade of life and it occurs more as columnar-type mucosa in the distal esophageal ep- frequently in whites (80% of patients) than patients ithelium that, on biopsy, are confirmed to have intes- of other races.3 Patients with BE are predisposed to esophageal adenocarcinoma (EAC; 5-year survival rate of 14%–22%), which has increased in incidence From the Department of Pathology and Oncological Sciences, Uni- during the last 30 years.4,5 versity of South Florida Morsani College of Medicine (SZ), and the Guidelines for diagnosis and surveillance inter- Departments of Anatomic Pathology (DC), Chemical Biology and Molecular Medicine, and Tumor Biology, H. Lee Moffitt Cancer vals for patients diagnosed with BE have been es- Center & Research Institute, Tampa, Florida. tablished and developed by the American College of Submitted July 8, 2014; accepted September 11, 2014. Gastroenterology (Table 1).1 Address correspondence to Domenico Coppola, MD, Department of Anatomic Pathology, Moffitt Cancer Center, 12902 Magnolia Etiology and Pathogenesis Drive, Tampa, FL 33612. E-mail: [email protected] Typically, patients acquire BE as a result of gastro- No significant relationships exist between the authors and the 6,7 companies/organizations whose products or services may be esophageal reflux. The factors putting patients at referenced in this article. risk for BE are several and include duodenogastric

April 2015, Vol. 22, No. 2 Cancer Control 177 Table 1. — ACG Guidelines for BE of BE varies between 14% and 40%,26 and 7% to Dysplasia Diagnosis Follow-Up 15% of patients will have EAC at the time of initial 27 None present 2 EGDs with biopsy in 1 y EGD every 3 y diagnosis of BE. Low grade Highest grade on repeat EGD 1-y interval with biopsies within 6 mo until no Pathology Expert pathology dysplasia (× 2) Grossly, the esophagogastric junction in BE is dis- confirmation placed by areas of velvety, gastric-type mucosa that 11 High grade Mucosal irregularity Continued 3-mo macroscopically appear salmon in color. These ar- Repeat EGD with biopsies surveillance or eas of gastric-type mucosa proximally extend from to rule out esophageal intervention their typical location. The area of BE may be ulcerat- carcinoma within 3 mo based on results and outcomes ed or contain islands of residual, whitish squamous Expert pathology confirmation mucosa surrounded by pink-colored, gastric-type Invasive Expert pathology confirmation Esophagectomy mucosa (Fig 2). It is possible that the health care adenocarcinoma professional may be unable to distinguish the ap- ACG = American College of Gastroenterology, BE = Barrett esophagus, EGD = esophagogastroduodenoscopy. pearance of simple BE — grossly and endoscopi- 28 Adapted from Macmillan Publishers Ltd: Wang KK, Sampliner RE; Practice Param- cally — from BE with early cancer and dysplasia. eters Committee of the American College of Gastroenterology. Updated guidelines With regard to its location, BE characteristically in- 2008 for diagnosis, surveillance and therapy of Barrett’s esophagus, Am J Gastro- enterol, 103(3):788-797, copyright 2008. volves the lower one-third of the esophagus, but it may involve the middle and upper esophagus as it reflux, which is a delay in the clearance of acid juice progresses.11 from the esophagus, a resting pressure in the sphinc- Previously, BE was histologically indicated by the ter of the lower esophagus less than the optimal rate, presence of intestinal metaplasia or the gastric, cardia, and the presence of hiatal hernia.8-11 or oxyntic type of columnar mucosa in the esopha- Other etiological factors include postgastrectomy gus.29,30 However, in the modern era of endoscopy, bile reflux12 and esophageal injury (eg, lye ingestion, the American Gastroenterological Association requires tobacco use).13,14 Alcohol intake does not promote the presence of unequivocal intestinal metaplasia for BE15 and Helicobacter pylori infection has a protec- there to be a diagnosis of BE.31,32 tive effect, particularly if Helicobacter possesses the The nonintestinalized, gastric-type mucosa was CagA gene.16 In infants, BE may arise from congen- previously thought to have no potential for malig- ital rests of gastric columnar epithelium reacting to nant transformation; however, in recent years, small gastroesophageal reflux.5,17 (presumably early) EACs arising in the absence of The way in which normal squamous epi- intestinal metaplasia have been reported.33-35 Thus, thelium transitions into specialized columnar these findings suggest that the identification of gob- epithelium is not yet clear.18 However, based let cells as a requirement for diagnosing BE should on the theory of restitution and replication probe challenged.26-38 For example, the requirement no posed by Jankowski et al,19 after the acid-pep- longer exists in Japan, which was the first country tic content of the stomach erodes the parabasal to propose the existence of malignant potential for and superficial layers of the squamous mucosa nonintestinalized BE.35,38 In recent years, this view (Fig 1A), the basal progenitor cells move into de- has been supported by study findings indicating that nuded areas.5 In the setting of persistent gas- goblet cell–containing epithelium has similar DNA troesophageal reflux, these multipotential stem abnormalities as metaplastic esophageal columnar cells become resistant to acid and bile, selective- epithelium without goblet cells.39,40 ly differentiating into columnar epithelium that Although goblet cell metaplasia is not always an secretes mucin (Fig 1B).5,19 By contrast, the up- obligate precursor of malignancy, it is important to ward movement of the columnar epithelium from point out that the recognition of goblet cell metapla- the stomach could cause metaplasia due to wound sia remains the most important criterion in clinical healing of the ulcerated mucosa.5 practice for the routine histological recognition and diagnosis of BE. Cancer Risk In BE, the intestinalized mucosa rarely has a vil- BE is a premalignant condition associated with EAC liform appearance and Paneth cells, absorptive cells, and dysplasia.20,21 During the last several decades, and intestinal-type endocrine cells are typically absent the incidence of EAC has risen.21-23 Approximately because of the incomplete form of metaplasia pres- 10% to 20% of patients with symptomatic gastro- ent. Uncommonly, the complete form of metaplasia esophageal reflux who undergo endoscopy have presents in conjunction with the usual incomplete BE.24 However, many patients with BE are asymp- form of metaplasia. tomatic.25 The prevalence of dysplasia in the setting Shaped like barrels, goblet cells have a distended

178 Cancer Control April 2015, Vol. 22, No. 2 Bile A Acid

Superficial compartment Desquamating cells

Parabasal compartment

Capillaries

Basal compartment Positively labeled nuclei/cells by autoradiography Submucosa Papillae region Basal region Smooth muscle

B Bile Acid

Superficial compartment Desquamating cells

Parabasal compartment

Capillaries

Positively labeled nuclei/cells Basal compartment by autoradiography

Submucosa Papillae region Basal region Smooth muscle

C Bile Acid Rapidly dividing mucin-secreting cells

Superficial compartment Desquamating cells

Parabasal compartment

Capillaries

Positively labeled nuclei/cells Basal compartment by autoradiography

Submucosa Papillae region Basal region Smooth muscle

Fig 1A–C. — (A) Basal progenitor cells move into denuded areas after the gastric acid-peptic content from the stomach has eroded the squamous mucosa. (B, C) Persistent gastroesophageal reflux causes multipotential stem cells to differentiate into columnar, mucin-secreting epithelium resistant to acid and bile. Reprinted from Am J Pathol, 154(4), Jankowski JA, Wright NA, Meltzer SJ, et al, Molecular evolution of the metaplasia-dysplasia-adenocarcinoma sequence in the esophagus, 965-973, Copyright 1999, with permission from Elsevier.

April 2015, Vol. 22, No. 2 Cancer Control 179 which it arises and within the superficial layer of epithelium is the defining characteristic of dysplasia.41-43 Criteria have been established for grading dysplasia in BE44 that consider a combination of features, such as gland architecture, the presence of inflammation, ulceration, or erosion, cytological features, and the level of maturation of the surface epithelium.28

Negative When cases of BE are negative for dysplasia, the underlying glands will not be as mature as the surface epithelium.30 The nuclear:cytoplasmic ratio is lower than that of the deeper glands.30 Lamina propria is abundantly present between the glands.30 Cytologically, nuclear polarity is retained, the nuclear membrane is smooth, and the nucleoli are inconspicuous. In the event of prominent inflammation, reactive changes will be present but surface maturation is maintained.44,45

Indefinite A diagnosis of indefinite for dysplasia is used for cases with worrisome changes but not diagnostic for dysplasia, typically because of the presence of marked inflammation. Normal glandular architecture may be observed but crowding will be present, and, cytologically, nuclear hyperchromasia, nuclear membrane irregularities, and increased mitoses at Fig 2. — Gross appearance of BE as a velvety, salmon-colored mucosa the base of the glands (but with surface maturation) extending above the gastroesophageal junction. BE = Barrett esophagus. may be present. Nuclear polarity is not lost in cases cytoplasm filled with acidic mucin and can be identi- of indefinite for dysplasia. Such a diagnosis may also fied using hematoxylin and eosin stain, and, if neces- be used in the presence of tangential embedding that sary, also highlighted with either Periodic acid–Schiff does not allow for the assessment of the superficial or Alcian blue (pH 2.5) stain.11 portion of the glands.44,46

Role of Pathologists Low Grade The role of pathologists when dealing with esoph- In cases of BE with low-grade dysplasia (LGD), the agogastric biopsy or resection is twofold: mucosal surface is similar to the underlying glands, 1. Identification of BE: The pathologist must which reveal nuclear stratification and nuclear polarity differentiate between nonmetaplastic (Fig 3).30 Glandular crowding can be seen, but the columnar epithelium (eg, hiatal hernia, lamina propria is still present between the glands. The ectopic gastric mucosa), intestinal nuclei are elongated, hyperchromatic, and have irregu- metaplasia of the cardia, and true BE. lar contours and sustain a moderate increase in mitotic 2. Identification and grading of dysplasia: activity. The nuclear:cytoplasmic ratio will be mildly This task will affect patient treatment. increased. In the setting of LGD, minimal inflammation When dealing with endoscopic mucosal resection may be present and nucleoli are not prominent.45-47 specimens, pathologists are responsible for assessing the resection margins. If invasive cancer is present, High Grade then pathologists are also required to report the dis- In the setting of BE with high-grade dysplasia (HGD), tance of the invasive carcinoma from the deep mar- surface maturation is lacking; glandular distortion gin, the presence of angiolymphatic invasion, and, if and glandular crowding are both present with a min- possible, the level of invasion within the muscularis imal amount of lamina propria between the glands.47 mucosa, submucosa, or both. The glands may become dilated with luminal necrot- ic debris. The nuclei display marked enlargement, Dysplasia anisocytosis, pleomorphism, hyperchromasia, mem- The presence of the neoplastic epithelium confined brane irregularity, and prominent nucleoli. The mitotic within the basement membrane of the gland from figures are numerous and the nuclear:cytoplasmic

180 Cancer Control April 2015, Vol. 22, No. 2 observing these findings.48,49

Clinical Significance One study found that HGD was a marker of metachronous or synchronous adenocarcinoma in approximately 40% of cases.50 In another study, Weston et al51 found that HGD progressed to adenocarcinoma or multifocal HGD in 53% of cases.

Pitfalls Associated With Grading The grading of dysplasia in BE is subjective due to sampling errors as well as interobserver and intraobserver variations. In addition, the biological behavior of the lesion and its degree of dysplasia may not correlate.34,52 However, the task of grading dysplasia in BE is important and can sometimes have crucial therapeutic implications (see Table 1).1 For this reason, biopsy findings in cases of BE, particularly in the setting of HGD and carcinoma, should be reviewed by a second pathologist with expertise in gastrointestinal pathology.53 Morphological and molecular data suggest that dysplasia in BE begins at the base of the crypts.54 Crypt dysplasia is thought to progress into the remaining portion of the crypt and surface epithelium.55 Histologically, crypt dysplasia may exhibit the same Fig 3. — BE with low-grade dysplasia. The glands exhibit mild architectural cytological changes characteristic of LGD or, rarely, distortion and are lined by elongated, stratified, and hyperchromatic nuclei have high-grade cytological features with very large with retained polarity (hematoxylin and eosin, × 200). BE = Barrett esophagus. nuclei, a large nuclear:cytoplasmic ratio, irregular nuclear membranes, nuclear polarity loss, and eosinophilic cytoplasm. One study had good interobserver agreement when diagnosing crypt dysplasia.55

Intramucosal Carcinoma In intramucosal adenocarcinoma, the neoplastic cells are limited to the mucosa (Fig 5); however, these cells will make their way to the basement membrane and

Fig 4. — BE with high-grade dysplasia. The epithelium is composed of cells containing enlarged, polygonal, and hyperchromatic nuclei that exhibit prominent nucleoli. Polarity is lost (hematoxylin and eosin, × 200). BE = Barrett esophagus. ratio will be large. Nuclear polarity is lost (Fig 4). Inflammation will be minimal. In the setting of HGD, the presences of superficial ulceration, cytologically high-grade nuclei, solid nests of dysplastic cells with multiple secondary lumina, and/or dilated dysplastic tubules containing granular eosinophilic and nuclear Fig 5. — BE with intramucosal carcinoma showing single cells and debris may indicate an unsampled invasive carcino- a syncytial arrangement of the cells infiltrating the lamina propria. ma. Additional biopsies should be performed when BE = Barrett esophagus.

April 2015, Vol. 22, No. 2 Cancer Control 181 into the muscularis mucosae or the lamina propria, example, following mucosal injury, the mucosa rapidly but they do not go beyond this point.5 Histologically, heals by restitution. During this phase, metaplastic an effacement of the architecture of the lamina propria clones with specialized intestinal metaplasia propa- can be seen,28 as well as back-to-back microglands, gate in the presence of persistent gastroesophageal re- small clusters, or infiltrating single cells.56 A syncy- flux, followed by the loss of cell-cycle checkpoints. In tial growth pattern may also be observed. Typically, addition, because of increasing proliferation, genomic desmoplasia is absent or not completely developed, instability is thought to be associated with slow clonal perhaps accounting for the poor intraobserver repro- expansion.19,63,64 The inhibition of apoptosis occurs ducibility in the diagnosis of this entity.45,46 late and only in cases of HGD. Altering cell adhesion can precede invasive cancer,65 and the accumulation of Invasive Adenocarcinoma genetic errors that follows may be associated with the Invasive adenocarcinoma has a variable gross ap- generation of multiple, transformed cell clones, thus pearance and size (≤ 10 cm) and typically involves expanding the number of altered cells with metastatic the distal one-third of the esophagus. Tumor cells or angiogenic potential.19 infiltrate beyond the muscularis mucosa in the setting of invasive adenocarcinoma and oftentimes evoke a Aberrant p53 desmoplastic reaction.5 The microscopic pattern is Several studies have observed alterations in p53 mostly of the intestinal form, however diffuse, and during the progression of BE.66-68 Typically, aberrant signet ring cell or mixed types can be observed. The p53 is accumulated in the nucleus of the neoplas- tumor cells may produce mucin, and the tumor itself tic cells, particularly in cases of HGD (Fig 6). When will generally be well to moderately differentiated. they were detected by polymerase chain reaction single-strand conformation polymorphism, the p53 Intestinal Metaplasia of the Cardia molecular alterations coincided with the level of p53 Intestinal metaplasia of the cardia has a prevalence protein overexpression detected by immunohisto- of 6% to 25%.57,58 Generally in this setting, the cardia chemistry in Barrett adenocarcinoma and BE-HGD contains areas of intestinal metaplasia with goblet but were discordant in BE-LGD.68 cells and a normal Z line. H pylori infection, gastro- Currently, the only accepted predictor for pro- esophageal reflux, and carditis can all be causes for gression is a histological diagnosis of LGD; its pre- intestinal metaplasia of the cardia. Patients with intes- dictive value is low.69,70 However, the immunohisto- tinal metaplasia of the cardia are at increased risk for chemistry status of p53 has diagnostic value and can gastric cardia cancer; however, this type of intestinal help predict neoplastic progression in patients with metaplasia may go into regression if its cause is treat- BE. Using more than 12,000 biopsy samples from ed; furthermore, research suggests that this condition 635 patients, including those with BE who developed is not typically associated with dysplasia.57,59 HGD, those with invasive adenocarcinoma, and those without neoplastic progression, Kastelein et al70 stud- Endoscopic Mucosal Resection ied the protein expression of p53 via immunohisto- Gastroesophageal endoscopic mucosal resection is chemistry. The researchers found that 49 patients (8%) a procedure to remove dysplastic or carcinomatous developed HGD or invasive adenocarcinoma during mucosa from the stomach, esophagus, or both.52 It the follow-up period.70 The results of the study in- is performed on low-risk adenocarcinomas using ei- dicated that patients with a diagnosis of LGD had a ther an injection- or cap-assisted endoscopic mucosal resection. Low-risk adenocarcinoma is defined as a stage 1A tumor no larger than 20 mm in size without vascular or lymphatic invasion and a histological grade of G1 or G2. Following endoscopic mucosal resection, a complete tumor response was reported in 99% of patients after a follow-up time of 36.7 months.60 Recurrent or metachronous lesions were present in 11% of patients and the 5-year survival rate was 98%.61,62

Molecular Alterations Jankowski et al19 highlight restitution, cell cycle and apoptosis, adhesion, DNA instability, and invasion as potential factors in the molecular pathways associated Fig 6. — p53 immunostaining of Barrett mucosa showing strong nuclear with the progression of BE to adenocarcinoma. For positivity and negativity in high- and low-grade dysplasia, respectively.

182 Cancer Control April 2015, Vol. 22, No. 2 positive predictive value for neoplastic progression of Table 2. — FISH Probe Sets 70 15%. Conversely, patients with LGD and concurrent Probe Fluorophore altered p53 expression had a positive predictive value Set 70 Spectrum Spectrum Spectrum Spectrum of 33%. Thus, these data show that p53 alterations Red Green Aqua Gold are a better predictor of tumor progression than BE I LSI 9p21 LSI 5p15 CEP 9 LSI 5q21–22 70 with LGD. (P16) (APC) II CEP Y LSI 17q11.2–12 CEP 17 LSI 17p13.1 ERBB2 Amplification and Overexpression (ERBB2a) (P53) 71 Using a tissue microarray, Hu et al looked at II CEP Y LSI 17q11.2–12 CEP 17 LSI 17p13.1 34 cases of BE, 18 cases of LGD, 15 cases of HGD, and (ERBB2a) (P53) 116 cases of EAC to study ERBB2 (formerly known as aFormerly known as HER2/neu. HER2 or HER2/neu) amplification using chromogenic CEP = centromere enumeration, FISH = fluorescence in situ hybridization, LSI = locus-specific indicator. in situ hybridization. They found ERBB2 amplification Reprinted from J Mol Diagn, 8(2), Brankley SM, Wang KK, Harwood AR, et al, in 21 of the cases of EAC (18.1%), ERBB2 overexpres- The development of a fluorescence in situ hybridization assay for the detection of dysplasia and adenocarcinoma in Barrett’s esophagus, 260-267, Copyright 2006, sion in another 14 cases of EAC (12.1%), and ERBB2 with permission from Elsevier. protein overexpression in 1 case of HGD (6.7%).71 All of the cases of LGD and BE had negative results for were similar in all cases of BE, which clustered to- fluorescence in situ hybridization (FISH) and immu- gether and were different from the EAC cases (also nohistochemistry.71 clustered together).75 The researchers also differenti- A recent study looked at the immunohistochemi- ated adenocarcinoma from squamous cell carcinoma cal expression of ERBB2 in BE, BE with dysplasia, and and signet ring cell carcinoma. adenocarcinoma.72 The overexpression of ERBB2 in a In another study, Helm et al76 analyzed microar- significant number of HGD cases was an unexpected rays so that they could obtain expression profiles finding that may have clinical implications. The low from individual genes based on biopsy samples expression level of ERBB2 in LGD was a novel find- (9 cases of EAC/BE from which 3 cancers had arisen). ing and may indicate the role of ERBB2 in the early To use as a reference point, the researchers pooled stages of Barrett carcinogenesis.72 samples of BE from 6 patients with BE but without cancer or dysplasia. Their findings showed that in- Molecular Testing creasing changes in gene expression were related Molecular testing has been proposed to accurately to the progression from BE to EAC. The early loss detect and grade dysplasia and adenocarcinoma in of gene function that governs differentiation began patients with BE.73 Among these tests, FISH can be prior to such histological change, whereas a gain in performed on endoscopic esophageal brushings, gene function was associated with the progression of which can be used to sample a large diseased area invasiveness and remodeling.76 of the esophagus. In a study by Brankley et al,74 Using DNA microarrays, Hao et al77 focused various probe combinations were used to select a on the contributions of the stroma to the neo- combination of FISH probes specific for identify- plastic process, studying the gene expression ing dysplasia and carcinoma in the setting of BE profiles of EAC or BE specimens obtained via bi- (Table 2). A total of 170 brushing specimens from opsy and the associated normal duodenum and 138 patients with BE were examined. Biopsy results esophagus among 17 volunteers. The research- were used as the gold standard. The authors found ers found a unique expression profile for each that probes to 8q24, 9p21, 17q11.2, and 20q13.2 tissue. Both EAC and BE had a unique set of detected LGD with sensitivity and specificity rates 37 stromal genes distinct from those found in normal of 70% and 89%, respectively.74 The same probes tissue.77 Collagen 5A2 and periostin were validated were also able to identify HGD with sensitivity and by in situ hybridization. No differences in gene specificity rates of 84% and 93%, respectively, and expression were seen between long- and identify adenocarcinoma with sensitivity and spec- short-segment-type BE.77 The authors concluded ificity rates of 94% and 93%, respectively.74 Thus, that stromal gene changes precede transformation. these probes have been proposed for testing in prospective clinical trials. Ultrasensitive Complementary DNA Selaru et al75 used global gene expression pro- Microarray Chip filing to study BE and esophageal cancer using DNA In 2007, Ito et al78 introduced an ultrasensitive cDNA microarrays. They examined 13 esophageal resections microarray chip for gene-expression profiling to be or biopsies, including 7 cases of BE and 6 cases of used in preoperative esophageal cancer biopsies EAC, using a microarray platform containing 8,000 that requires 0.1 to 0.01 µg of total ribonucleic acid complementary DNA (cDNA) clones. Gene profiles (RNA). However, no RNA amplification is needed.

April 2015, Vol. 22, No. 2 Cancer Control 183 Nucleus Cytoplasm the personalized management of Barrett esophagus. RNase III Dicer Pathologists are essential for planning the most ap- Pre-miRNA TRBP propriate follow-up period needed for each patient. RAN- The advent of molecular testing for Barrett esopha- GTP gus also provides new tools for selecting the most Exportin 5 Mature-miRNA appropriate therapy for each patient. RISC DGCR8 References Pri-miRNA Drosha Imperfect 1. Wang KK, Sampliner RE; Practice Parameters Committee of the Perfect complementarity RNase III complementarity American College of Gastroenterology. Updated guidelines 2008 for diag- Polymerase II, III nosis, surveillance and therapy of Barrett’s esophagus. Am J Gastroenterol. 2008;103(3):788-797. Cap AAAAAA 5’ 3’ Cap ORF AAAA 2. Rashid F, Khan RN, Iftikhar SY. Probing the link between oestrogen receptors and oesophageal cancer. World J Surg Oncol. 2010;8:9. 5’ 3’ Translational 3. Pohl H, Welch HG. The role of overdiagnosis and reclassification in miRNA gene Cap ORF AAAA repression the marked increase of esophageal adenocarcinoma incidence. J Natl Cancer mRNA cleavage Inst. 2005;97(2):142-146. 4. Gamliel Z, Krasna MJ. Multimodality treatment of esophageal cancer. Fig 7. — The expression of short, noncoding RNA influences genes by Surg Clin North Am. 2005;85(3):621-630. blocking the translation of target messenger RNAs or by quickening their 5. Coppola D, Nasir NA, Turner L. Genesis of Barrett’s neoplasia: current degradation. RNA = ribonucleic acid. Reprinted from Curr Opin Pharmacol, concepts. In: Coppola D, ed. Mechanisms of Oncogenesis. The Netherlands: Springer; 2010:213-224. 9(6), Kan T, Meltzer SJ, MicroRNAs in Barrett’s esophagus and esophageal 6. Bremner CG, Lynch VP, Ellis FH Jr. Barrett’s esophagus: congenital adenocarcinoma, 727-732, Copyright 2009, with permission from Elsevier. or acquired? An experimental study of esophageal mucosal regeneration in the dog. Surgery. 1970;68(1):209-216. 7. Hamilton SR, Yardley JH. Regenerative of cardiac type mucosa and This method is highly sensitive. It might be suit- acquisition of Barrett mucosa after esophagogastrostomy. Gastroenterology. able for clinical diagnosis because it discriminates 1977;72(4 pt 1):669-675. 8. Dent J, Bremner CG, Collen MJ, et al. Barrett’s oesophagus. J Gas- noncancerous tissue from cancerous tissue and has troenterol Hepatol. 1991;6(1):1-22. accuracy, sensitivity, and specificity rates of 95.2%, 9. Collen MJ, Lewis JH, Benjamin SB. Gastric acid hypersecretion in refractory gastroesophageal reflux disease. Gastroenterology. 1990;98(3):654-661. 95.7%, and 94.7%, respectively. 10. Mulholland MW, Reid BJ, Levine DS, et al. Elevated gastric acid secretion in patients with Barrett’s metaplastic epithelium. Dig Dis Sci. 1989;34(9):1329-1334. MicroRNA 11. Coppola D, Karl RC. Barrett’s esophagus and Barrett’s-associated The expression of short, noncoding RNA influenc- neoplasia: etiology and pathologic features. Cancer Control. 1999;6(1):21-27. 12. Meyer W, Vollmar F, Bär W. Barrett-esophagus following total gastrec- es genes by blocking the translation of target mes- tomy. A contribution to its pathogenesis. Endoscopy. 1979;11(2):121-126. senger RNAs or by quickening their degradation 13. Fisher DR, Preston DF, Robinson RG, et al. Barrett’s esophagus com- 79,80 81 plicating lye ingestion. Demonstration by pertechnetate scintigraphy. Clin Nucl (Fig 7). Yang et al examined microRNA expression Med. 1983;8(11):550-552. in 5 cases of BE with LGD, 5 cases of BE with HGD, and 14. Kubo A, Levin TR, Block G, et al. Cigarette smoking and the risk of Barrett’s esophagus. Cancer Causes Control. 2009;20(3):303-311. 6 cases of EAC using a human microRNA microar- 15. El-Serag HB, Lagergren J. Alcohol drinking and the risk of Bar- ray kit and then validated by real-time reverse tran- rett’s esophagus and esophageal adenocarcinoma. Gastroenterology. 2009;136(4):1155-1157. scription–polymerase chain reaction. Unsupervised 16. Rubenstein JH, Inadomi JM, Scheiman J, et al. Association between hierarchical clustering and class comparison analy- Helicobacter pylori and Barrett’s esophagus, erosive esophagitis, and gastroesophageal reflux symptoms. Clin Gastroenterol Hepatol. 2014;12(2):239-245. ses showed different microRNA expression profiles 17. Othersen HB Jr, Ocampo RJ, Parker EF, et al. Barrett’s esophagus in between either EAC or BE with HGD and their cor- children. Diagnosis and management. Ann Surg. 1993;217(6):676-681. 81 18. Grotenhuis BA, van Lanschot JB, Dinjens WN, et al. The pathogenesis responding normal tissues. No differences in mi- of Barrett’s metaplasia and the progression to esophageal adenocarcinoma. croRNA expression profiles were observed between In: Schneider PM, ed. Adenocarcinoma of the Esophagogastric Junction. Berlin: Springer; 2010:39-63. BE with LGD and their normal counterparts (refer 19. Jankowski JA, Wright NA, Meltzer SJ, et al. Molecular evolution of the to Table 4 in Yang et al).81 metaplasia-dysplasia-adenocarcinoma sequence in the esophagus. Am J Pathol. 1999;154(4):965-973. In addition, microRNA signatures specific to 20. Haggitt RC, Tryzelaar J, Ellis FH, et al. Adenocarcinoma compli- each stage of Barrett carcinogenesis have been cating columnar epithelium-lined (Barrett’s) esophagus. Am J Clin Pathol. 1978:70(1):1-5. identified; each correlated with their target gene 21. Goldblum JR. Barrett’s esophagus and Barrett’s-related dysplasia. Mod expression.69 Pathol. 2003;16(4):316-324. 22. Haggitt RC. Adenocarcinoma in Barrett’s esophagus: a new epidemic? Hum Pathol. 1992;23(5):475-476. Conclusions 23. Blot WJ. Esophageal cancer trends and risk factors. Semin Oncol. 1994:21(4):403-410. We presented a review of the clinical and pathologi- 24. Winters C Jr, Spurling TJ, Chobanian SJ, et al. Barrett’s esophagus. A cal characteristics of Barrett esophagus, underscoring prevalent, occult complication of gastroesophageal reflux disease. Gastroen- terology. 1987;92(1):118-124. the role of pathologists in recognizing and grading 25. Spechler SJ, Souza RF. Barrett’s esophagus. N Engl J Med. this disease. Potential pitfalls during diagnosis, de- 2014;371(9):836-845. 26. DeMeeter TR, Attwood SE, Smyrk TC, et al. Surgical therapy in Barrett’s tection, and grading of Barrett esophagus, including esophagus. Ann Surg. 1990;212(4):528-542. associated dysplasia, were also topics addressed. Fi- 27. Palley SL, Sampliner RE, Garewal HS. Management of high-grade dysplasia in Barrett’s esophagus. J Clin Gastroenterol.1989;11(4):369-372. nally, we discussed select molecular techniques that 28. Montgomery E, Canto MI. Management of high-grade dysplasia in patients may soon be introduced in the clinical setting for with Barrett’s esophagus. Clin Gastroenterol Hepatol. 2006;4(12):1434-1439.

184 Cancer Control April 2015, Vol. 22, No. 2 29. Woolf GM, Riddell RH, Irvine EJ, et al. A study to examine agreement 1980;302(15):844-848. between endoscopy and histology for the diagnosis of columnar-lined (Bar- 60. Kojima T, Parra-Blanco A, Takahashi H, et al. Outcome of endoscopic rett’s) esophagus. Gastrointest Endosc. 1989;35(6):541-544. mucosal resection for early gastric cancer: review of the Japanese literature. 30. Montgomery E, Kalloo AN. Endoscopic mucosal biopsy: histopatho- Gastrointest Endosc. 1998;48(5):550-550. logical interpretation. In: Yamada T, ed. Textbook of Gastroenterology. Oxford: 61. Schnell TG, Sontag SJ, Chejfec G, et al. Long-term nonsurgical man- Blackwell Publishing; 2009. agement of Barrett’s esophagus with high-grade dysplasia. Gastroenterology. 31. Brown IS, Whiteman DC, Lauwers GY. Foveolar type dysplasia in Barrett 2001;120(7):1607-1619. esophagus. Mod Pathol. 2010;23(6):834-843. 62. Ell C, May A, Pech O, et al. Curative endoscopic resection of ear- 32. Spechler SJ, Sharma P, Souza RF, et al; American Gastroenterolog- ly esophageal adenocarcinomas (Barrett’s cancer). Gastrointest Endosc. ical Association. American Gastroenterological Association medical position 2007;65(1):3-10. statement on the management of Barrett’s esophagus. Gastroenterology. 63. Huang Y, Boynton RF, Blount PL, et al. Loss of heterozygosity involves 2011;140(3):1084-1091. multiple tumour suppressor genes in human esophageal cancers. Cancer Res. 33. Cameron AJ, Souto EO, Smyrk TC. Small adenocarcinomas of the 1992;52(23):6525-6530. esophagogastric junction: association with intestinal metaplasia and dysplasia. 64. Reid BJ, Sanchez CA, Blount PL, et al. Barrett’s esophagus: cell cycle Am J Gastroenterol. 2002;97(6):1375-1380. abnormalities in advancing stages of neoplastic progression. Gastroenterology. 34. Cameron AJ, Lomboy CT, Pera M, et al. Adenocarcinomas of the esoph- 1993;105(1):119-129. agogastric junction and Barrett’s esophagus. Gastroenterology. 2009;40(1):65-74. 65. Bailey T, Biddlestone L, Shepherd N, et al. Altered cadherin/catenin 35. Takubo K, Aida J, Naomoto Y, et al. Cardiac rather than intestinal-type complexes in the Barrett’s esophagus dysplasia-adenocarcinoma sequence: background in endoscopic resection specimens of minute Barrett adenocar- correlation with disease progression and dedifferentiation. Am J Pathol. cinoma. Hum Pathol. 2009;40(1):65-74. 1998;152(1):135-144. 36. Chandrasoma P, Wickramasinghe K, Ma Y, et al. Is intestinal metaplasia 66. Ohbu M, Kobayashi N, Okayasu I. Expression of cell cycle regulatory a necessary precursor lesion for adenocarcinoma of the distal esophagus, gas- proteins in the multistep process of oesophageal carcinogenesis: stepwise troesophageal junction and gastric cardia? Dis Esophagus. 2007;20(1):36-41. overexpression of cyclin E and p53, reduction of p21(WAF1/CIP1) and dys- 37. Kelty C, Gough MD, Van Wyk Q, et al. Barrett’s oesophagus: in- regulation of cyclin D1 and p27(KIP1). Histopathology. 2001;39(6):589-596. testinal metaplasia is not essential for cancer risk. Scand J Gastoenterol. 67. Novotna K, Trkova M, Pazdro A, et al. TP53 gene mutations are rare 2007;42(11):1271-1274. in nondysplastic Barrett’s esophagus. Digest Dis Sci. 2006;51(1):110-113. 38. Barr H, Upton MP, Orlando RC, et al. . Barrett’s esophagus: histology 68. Bian YS, Osterheld MC, Bosman FT, et al. p53 gene mutation and and immunohistology. Ann N Y Acad Sci. 2011;1232:76-92. protein accumulation during neoplastic progression in Barrett’s esophagus. 39. Liu W, Hahn H, Odze RD, et al. Metaplastic esophageal columnar Mod Pathol. 2001;14(5):397-403. epithelium without goblet cells shows DNA content abnormalities similar to 69. Srivastava A, Hornick JL, Li X, et al. Extent of low-grade dysplasia is goblet cell–containing epithelium. Am J Gastroenterol. 2009;104(4):816-824. a risk factor for the development of esophageal adenocarcinoma in Barrett’s 40. Romagnoli S, Roncalli M, Graziani D, et al. Molecular alterations of esophagus. Am J Gastroenterol. 2007;102(3):483-493. Barrett’s esophagus on microdissected endoscopic biopsies. Lab Invest. 70. Kastelein F, Biermann K, Steyerberg EW, et al; ProBar-study group. Ab- 2001;81(2):241-247. errant p53 protein expression is associated with an increased risk of neoplastic 41. Riddell RH, Goldman H, Ransohoff DF, et al. Dysplasia in inflammatory progression in patients with Barrett’s oesophagus. Gut. 2013;62(12):1676-1683. bowel disease: standardized classification with provisional clinical applications. 71. Hu Y, Bandla S, Godfrey TE, et al. HER2 amplification, overexpression and Hum Pathol. 1983;14(11):931-968. score criteria in esophageal adenocarcinoma. Mod Pathol. 2011;24(7):899-907. 42. Horvath B, Singh P, Xie H, et al. Risk for esophageal in Barrett’s esoph- 72. Jiang K, Henderson-Jackson E, Almhanna K, et al. Immunohistochemi- agus patients with mucosal changes indefinite for dysplasia. J Gastroenterol cal expression of HER2-neu in gastroesophageal intestinal metaplasia, dyspla- Hepatol. 2015;30(2):262-267. sia and adenocarcinoma. Abstract presented at: American Society for Clinical 43. Goldblum JR, Liu X. Role of histology and cytology in esophageal diseases. Pathology Annual Meeting; Tampa, FL; October 8–10, 2014. In: Richter JE, Castell DO, eds. The Esophagus. Oxford: Wiley-Blackwell; 2012. 73. Tischoff I, Tannapfel A. Barrett’s esophagus: can biomarkers predict pro- 44. Montgomery E, Bronner MP, Goldblum JR, et al. Reproducibility of gression to malignancy? Expert Rev Gastroenterol Hepatol. 2008;2(5):653-663. the diagnosis of dysplasia in Barrett esophagus: a reaffirmation. Hum Pathol. 74. Brankley SM, Wang KK, Harwood AR, et al. The development of a 2001;32(4):368-378. fluorescence in situ hybridization assay for the detection of dysplasia and 45. Odze RD. Diagnosis and grading of dysplasia in Barrett’s oesophagus. adenocarcinoma in Barrett’s esophagus. J Mol Diagn. 2006;8(2):260-267. J Clin Pathol. 2006;59(10):1029-1038. 75. Selaru FM, Zou T, Xu Y, et al. Global gene expression profiling in 46. Montgomery EA. Barrett esophagus and esophageal neoplasms. In: Barrett’s esophagus and esophageal cancer: a comparative analysis using Montgomery EA, Voltaggio L. Biopsy Interpretation of the Gastrointestinal Tract cDNA microarrays. Oncogene. 2002;21(3):475-478. Mucosa. 2nd ed., vol. 1, non-neoplastic. Philadelphia: Lippincott Williams & 76. Helm J, Enkemann SA, Coppola D, et al. Dedifferentiation precedes Wilkins; 2012. invasion in the progression from Barrett’s metaplasia to esophageal adeno- 47. Montgomery E. Tumors of the esophagus. In: Iacobuzio-Donahue carcinoma. Clin Cancer Res. 2005;11(7):2478-2485. CA, Montgomery E. Gastrointestinal and Livery Pathology. Philadelphia: 77. Hao Y, Triadafilopoulos G, Sahbaie P, et al. Gene expression profiling Saunders; 2012. reveals stromal genes expressed in common between Barrett’s esophagus 48. Montgomery E, Bronner MP, Greenson JK, et al. Are ulcers a marker for and adenocarcinoma. Gastroenterology. 2006;131(3):925-933. invasive carcinoma in Barrett’s esophagus? Data from a diagnostic variability 78. Ito T, Tanaka E, Kadowaki T, et al. An ultrasensitive new DNA microarray study with clinical follow-up. Am J Gastroenterol. 2002;97(1):27-31. chip provides gene expression profiles for preoperative esophageal cancer 49. Zhu W, Appelman HD, Greenson JK, et al. A histologically defined biopsies without RNA amplification. Oncology. 2007;73(5-6):366-375. subset of high-grade dysplasia in Barrett mucosa is predictive of associated 79. Kan T, Meltzer SJ. MicroRNAs in Barrett’s esophagus and esophageal carcinoma. Am J Clin Pathol. 2009;132(1):94-100. adenocarcinoma. Curr Opin Pharmacol. 2009;9(6):727-732. 50. Konda VJ, Ross AS, Ferguson MK, et al. Is the risk of concomitant 80. Pacifici M, Delbue S, Ferrante P, et al. Cerebrospinal fluid miRNA invasive esophageal cancer in high-grade dysplasia in Barrett’s esophagus profile in HIV-encephalitis. J Cell Physiol. 2013;228(5):1070-1075. overestimated? Clin Gastroenterol Hepatol. 2008;6(2):159-164. 81. Yang H, Gu J, Wang KK, et al. MicroRNA expression signatures in 51. Weston AP, Sharma P, Topalovski M, et al. Long-term follow-up of Barrett’s esophagus and esophageal adenocarcinoma. Clin Cancer Res. Barrett’s high-grade dysplasia. Am J Gastroenterol. 2000;95(8):1888-1893. 2009;15(18):5744-5752. 52. Bani-Hani KE, Bani-Hani BK. Columnar lined (Barrett’s) esophagus: future perspectives. J Gastroenterol Hepatol. 2008;23(2):178-191. 53. Booth CL, Thompson KS. Barrett’s esophagus: a review of diagnostic criteria, clinical surveillance practices and new developments. J Gastrointest Oncol. 2012;3(3):232-242. 54. Lomo LC, Blount PL, Sanchez CA, et al. Crypt dysplasia with surface maturation: a clinical, pathologic, and molecular study of a Barrett’s esophagus cohort. Am J Surg Pathol. 2006;30(4):423-435. 55. Coco DP, Goldblum JR, Hornick JL, et al. Interobserver variability in the diagnosis of crypt dysplasia in Barrett esophagus. Am J Surg Pathol. 2011;35(1):45-54. 56. Voltaggio L, Montgomery EA. Histopathology of Barret esophagus. Diagn Histopoathol. 2011;17(2):41-49. 57. Morales TG, Sampliner RE, Bhattachayya A. Intestinal metaplasia of the gastric cardia. Am J Gastroenterol. 1997;92(3):414-418. 58. Spechler SJ. The role of gastric carditis in metaplasia and neoplasia at the gastroesophageal junction. Gastroenterology. 1999;117(1):218-28. 59. Brand DL, Ylvisaker JT, Gelfand M, et al. Regression of columnar esophageal (Barrett’s) epithelium after anti-reflux surgery. N Engl J Med.

April 2015, Vol. 22, No. 2 Cancer Control 185 Pathologists play a crucial role

in the multidisciplinary care

of patients with sarcoma.

A Quiet Sound. Photograph courtesy of Craig Damlo. www.soapboxrocket.com.

Molecular Pathology of Soft-Tissue Neoplasms and Its Role in Clinical Practice Evita B. Henderson-Jackson, MD, and Marilyn M. Bui, MD, PhD

Background: Soft-tissue neoplasms embody a histologically diverse group of mesenchymal tumors. Oftentimes the histopathological diagnosis of soft-tissue tumors is challenging due to overlapping pathological features. Methods: We reviewed the current and most importantly known recurrent or tumor-specific genetic abnormalities involving soft-tissue tumors, focusing on how they are useful in working up differential diagnoses and the relevance of potentially targeted therapies. Results: Molecular diagnostic tools have shown great advantage as an aid in the differentiation between different soft-tissue tumor entities, providing a potential avenue in the identification of novel therapeutic targets. Gastrointestinal stromal tumor is a well-known example of a soft-tissue tumor with a successful, molecularly driven treatment with response rates of more than 80% in stable disease and partial remission. Classifying soft-tissue neoplasms by their molecular genetic pathology has been considered as molecular testing becomes more integrated into various diagnostic and prognostic algorithms. Conclusions: Molecular pathology provides a unique opportunity for pathologists to play a crucial role in the multidisciplinary care of patients with sarcoma. These opportunities include but are not limited to the appropriate triage of tissue for molecular testing and the integration of molecular testing results, with histological and immunohistochemical findings providing actionable information for the diagnosis, prognosis, and choice of therapeutic modality.

Introduction Benign and malignant soft-tissue neoplasms are rare and have diverse clinical and biological behavior. This From the Departments of Anatomic Pathology (EBH, MMB) and heterogeneous group of mesenchymal tumors are di- Chemical Biology & Molecular Medicine Program (MMB), H. Lee agnosed and classified using criteria from the World Moffitt Cancer Center & Research Institute, the Departments of Health Organization (WHO); more than 100 benign Pathology and Cell Biology (EBH, MMB), Oncologic Sciences (EBH, MMB), University of South Florida Morsani College of Medicine, and malignant soft-tissue tumor entities have been Tampa, Florida. defined.1 These tumors are predominantly classified Submitted September 10, 2014; accepted December 28, 2014. by their histogenesis and incorporated cytogenetic Address correspondence to Marilyn M. Bui, MD, PhD, Department and molecular genetic information. In practice, pa- of Anatomic Pathology, Moffitt Cancer Center, 12902 Magnolia thologists use a morphological pattern recognition Drive, Tampa, FL 33612. E-mail: [email protected] approach to analyze the histological information. The No significant relationships exist between the authors and the companies/organizations whose products or services may be tumors can be arbitrarily divided into different mor- referenced in this article. phological groups such as round cell, spindle, epi-

186 Cancer Control April 2015, Vol. 22, No. 2 thelioid/polygonal, pleomorphic, adipocytic, myxoid, Ewing sarcoma, the second most common sarcoma and giant cell, among others. Each group consists of of bone in children and young adults, occurs in ex- a collection of pertinent differential diagnoses from traskeletal sites in about 10% to 20% of cases.1,7 The various histogenesis. Immunohistochemistry (IHC) is histopathology and IHC profile may mimic poorly an important ancillary tool to aid in diagnosis because differentiated synovial sarcoma with round cell mor- it can provide information about tumor histogenesis.2 phology.3,8-10 Molecular testing is often necessary Cytogenetic and molecular genetic analyses to establish an accurate diagnosis. Ewing sarcoma of mesenchymal neoplasms have revealed 2 major is known to demonstrate recurrent translocations. groups: (1) soft-tissue neoplasms associated with t(11;22)(q24;q12) is detected in approximately 85% complex karyotypes, and (2) soft-tissue neoplasms of Ewing sarcomas and results in the fusion of EWSR characterized by recurrent chromosomal structural on chromosome 22 to a member of the ETS gene abnormalities, gene amplification, mutations, or loss family of transcription factors, FLI1, on chromosome of heterozygosity. Although soft-tissue neoplasms with 11.11,12 Approximately 5% to 10% of Ewing sarcoma complex karyotypes do not demonstrate reproducible cases are associated with t(21;22)(q22;12), which is molecular alterations precluding the use of routine a fusion of EWSR1 and ERG.13 Rare cases (< 1%) of molecular testing for diagnostic purposes, the latter Ewing sarcoma demonstrate FUS-ERG or FUS-FEV fu- group can benefit from routine molecular testing to sions, likely because the fused in sarcoma (FUS) pro- facilitate a definitive diagnosis. tein amino-acid sequence is similar to Ewing sarcoma With the continuous advancement in our under- breakpoint region 1 (EWSR1) and is considered part standing of the cytogenetic and molecular genetic of the tet methylcytosine dioxygenase (TET) family of pathology of soft-tissue neoplasms, pathologists eval- proteins.14 The EWSR1 gene rearrangement is shared uating soft-tissue tumors must rely on histopathology by numerous sarcomas, including Ewing sarcoma, for diagnosis and integrate cytogenetic and molec- myxoid liposarcoma, angiomatoid fibrous histiocyto- ular genetic information into diagnostic algorithms ma, myoepithelioma, myoepithelial carcinoma, mixed for accurate diagnoses. Fusion genes resulting from tumor of soft tissue, clear cell sarcoma of soft tissue, chromosomal rearrangements in soft-tissue neoplasms extraskeletal myxoid chondrosarcoma, malignant represent an important part of the pathologist’s di- gastrointestinal neuroectodermal tumor (previously agnostic algorithm for tumor classification.3 Typical- referred to as clear cell sarcoma-like gastrointestinal ly, chromosomal translocations and fusion genes are tumor), low-grade fibromyxoid sarcoma, and des- present in the early phase of disease and persist in moplastic small round cell tumor.1,5,6,15 Recently, the disease progression, metastatic development, or both, WHO classification described a group of small round demonstrating that molecular testing for these types cell sarcomas with features similar to Ewing sarcoma of genetic alterations would be diagnostically useful. (“Ewing-like” sarcoma) but classified the group as Furthermore, the identification of particular soft-tissue undifferentiated round cell sarcomas because the tu- tumoral genetic aberrations is important for therapy. mors either demonstrated rearrangements of EWSR1 In this article, we provide an overview of the with non-ETS gene partners or lacked rearrangement important recurrent or tumor-specific molecular ab- of EWSR1 or other TET family members.14,16-22 A small errations associated with soft-tissue tumors, their dif- subset of these Ewing-like sarcomas (also known as ferential diagnoses, the role of molecular testing in undifferentiated round cell sarcomas) demonstrates the clinical management of soft-tissue tumors, and CIC-DUX4 gene fusion resulting from t(4:19)(q35;q13) potentially relevant therapeutic targets. or t(10;19)(q26;q13) and is primarily described within the pediatric population.19,23-25 Moreover, the Round Cell Tumors CIC-FOXO4 gene fusion, t(X;19)(q13;q13.3), has also Small round blue cell tumors encompass a group of been identified in 2 cases of Ewing-like sarcoma.26,27 In primitive-appearing neoplasms difficult to diagnosis addition, an emerging group of Ewing-like sarcomas on histopathology alone. The differential diagnoses with the BCOR-CCNB3 fusion gene, arising from an include a neoplasm of epithelial, mesenchymal, he- X chromosome paracentric inversion, may represent matopoietic, or melanocytic origin. IHC may be help- a biologically distinct entity within undifferentiated ful in narrowing the differential, as it is particularly round cell sarcomas.28-30 Simple reverse transcription effective for identifying carcinoma, neuroendocrine polymerase chain reaction (RT-PCR) assay in con- carcinoma, lymphoma, plasmacytoma, and melanoma. junction with cyclin B3 (CCNB3) IHC can be useful However, IHC is often ineffective in distinguishing in diagnosing these tumors. Whether these Ewing-like between different round cell sarcomas.4 Most round sarcoma cases should be classified as Ewing sarcoma blue cell tumors are associated with a tumor-specific or represent separate tumor types is unknown, but chromosomal translocation, which is diagnostically they have similar treatment.31 Further identification valuable to pathologists.1,5,6 and clinical follow-up of Ewing-like sarcomas with

April 2015, Vol. 22, No. 2 Cancer Control 187 distinct gene rearrangements are warranted to evaluate patient outcomes. Alveolar rhabdomyosarcoma (ARMS), often seen in adolescents and young adults, carries 2 specific chromosomal translocations.32 t(2;13)(q35;q14) occurs in approximately 60% of cases, whereas t(1;13) (p36;q14) occurs in a smaller subset.33-36 These translocations involve FOXO1A on chromosome 13q14 and either PAX3 (2q35) or PAX7 (1p36).33 No specific chromosomal abnormality has been described in embryonal rhabdomyosarcoma.37 Thus, the identification of t(2;13) or t(1;13) is diagnostically valuable and prognostically significant because the PAX-FOXO1A Fig. — Tumor positive for rearrangement of the EWSR1 (22q12) locus. fusion status imparts an unfavorable outcome for chil- This can be seen in DSRCT, but it is not specific because other sarcomas dren with rhabdomyosarcoma.3 A report of event-free may share this marker. EWSR1-WT1 fusion (yellow signal) is detected by survival and overall survival rates at 5 years correlated fluorescence in situ hybridization, which is specific and confirms the diagnosis of DSRCT. © Moffitt Cancer Center. Photographer: Julia A. Bridge, MD. histopathological subtype with PAX-FOXO1A status DSCRT = desmoplastic small round cell tumor. and showed that fusion-negative ARMS (lacking a detectable fusion of PAX3 or PAX7 with FOXO1A) had which represents a component of t(9:22) located on an outcome similar to embryonal rhabdomyosarco- chromosome 9, would confirm the diagnosis of EMC. ma and superior event-free survival rates compared This is because NR4A3 fusions are unique or specific with ARMS with either fusions of PAX3 or PAX7 with for EMC and are not present in other sarcomas; in FOXO1A when given therapy designed for children fact, NR4A3 fusions are present in more than 90% of with intermediate-risk rhabdomyosarcoma.38,39 These EMC cases.1,5 findings support the incorporation of PAX-FOXO1A fusion status into risk stratification and treatment.40 Spindle Cell Tumors Desmoplastic small round cell tumor primarily af- Clinically important, tumor-specific genetic abnormal- fects children and young adults and typically presents ities have been identified in spindle cell neoplasms. with widespread abdominal serosal involvement, ex- These genetic alterations are diagnostically helpful hibits polyphenotypic differentiation, and has a con- because the differential diagnoses of spindle cell sistent association with t(11;22)(p13;q12) and fusion neoplasms in soft tissue are diverse. Differentiation of EWSR1 and WT1.41-43 Because EWSR1 is shared by between tumor subtypes, such as monophasic syno- other sarcomas, identification of the partner gene is vial sarcoma, leiomyosarcoma, malignant peripheral warranted for a specific diagnosis (Fig). nerve sheath tumor, fibrosarcoma, and gastrointestinal Extraskeletal myxoid chondrosarcoma (EMC) oc- stromal tumor (GIST), may be challenging depending curs in adults and, rarely, children, demonstrating on specimen adequacy (quantity and quality), the no convincing evidence of cartilaginous differentia- immunostaining profile, and clinical presentation. tion; it is characterized by NR4A3 rearrangement.44-46 Once the findings on the standard histopathological More than 90% of cases of EMC harbor either t(9:22) examination and IHC workup are inconclusive, mo- (q22;q12) or, less frequently, t(9;17)(q22;q11) result- lecular testing is a necessary adjunct. For example, ing in the fusion of NR4A3 (9q22) to either EWSR1 synovial sarcoma can be difficult to discriminate from (22q12) or TAF15 (17q12).47-50 IHC is not helpful in solitary fibrous tumor, malignant peripheral nerve the diagnosis of EMC; therefore, molecular testing sheath tumor, or sarcomatoid carcinoma. Given the is essential. For example, EMC and mixed tumor of ability to detect 95% of cases of synovial sarcoma soft-tissue and myoepithelioma have overlapping his- through identification of its recurrent reciprocal tological and IHC features; thus, these entities present t(X;18)(p11.2;q11.2), which fuses SYT (18q11) to 1 a diagnostic challenge. Considering the difference in of the 3 homologous genes on Xp11 (SSX1, SSX2, or therapies and outcomes of EMC and mixed tumor SSX4), molecular testing is essential for classification of soft-tissue and myoepithelioma, an accurate and and treatment.1,52,53 definitive diagnosis is critical for the management of Dermatofibrosarcoma protuberans, a low-grade the condition. Because both tumors may show EWSR1 spindle cell neoplasm, harbors either supernumer- gene rearrangement and 75% of EMC cases and 45% of ary ring chromosomes or unbalanced derivatives of mixed tumor of soft-tissue and myoepithelioma cases t(17;22)(q22;q13). Both aberrations contain a chimeric exhibit this rearrangement,51 fluorescence in situ hy- gene that fuses COL1A1 with PDGFB commonly iden- bridization (FISH) testing of NR4A3 would be the ideal tified by multiplex RT-PCR or, preferably, FISH.54-59 test to perform. Positive NR4A3 gene rearrangement, Typically, dermatofibrosarcoma protuberans is treated

188 Cancer Control April 2015, Vol. 22, No. 2 by wide local excision; however, in cases of unresect- expression, which can be a potential pitfall, particular- able and metastatic disease, targeted therapy with ly in retroperitoneal masses for which the differential imatinib mesylate may be clinically useful.60,61 Molec- diagnosis also includes SFT.77 ular testing is important for identifying patients who GIST is a mesenchymal tumor characterized by may have a clinical response.62 A total of 10% to 15% activating oncogenic mutations rather than specific of cases demonstrate fibrosarcomatous progression translocations or fusion genes and has a clinical spec- and may not show a response.1 trum that ranges from benign to malignant.1,78 Most A rare malignant fibroblastic neoplasm, low- GISTs harbor a mutation in 1 of 3 sites: KIT exon 9, grade fibromyxoid sarcoma, consistently has either KIT exon 11, or PDGFRA exon 18.79,80 Tyrosine kinase FUS-CREB3L2 gene fusion or, less frequently, inhibitors (eg, imatinib mesylate, sunitinib malate) FUS-CREB3L1 gene fusion and results in t(7;16) have been used to successfully treat GISTs.81 Mutation- or t(11;16), respectively.63,64 A small number of al analysis has become essential to evaluate patient low-grade fibromyxoid sarcomas harbor EWSR1-CRE- prognosis, make treatment decisions, predict treat- B3L1 gene fusion.65 Sclerosing epithelioid fibrosarco- ment response, and select appropriate dosages.78,82-84 ma (SEF), which is another rare malignant fibroblastic Most laboratories utilize PCR to amplify the most tumor with considerable morphological overlap with commonly mutated exons with subsequent direct se- low-grade fibromyxoid sarcoma, harbors frequent quencing analysis of the amplified exon.78 Recently, EWSR1 gene rearrangements; a minority of cases BRAF V600E mutation was identified in patients with exhibits FUS-CREB3L2 fusions.66 Both low-grade fibro- GIST lacking KIT and PDGFRA mutations.85,86 In ad- myxoid sarcoma and SEF show mucin 4 expression by dition, cases of GIST are typically associated with the IHC.67 Thus, the presence of morphological features Carney triad and Carney–Stratakis syndrome when reminiscent of low-grade fibromyxoid sarcoma in SEFs they show mutations in SDH-related genes.87,88 These as well as the presence of FUS rearrangements suggest tumors have a distinct morphology. that a possible relationship may exist between the 2 entities and that they are perhaps part of a diseas Lipomatous Tumors e spectrum. A few genetic alterations exist that involve lipomatous Congenital or infantile fibrosarcoma, which is his- neoplasms and aid in discriminating tumor subtypes tologically similar to adult fibrosarcoma, has a distinc- with similar morphological features. Myxoid/round tive t(12;15)(p13;q26) resulting in the ETV6-NTRK3 cell liposarcoma may be challenging to diagnose fusion. This neoplasm occurs in children within the when the differential diagnosis includes other myxoid first year of life and the histological features may neoplasms such as EMC or myxofibrosarcoma — this mimic other pediatric spindle cell neoplasms, such as is particularly true on biopsy specimens. However, infantile fibromatosis and infantile myofibromatosis approximately 95% of cases of myxoid liposarcoma or myofibroma.3,68,69 It is often difficult to cytogeneti- demonstrate the FUS-DDIT3 chimeric gene due to cally identify the ETV6-NTRK3 fusion, so it is typically reciprocal t(12;16)(q13;p11).43,89,90 Incorporating mo- detected by FISH or PCR.70,71 lecular testing with histopathology is useful in these Congenital or infantile spindle cell rhabdomyo- cases because myxoid liposarcomas are sensitive to sarcoma is another differential diagnosis in this radiation therapy and select patients receive neoad- group. This tumor typically lacks PAX3-FOXO1 and juvant therapy.91 PAX7-FOXO1 fusions but exhibits recurrent NCOA2 Another diagnostic challenge is the differentia- arrangements.72 tion between lipoma and atypical lipomatous tumor A breakthrough in the molecular pathology of (ALT)/well-differentiated liposarcoma (WDL). Lipoma solitary fibrous tumor (SFT) is the identification of is a mass-forming tumor with mature-appearing fat; the recurrent NAB2-STAT6 fusion by integrative se- however, differently from normal fat, lipoma exhibits quencing.73 NAB2-STAT6 has been established as a HMGA2 translocation.92 The diagnostic atypical cells the defining driver mutation of SFT. Fusion variants in fibrous septa can be scarce, the atypia may be cyto- NAB2ex4-STAT6ex2/3 and NAB2ex6-STAT6ex16/17 logically subtle, or the ALT/WDL may demonstrate a have been identified; the former fusion occurs in pleu- lipoma-like morphological pattern. ALT/WDL is like- ropulmonary SFTs and mostly exhibits a benign be- ly to recur and carries the risk of dedifferentiation, havior, whereas the latter fusion occurs in deep-seated which results in a poor prognosis depending on the extrapleural SFTs and has more aggressive behav- anatomical location. For clinical purposes, molecular ior.74 STAT6 IHC can be a useful adjunct tool in the testing is useful in discriminating between lipoma and diagnosis of SFT, particularly in cases with aberrant ALT/WDL, and this is particularly true if clinicora- morphology and limited material.75,76 Of note, STAT6 diological information is inconclusive. Furthermore, amplification has been described in a small subset of dedifferentiated liposarcoma with predominant, dedifferentiated liposarcomas resulting in STAT6 IHC high-grade dedifferentiated areas may be difficult

April 2015, Vol. 22, No. 2 Cancer Control 189 to discriminate from other high-grade pleomorphic Table 2. — Genetics of Myxoid Sarcomas sarcomas. Molecular testing that identifies super- Tumor Type Defect Gene numerary ring chromosomes and/or giant-marker Myxoma Activating Gs-α GNAS chromosomes corresponding to amplification of the mutations 12q13-15 band support a diagnosis of ALT/WDL or Low-grade fibromyxoid t(7;16)(q33;p11) CREB3L2-FUS dedifferentiated liposarcoma depending on histopa- sarcoma/hyalinizing spindle t(11;16)(p11;p11) CREB3L1-FUS 3,52 cell tumor with giant rosettes thology. Of note, some lipomas have gains of the t(16;xx)(p11) EWSR1-CREB3L1 mouse double minute 2 homolog (MDM2) and some Myxoid liposarcoma t(12;16)(q13;p11) DDIT3-FUS other mesenchymal tumors can have MDM2 amplifi- t(12;22)(q13;q12) DDIT3-EWSR1 cation or expression (Table 1).23,93 Extraskeletal myxoid t(9;22)(q22;q12) NR4A3-EWSR1 chondrosarcoma t(9;17)(q22;q11) NR4A3-TAF2N Tumor of Uncertain Histogenesis t(9;15)(q22;q21) NR4A3-TCF12 Tumors of uncertain histogenesis have a range of t (3;9(q11;q22) NR4A3-TFG morphological features from round cells to spindle Myxofibrosarcoma None None cells to epithelioid/polygonal cells. Some have char- characteristic acteristic translocations as well as unique and recog- Myxoinflammatory t(1;10) TGFBR3-MGEA5 nizable histopathology and clinical features. However, fibroblastic sarcoma significant histopathological and IHC overlap exists between clear cell sarcoma (malignant melanoma of histologic entity characterized as a benign, soft parts) and conventional melanoma. Clear cell fibrovascular soft-tissue tumor of uncertain cel- sarcomas possess a recurrent EWSR1-ATF1 fusion in lular origin. This tumor is characterized by an more than 90% of cases from a reciprocal t(12;22) AHRR-NCOA2 fusion resulting from t(5;8) (q13;q12).94,95 A related variant, t(2;22)(q32.3;q12), (p15;q13).105,106 The evaluation of NCOA2 rearrange- the EWSR-CREB1 fusion, has been reported in a small ments using FISH is a practical method to confirm a subset of clear cell sarcomas.96,97 Molecular testing is diagnosis of soft-tissue angiofibroma.105 necessary to establish an unequivocal diagnosis of clear cell sarcoma.98 In addition, rare clear cell sarco- Myxoid Tumors mas have been reported to contain BRAF mutations, Myxoid tumors do not represent a standalone group possibly representing a therapeutic target.99,100 of tumors per the WHO criteria.1 However, this type Of note, alveolar soft-part sarcoma, which is a of tumor is commonly encountered in clinical prac- rare sarcoma composed of nests of tumor cells lined tice. The tumor cells may have various morphological by sinusoidal vascular channels that may have eosin- features and they produce abundant myxoid stroma. ophilic, granular-to-clear cytoplasm, may be mistaken The common genetic change of this group of tumor for renal cell carcinoma or perivascular epithelioid is summarized in Table 2. Among this group, myx- cell tumor.1 On IHC, alveolar soft-part sarcoma is neg- oinflammatory fibroblastic sarcoma exhibits t(1;10) ative for keratin, epithelial membrane antigen, and involving TGFBR3 and MGEA5.107,108 HMB-45. Molecular testing is diagnostically helpful to confirm the diagnosis. Alveolar soft-part sarcoma Conclusions is defined by recurrent, unbalanced der(17)t(X;17) Soft-tissue tumors form a heterogeneous and complex (p11;q25) involving the fusion of TFE3 (Xp11) and group with diverse morphology and, at times, a non- ASPSCR1 (17q25).101,102 Although ASPSCR1-TFE3 ap- specific immunohistochemical profile. The availability pears specific for alveolar soft-part sarcoma, this same of tumor-specific genetic markers has modified the gene fusion has been identified in a small subset of routine diagnostic workup of soft-tissue neoplasms. renal cell carcinomas.103,104 The cytogenetic and molecular identification of chro- Soft-tissue angiofibroma is a relatively new mosomal translocations and their associated gene fusions, loss, and gains of specific chromosomal regions Table 1. — MDM2-Positive Soft-Tissue Tumors and activating or inactivating mutations of select on- Tumor Type MDM2 by FISH MDM2 by IHC cogenes or tumor-suppressor genes have contributed to therapeutic and prognostic assessments and the Lipoma – – classification of soft-tissue neoplasms. The integration Atypical lipomatous tumor/ + + (nuclear) well-differentiated liposarcoma of morphology, immunohistochemistry, and molecular pathology is essential for appropriate and accurate Dedifferentiated liposarcoma + + (diffuse, nuclear) diagnostic algorithms. A better understanding of how Intimal sarcoma + + (≤ 70%) to apply the molecular pathology of soft-tissue neo- FISH = fluorescence in situ hybridization, IHC = immunohistochemistry, plasms will provide new roles for pathologists in the MDM2 = mouse double minute 2 homolog. diagnostics and targeted therapy of these tumors.3,109

190 Cancer Control April 2015, Vol. 22, No. 2 33. Barr FG, Galili N, Holick J, et al. Rearrangement of PAX3 paired box References gene in the paediatric solid tumour alveolar rhabdomyosarcoma. Nat Genet. 1. Fletcher CD, Bridge JA, Hogendoor PC, eds. Classification of Tumours 1993;3(2):113-117. of Soft Tissue and Bone. 4th ed. Lyon, France: IARC Press; 2013. 34. Douglass EC, Shapiro DN, Valentine M. Alveolar rhabdomyosarcoma 2. Coppola D, ed. Molecular Pathology and Diagnostics of Cancer. New with the t(2;13): cytogenetic findings and clinicopathologic correlations. Med York: Springer; 2013. Pediatr Oncol. 1993;21(2):83-87. 3. Bridge JA. The role of cytogenetics and molecular diagnostics in the 35. Wachtel M, Dettling M, Koscielniak E, et al. Gene expression signa- diagnosis of soft-tissue tumors. Mod Pathol. 2014;27(suppl 1):S80-S97. tures identify rhabdomyosarcoma subtypes and detect a novel t(2;2)(q35;p23) 4. Meis-Kindblom JM, Stenman G, Kindblom LG. Differential diagnosis translocation fusing PAX3 to NCOA1. Cancer Res. 2004;64(16):5539-5545. of small round cell tumors. Sem Diagn Pathol. 1996;13(3):213-241. 36. Barr FG. Gene fusions involving PAX and FOX family members in 5. Fletcher C, Uni K, Mertens F, eds. Pathology and Genetics of Tumours alveolar rhabdomyosarcoma. Oncogene. 2001;20(40):5736-5746. of Soft Tissue and Bone. Lyon, France: IARC Press; 2002. 37. Bridge JA, Liu J, Qualman SJ, et al. Genomic gains and losses are 6. Wei S, Siegal GP. Round cell tumors of bone: an update on recent similar in genetic and histologic subsets of rhabdomyosarcoma, whereas molecular genetic advances. Adv Anat Pathol. 2014;21(5):359-372. amplification predominates in embryonal with anaplasia and alveolar subtypes. 7. Gurney JG, Davis S, Severson RK, et al. Trends in cancer incidence Gene Chromosomes Cancer. 2002;33(3):310-321. among children in the U.S. Cancer. 1996;78(3):532-541. 38. Raney RB, Anderson JR, Barr FG, et al. Rhabdomyosarcoma and undif- 8. Bridge JA, Fidler ME, Neff JR, et al. Adamantinoma-like Ewing’s sarcoma: ferentiated sarcoma in the first two decades of life: a selective review of intergroup genomic confirmation, phenotypic drift. Am J Surg Pathol. 1999;23(2):159-165. rhabdomyosarcoma study group experience and rationale for Intergroup Rhab- 9. Gu M, Antonescu CR, Guiter G, et al. Cytokeratin immunoreactivity in domyosarcoma Study V. J Pediatr Hematol Oncol. 2001;23(4):215-220. Ewing’s sarcoma: prevalence in 50 cases confirmed by molecular diagnostic 39. Parham DM, Ellison DA. Rhabdomyosarcomas in adults and children: studies. Am J Surg Pathol. 2000;24(3):410-416. an update. Arch Pathol Lab Med. 2006;130(10):1454-1465. 10. Fisher C. Synovial sarcoma. Ann Diagn Pathol. 1998;2(6):401-421. 40. Skapek SX, Anderson J, Barr FG, et al. PAX-FOXO1 fusion status 11. Delattre O, Zucman J, Plougastel B, et al. Gene fusion with an ETS drives unfavorable outcome for children with rhabdomyosarcoma: a children’s DNA-binding domain caused by chromosome translocation in human tumours. oncology group report. Pediatr Blood Cancer. 2013;60(9):1411-1417. Nature. 1992;359(6391):162-165. 41. Ladanyi M, Gerald WL. Specificity of the EWS/WT1 gene fusion for 12. Turc-Carel C, Aurias A, Mugneret F, et al. Chromosomes in Ewing’s desmoplastic small round cell tumor. J Pathol. 1996;180(4):462. sarcoma. I. An evaluation of 85 cases of remarkable consistency of t(11;22) 42. Gerald WL, Miller HK, Battifora H, et al. Intra-abdominal desmoplastic (q24;q12). Cancer Genet Cytogenet. 1988;32(2):229-238. small round-cell tumor. Report of 19 cases of a distinctive type of high-grade 13. Sorensen PH, Lessnick SL, Lopez-Terrada D, et al. A second Ewing’s polyphenotypic malignancy affecting young individuals. Am J Surg Pathol. sarcoma translocation, t(21;22), fuses the EWS gene to another ETS-family 1991;15(6):499-513. transcription factor, ERG. Nat Genet. 1994;6(2):146-151. 43. Ordóeñz J, Osuna D, García-Domínguez DJ, et al. The clinical relevance of 14. Tan AY, Manley JL. The TET family of proteins: functions and roles in molecular genetics in soft tissue sarcomas. Adv Anat Pathol. 2010;17(3):162-181. disease. J Mol Cell Biol. 2009;1(2):82-92. 44. Hachitanda Y, Tsuneyoshi M, Daimaru Y, et al. Extraskeletal myxoid 15. Kong J, Li N, Wu S, et al. Malignant gastrointestinal neuroectodermal chondrosarcoma in young children. Cancer. 1988;61(12):2521-2526. tumor: a case report and review of the literature. Oncol Lett. 2014;8(6):2687-2690. 45. Meis-Kindblom JM, Bergh P, Gunterberg B, et al. Extraskeletal myxoid 16. Yamaguchi S, Yamazaki Y, Ishikawa Y, et al. EWSR1 is fused to POU5F1 chondrosarcoma: a reappraisal of its morphologic spectrum and prognostic in a bone tumor with translocation t(6;22)(p21;q12). Genes Chromosomes factors based on 117 cases. Am J Surg Pathol. 1999;23(6):636-650. Cancer. 2005;43(2):217-222. 46. Drilon AD, Popat S, Bhuchar G, et al. Extraskeletal myxoid chondro- 17. Szuhai K, Ijszenga M, de Jong D, et al. The NFATc2 gene is involved sarcoma: a retrospective review from 2 referral centers emphasizing long-term in a novel cloned translocation in a Ewing sarcoma variant that couples its outcomes with surgery and chemotherapy. Cancer. 2008;113(12):3364-3371. function in immunology to oncology. Clin Cancer Res. 2009;15(7):2259-2268. 47. Antonescu C, Argani P, Erlandson RA, et al. Skeletal and extraskeletal 18. Mastrangelo T, Modena P, Tornielli S, et al. A novel zinc finger gene is myxoid chondrosarcoma: a comparative clinicopathologic, ultrastructural, and fused to EWS in small round cell tumor. Oncogene. 2000;19(33):3799-3804. molecular study. Cancer. 1998;83(8):1504-1521. 19. Kawamura-Saito M, Yamazaki Y, Kaneko K, et al. Fusion between CIC 48. Panagopoulos I, Mertens F, Isaksson M. Molecular genetic characteri- and DUX4 up-regulates PEA3 family genes in Ewing-like sarcomas with t(4;19) zation of the EWS/CHN and RBP56/CHN fusion genes in extraskeletal myxoid (q35;q13) translocation. Hum Mol Genet. 2006;15(13):2125-2137. chondrosarcoma. Gene Chromosomes Cancer. 2002;35(4):340-352. 20. Deng FM, Galvan K, de la Roza G, et al. Molecular characterization 49. Hisaoka M, Hashimoto H. Extraskeletal myxoid chondrosarcoma: of an EWSR1-POU5F1 fusion associated with a t(6;22) in an undifferentiated updated clinicopathological and molecular genetic characteristics. Pathol Int. soft tissue sarcoma. Cancer Genet. 2011;204(8):423-429. 2005;55(8):453-463. 21. National Cancer Institute; Cancer Genome Anatomy Project. Mitelman 50. Sjögren H, Meis-Kindblom JM, Orndal C, et al. Studies on the molecular Database of Chromosome Aberrations and Gene Fusions in Cancer: searching pathogenesis of extraskeletal myxoid chondrosarcoma-cytogenetic, molecular the database. Updated February 2015. http://cgap.nci.nih.gov/Chromosomes/ genetic, and cDNA microarray analyses. Am J Surg Pathol. 2003;162(3):781-792. Mitelman. Accessed March 5, 2015. 51. Nishio J, Iwasaki H, Nabeshima K, Naito M. Cytogenetics and molecular 22. Wang L, Bhargava R, Zheng T, et al. Undifferentiated small round cell genetics of myxoid soft-tissue sarcomas. Genet Res Int. 2011;2011:497148. sarcomas with rare EWS gene fusions: identification of a novel EWS-SP3 52. Igbokwe A, Lopez-Terrada DH. Molecular testing of solid tumors. Arch fusion and of additional cases with the EWS-ETV1 and EWS-FEV fusions. Pathol Lab Med. 2011;135(1):67-82. J Mol Diagn. 2007;9(4):498-509. 53. de Leeuw B, Balemans M, Olde Weghuis D, et al. Identification of two 23. Italiano A, Cardot N, Dupré F, et al. Gains and complex rearrangements alternative fusion genes, SYT-SXX1 and SYT-SSX2, in t(X;18)(p11.2;q11.2)-pos- of the 12q13-15 chromosomal region in ordinary lipomas: the “missing link” itive synovial sarcomas. Hum Mol Genet. 1995;4(6):1097-1099. between lipomas and liposarcomas? Int J Cancer. 2007;121(2):308-315. 54. Patel KU, Szabo SS, Hernandez VS, et al. Dermatofibrosarcoma protu- 24. Richkind KE, Romansky SG, Finklestein JZ. t(4;19)(q35;q13.1): a berans COL1A1-PDGFB fusion is identified in virtually all dermatofibrosarcoma recurrent change in primitive mesenchymal tumors? Cancer Genet Cytogenet. protuberans cases when investigated by newly developed multiplex reverse 1996;87(1):71-74. transcription polymerase chain reaction and fluoresence in situ hybridization 25. Yoshimoto M, Graham C, Chilton-MacNeill S, et al. Detailed cytogenetic assays. Hum Pathol. 2008;39(2):184-193. and array analysis of pediatric primitive sarcomas reveals a recurrent CIC- 55. Naeem R, Lux ML, Huang SF, et al. Ring chromosomes in derma- DUX4 fusion gene event. Cancer Genet Cytogenet. 2009;195(1):1-11. tofibrosarcoma protuberans are composed of interspersed sequences from 26. Solomon DA, Brohl AS, Khan J, et al. Clinicopathologic features of a chromosomes 17 and 22. Am J Pathol. 1995;147(6):1553-1558. second patient with Ewing-like sarcoma harboring CIC-FOXO4 gene fusion. 56. Terrier-Lacombe MJ, Guillou L, Maire G, et al. Dermatofibrosarcoma Am J Surg Pathol. 2014;38(12):1724-1725. protuberans, giant cell fibroblastoma, and hybrid lesions in children: clinico- 27. Sugita S, Arai Y, Tonooka A, et al. A novel CIC-FOXO4 gene fusion pathologic comparative analysis of 28 cases with molecular data--a study from in undifferentiated small round cell sarcoma: a genetically distinct variant of the French Federation of Cancer Centers Sarcoma Group. Am J Surg Pathol. Ewing-like sarcoma. Am J Surg Pathol. 2014;38(11):1571-1576. 2003;27(1):27-39. 28. Pierron G, Tirode F, Lucchesi C, et al. A new subtype of bone sarcoma 57. Simon MP, Pedeutour F, Sirvent N, et al. Deregulation of the plate- defined by BCOR-CCNB3 gene fusion. Nat Genet. 2012;44(4):461-466. let-derived growth factor B-chain gene via fusion with collagen gene COL1A1 29. Puls F, Niblett A, Marland G, et al. BCOR-CCNB3 (Ewing-like) sarcoma: in dermatofibrosarcoma protuberans and giant-cell fibroblastoma. Nat Genet. a clinicopathologic analysis of 10 cases, in comparison with conventional Ewing 1997;15(1):95-98. sarcoma. Am J Surg Pathol. 2014;38(10):1307-1318. 58. Salgado R, Llombart B, M Pujol R, et al. Molecular diagnosis of derma- 30. Peters TL, Kumar V, Polikepahad S, et al. BCOR-CCNB3 fusions tofibrosarcoma protuberans: a comparison between reverse transcriptase-poly- are frequent in undifferentiated sarcomas of male children. Mod Pathol. merase chain reaction and fluorescence in situ hybridization methodologies. 2015;28(4):575-586. Gene Chromosomes Cancer. 2011;50(7):510-517. 31. Sankar S, Lessnick SL. Promiscuous partnerships in Ewing’s sarcoma. 59. Llombart B, Sanmartin O, López-Guerrero JA, et al. Dermatofibrosar- Cancer Genet. 2011;204(7):351-365. coma protuberans: clinical, pathological, and genetic (COL1A1-PDGFB) study 32. Parham DM. Pathologic classification of rhabdomyosarcomas and with therapeutic implications. Histopathology. 2009;54(7):860-872. correlations with molecular studies. Mod Pathol. 2001;14(5):506-514. 60. Rubin BP, Schuetze SM, Eary JF, et al. Molecular targeting of plate-

April 2015, Vol. 22, No. 2 Cancer Control 191 let-derived growth factor B by imatinib mesylate in a patient with metastatic gastrointestinal stromal tumours. J Clin Pathol. 2009;62(7):613-616. dermatofibrosarcoma protuberans. J Clin Oncol. 2002;20(17):3586-3591. 87. Janeway KA, Kim SY, Lodish M, et al. Defects in succinate dehydroge- 61. McArthur GA, Demetri GD, van Oosterom A, et al. Molecular and clinical nase in gastrointestinal stromal tumors lacking KIT and PDGFRA mutations. analysis of locally advanced dermatofibrosarcoma protuberans treated with Proc Natl Acad Sci U S A. 2011;108(1):314-318. imatinib: Imatinib Target Exploration Consortium Study B2225. J Clin Oncol. 88. Wagner AJ, Remillard SP, Zhang YX, et al. Loss of expression of SDHA 2005;23(4):866-873. predicts SDHA mutations in gastrointestinal stromal tumors. Mod Pathol. 62. Rutkowski P, Van Glabbeke M, Rankin CJ et al; European Organisation 2013;26(2):289-294. for Research and Treatment of Cancer Soft Tissue/Bone Sarcoma Group; 89. Sandberg AA. Updates on the cytogenetics and molecular genet- Southwest Oncology Group. Imatinib mesylate in advanced dermatofibrosar- ics of bone and soft tissue tumors: liposarcoma. Cancer Genet Cytogenet. coma protuberans: pooled analysis of two phase II clinical trials. J Clin Oncol. 2004;155(1):1-24. 2010;28(10):1772-1779. 90. Powers MP, Wang WL, Hernandez VS,e t al. Detection of myxoid 63. Storlazzi CT, Mertens F, Nascimento A, et al. Fusion of the FUS liposarcoma-associated FUS-DDIT3 rearrangement variants including a and BBF2H7 genes in low grade fibromyxioid sarcoma. Hum Molec Genet. newly identified breakpoint using an optimized RT-PCR assay. Mod Pathol. 2003;12(18):2349-2358. 2010;23(10):1307-1315. 64. Mertens F, Fletcher CD, Antonescu CR, et al. Clinicopathologic and 91. Flanagan AM, Delaney D, O’Donnell P. The benefits of molecular pa- molecular genetic characterization of low-grade fibromyxoid sarcoma, and thology in the diagnosis of musculoskeletal disease: part 1 of a two part review: cloning of a novel FUS/CREB3L1 fusion gene. Lab Invest. 2005;85(3):408-415. soft tissue tumors. Skeletal Radiol. 2010;39(2):105-115. 65. Lau PP, Lui PC, Lau GT, et al. EWSR1-CREB3L1 gene fusion: a novel 92. Bianchini L, Birtwisle L, Saâda E, et al. Identification of PPAP2B as alternative molecular aberration of low-grade fibromyxoid sarcoma. Am J Surg a novel recurrent translocation partner gene of HMGA2 in lipomas. Gene Pathol. 2013;37(5):734-738. Chromosomes Cancer. 2013;52(6):580-590. 66. Prieto-Granada C, Zhang L, Chen HW, et al. A genetic dichotomy 93. Mandahl N, Bartuma H, Magnusson L, et al. HMGA2 and MDM2 ex- between pure sclerosing epithelioid fibrosarcoma (SEF) and hybrid SEF/low- pression in lipomatous tumors with partial, low-level amplification of sequences grade fibromyxoid sarcoma: a pathologic and molecular study of 18 cases. from the long arm of chromosome 12. Cancer Genet. 2011;204(10):550-556. Gene Chromosomes Cancer. 2015;54(1):28-38. 94. Bridge JA, Sreekantaiah C, Neff JR, et al. Cytogenetic findings in clear 67. Doyle LA, Wang WL, Dal Cin P, et al. MUC4 is a sensitive and extremely cell sarcoma of tendons and aponeuroses. Malignant melanoma of soft parts. useful marker for sclerosing epithelioid fibrosarcoma: association with FUS Cancer Genet Cytogenet. 1991;52(1):101-106. gene rearrangement. Am J Surg Pathol. 2012;36(10):1444-1451. 95. Zucman J, Delattre O, Desmaze C, et al. EWS and ATF-1 gene fusion 68. Coffin CM, Dehner LP. Fibroblastic-myofibroblastic tumors in children induced by t(12;22) translocation in malignant melanoma of soft parts. Nat and adolescents: a clinicopathologic study of 108 examples in 103 patients. Genet. 1993;4(4):341-345. Pediatr Pathol. 1991;11(4):569-588. 96. Hisaoka M, Ishida T, Kuo TT, et al. Clear cell sarcoma of soft tissue: a 69. Alaggio R, Barisani D, Ninfo V, et al. Morphologic overlap between clinicopathologic, immunohistochemical, and molecular analysis of 33 cases. infantile myofibromatosis and infantile fibrosarcoma: a pitfall in diagnosis. Pe- Am J Surg Pathol. 2008;32(3):452-460. diatr Dev Pathol. 2008;11(5):355-362. 97. Wang WL, Mayordomo E, Zhang W, et al. Detection and characteri- 70. Schofield DE, Fletcher JA, Grier HE, et al. Fibrosarcoma in infants and zation of EWSR1/ATF1 and EWSR1/CREB1 chimeric transcripts in clear cell children. Application of new techniques. Am J Surg Pathol. 1994;18(1):14-24. sarcoma (melanoma of soft parts). Mod Pathol. 2009;22(9):1201-1209. 71. Argani P, Fritsch M, Kadkol SS, et al. Detection of the ETV6-NTRK3 chi- 98. Sandberg AA, Bridge JA. Updates on the cytogenetics and molecular meric RNA of infantile fibrosarcoma/cellular congenital mesoblastic nephroma genetics of bone and soft tissue tumors: clear cell sarcoma (malignant mela- in paraffin-embedded tissue: application to challenging pediatric renal stromal noma of soft parts). Cancer Genet Cytogenet. 2001;130(1):1-7. tumors. Mod Pathol. 2000;13(1):29-36. 99. Park BM, Jin SA, Choi YD, et al. Two cases of clear cell sarcoma with 72. Mosquera JM, Sboner A, Zhang L, et al. Recurrent NCOA2 gene re- different clinical and genetic features: cutaneous type with BRAF mutation and arrangements in congenital/infantile spindle cell rhabdomyosarcoma. Gene subcutaneous type with KIT mutation. Br J Dermatol. 2013;169(6):1346-1352. Chromosomes Cancer. 2013;52(6):538-550. 100. Negri T, Brich S, Conca E, et al. Receptor tyrosine kinase pathway 73. Robinson DR, Wu YM, Kalyana-Sundaram S, et al. Identification of analysis sheds light on similarities between clear-cell sarcoma and metastatic recurrent NAB2-STAT6 gene fusions in solitary fibrous tumor by integrative melanoma. Gene Chromosomes Cancer. 2012;51(2):111-126. sequencing. Nat Genet. 2013;45(2):180-185. 101. Heimann P, Devalck C, Debusscher C, et al. Alveolar soft-part sarcoma: 74. Barthelmeß S, Geddert H, Boltze C, et al. Solitary fibrous tumors/ further evidence by FISH for the involvement of chromosome band 17q25. hemangiopericytomas with different variants of the NAB2-STAT6 gene fusion Gene Chromosomes Cancer. 1998;23(2):194-197. are characterized by specific histomorphology and distinct clinicopathological 102. Ladanyi M, Lui MY, Antonescu CR, et al. The der(17)t(X;17)(p11;q25) features. Am J Pathol. 2014;184(4):1209-1218. of human alveolar soft part sarcoma fuses the TFE3 transcription factor gene 75. Vogels R, Vlenterie M, Versleijen-Jonkers Y, et al. Solitary fibrous tumor: to ASPL, a novel gene at 17q25. Oncogene. 2001;20(1):48-57. clinicopathologic, immunohistochemical and molecular analysis of 28 cases. 103. Argani P, Antonescu CR, Illei PB, et al. Primary renal neoplasms with Diagn Pathol. 2014;9(1):224. the ASPL-TFE3 gene fusion of alveolar soft part sarcoma: a distinctive tumor 76. Cheah AL, Billings SD, Goldblum JR, et al. STAT6 rabbit monoclonal entity previously included among renal cell carcinomas of children and ado- antibody is a robust diagnostic tool for the distinction of solitary fibrous tumour lescents. Am J Pathol. 2001;159(1):179-192. from its mimics. Pathology. 2014;46(5):389-395. 104. Heimann P, El Housni H, Ogur G, et al. Fusion of a novel gene, RCC17, 77. Doyle LA, Tao D, Mariño-Enríquez A. STAT6 is amplified in a subset to the TFE3 gene in t(X;17)(p11.2;q25.3)-bearing papillary renal cell carcino- of dedifferentiated liposarcoma. Mod Pathol. 2014;27(9):1231-1237. mas. Cancer Res. 2001;61(10):4130-4135. 78. Bridge JA, Cushman-Vokoun AM. Molecular diagnostics of soft tissue 105. Sugita S, Aoyama T, Kondo K, et al. Diagnostic utility of NCOA2 flu- tumors. Arch Pathol Lab Med. 2011;135(5):588-601. orescence in situ hybridization and Stat6 immunohistochemistry staining for 79. Debiec-Rychter M, Sciot R, Le Cesne A, et al; EORTC Soft Tissue and soft tissue angiofibroma and morphologically similar fibrovascular tumors. Hum Bone Sarcoma Group; Italian Sarcoma Group Australasian GastroIntestinal Trials Pathol. 2014;45(8):1588-1596. Group. KIT mutations and dose selection for imatinib in patients with advanced 106. Jin Y, Möller E, Nord KH, et al. Fusion of the AHRR and NCOA2 genes gastrointestinal stromal tumours. Eur J Cancer. 2006;42(8):1093-1103. through a recurrent translocation t(5;8)(p15;q13) in soft tissue angiofibroma 80. Andersson J, Bümming P, Meis-Kindblom JM, et al. Gastrointestinal results in upregulation of aryl hydrocarbon receptor target genes. Gene Chro- stromal tumors with KIT exon 11 deletions are associated with poor prognosis. mosomes Cancer. 2012;51(5):510-520. Gastroenterology. 2006;130(6):1573-1581. 107. Antonescu CR, Zhang L, Nielsen GP, et al. Consistent t(1;10) with 81. Quek R, George S. Update on the treatment of gastrointestinal stromal rearrangements of TGFBR3 and MGEA5 in both myxoinflammatory fibroblas- tumors (GISTs): role of imatinib. Biologics. 2010;4:19-31. tic sarcoma and hemosiderotic fibrolipomatous tumor. Gene Chromosomes 82. Heinrich MC, Corless CL, Blanke CD, et al. Molecular correlates Cancer. 2011;50(10):757-764. of imatinib resistance in gastrointestinal stromals tumors. J Clin Oncol. 108. Ieremia E, Thway K. Myxoinflammatory fibroblastic sarcoma: morpho- 2006;24(29):4764-4774. logic and genetic updates. Arch Pathol Lab Med. 2014;138(10):1406-1411. 83. Debiec-Rychter M, Dumez H, Judson I, et al; EORTC Soft Tissue and 109. Al-Zaid T, Somaiah N, Lazar AJ. Targeted therapies for sarcomas: new Bone Sarcoma Group. Use of c-KIT/PDGFRA mutational analysis to predict roles for the pathologist. Histopathology. 2014;64(1):119-133. the clinical response to imatinib in patients with advanced gastrointestinal stromal tumours entered on phase I and II studies of the EORTC Soft Tissue and Bone Sarcoma Group. Eur J Cancer. 2004;40(5):689-695. 84. Heinrich MC, Corless CL, Demetri GD, et al. Kinase mutations and imatinib response in patients with metastatic gastrointestinal stromal tumor. J Clin Oncol. 2003;21(23):4342-4349. 85. Agaram NP, Wong GC, Guo T, et al. Novel V600E BRAF mutations in imatinib-naive and imatinib-resistant gastrointestinal stromal tumors. Gene Chromosomes Cancer. 2008;47(10):853-859. 86. Agaimy A, Terracciano LM, Dirnhofer S, et al. V600E BRAF mutations are alternative early molecular events in a subset of KIT/PDGFRA wild-type

192 Cancer Control April 2015, Vol. 22, No. 2 Measuring EGFR mutation status in lung

cancer is standard for therapy selection.

The Old Barn. Photograph courtesy of Craig Damlo. www.soapboxrocket.com.

Advances in EGFR as a Predictive Marker in Lung Adenocarcinoma Farah K. Khalil, MD, and Soner Altiok, MD, PhD

Background: Worldwide, lung cancer is the most common cause of mortality. Toxins from tobacco smoke are known to increase the risk of lung cancer; however, up to 15% of lung cancer–related deaths in men and up to 50% of lung cancer–related deaths in women occur in people who do not smoke. Despite the fact that chemotherapy generally provides a survival benefit for non–small-cell lung cancer, not every patient will respond to therapy and many experience therapy-related adverse events. Thus, predictive markers are used to determine which patients are more likely to respond to a given regimen. Methods: We reviewed the current medical literature in English relating to predictive markers that may be positive, such as the presence of an activating EGFR mutation. Results: The advances in using EGFR as a molecular predictive marker were summarized. This biomarker influences therapeutic response in patients with lung adenocarcinoma. Clinical evidence supporting its value is also reviewed. Conclusions: The use of EGFR as a predictive factor in lung adenocarcinoma may help target therapy to individual tumors to achieve the best likelihood for long-term survival and to avoid adverse events from medications unlikely to be effective.

Survival and Response Rates with erlotinib, carboplatin, paclitaxel, and bevacizum- Across all treatments and tumor stages, the 5-year ab in addition to radiation and consolidation thera- overall survival (OS) rate for non–small-cell lung py with erlotinib and bevacizumab have objective cancer (NSCLC) is 16%.1 Patients receiving induc- response rates between 39% and 60% and median tion chemotherapy with carboplatin, paclitaxel, and progression-free survival (PFS) of 10.2 months to bevacizumab followed by concurrent chemotherapy 18.4 months.2 Response rates for patients with metastatic NSCLC provided treatment are 17% to 37%; From the Department of Anatomic Pathology, H. Lee Moffitt Cancer median survival rates are 6.7 to 11.3 months; and Center & Research Institute, and the Departments of Pathology and Cell Biology and Oncological Sciences, Morsani College of Medicine, 1- and 2-year survival rates are 31% to 46% and 9% University of South Florida, Tampa, Florida. to 21%, respectively.1,3-5 Time to progression averag- Submitted February 11, 2015; accepted April 16, 2015. es 4 to 6 months and response rate to second-line Address correspondence to Farah K. Khalil, MD, Department of therapy is 8%.6 Anatomic Pathology, Moffitt Cancer Center, 12902 Magnolia Drive, Tampa, FL 33612. E-mail: [email protected] EGFR and Tyrosine Kinase Inhibitors No significant relationships exist between the authors and the companies/organizations whose products or services may be The most promising and widely reported molecular referenced in this article. predictive factor in NSCLC is EGFR, which resides on

April 2015, Vol. 22, No. 2 Cancer Control 193 the short (p) arm of chromosome 7 at position 12; epidermal growth factor (EGF) was first identified 7-9 by Cohen in 1962. When EGF binds to epidermal EGF EGF growth factor receptor (EGFR), the receptor dimerizes, autophosphorylates, and activates several pathways, including mitogen-activated protein kinase, Janus kinase 2/signal transducer and activator of transcription (STAT) 3, STAT5, and phosphatidylinositol 3-kinase (PI3K)/protein kinase B pathways, which lead to cell proliferation, metastasis, and migration while EGFR preventing apoptosis.6,10-14 Because the EGF normally favors growth and proliferation, mutations of its receptor render it constitutively active independent of ligand-binding, which may lead to a malignant phenotype (Fig 1).15,16 An increased EGFR copy number src PTEN or overexpression may play a role in oncogenesis. For example, based on their mechanisms of action, EGFR RAF PI3K monoclonal antibodies (eg, cetuximab, matuzumab, panitumumab), which bind the extracellular domain MEK Akt of EGFR and prevent ligand binding, may be beneficial to patients with tumors overexpressing surface EGFR, MAPK whereas tyrosine kinase inhibitors (TKIs) may benefit patients whose tumor proliferation and metastases are 7,17,18 Proliferation Invasion driven by EGFR autophosphorylation. Blocking Metastasis Survival the catalytic site of the receptor, which is responsible Angiogenesis for activating downstream molecules, may prevent growth and proliferation and possibly favor apoptosis Fig 1. — EGFR involved in tumorigenesis of non–small-cell lung in cancer cells. cancer. Akt = protein kinase B, EGF = epidermal growth factor, Erlotinib and gefitinib, which are first-generation EGFR = epidermal growth factor receptor, MEK/MAPK, mitogen-activated reversible TKIs, target the catalytic domain of EGFR protein kinase, PI3K = phosphatidylinositol 3-kinase, PTEN = phosphatase 6,17,19 and tensin homolog, RAF = v-raf 1 murine leukemia viral oncogene homolog by competing with adenosine triphosphate. Com- 1, src = proto-oncogene tyrosine-protein kinase Src. pared with placebo, erlotinib has been shown in a randomized, placebo-controlled, phase 3 trial to pro- exon 19 (codons 746–750) or substitution of argi- vide significant survival benefit in unselected individ- nine for leucine at position 858 in exon 21 (L858R).25 uals with advanced-stage NSCLC who received prior Another 3% are substitutions of a variety of amino treatment.8 The researchers also found that those with acids in place of glycine at codon 719 (G719X) in exon certain demographic characteristics responded better 18 and 3% are in-frame insertions in exon 20.12,19,26-30 to erlotinib.8 Other studies have demonstrated that The overall incidence of EGFR mutations in NSCLC patients who are Japanese, women, and those with ranges from 12.1% to 49%, depending on the patient adenocarcinomas have high response rates to gefi- population.19,28-31 Tyrosine kinase domain mutations tinib.12,20-22 Patients who have never smoked also have of EGFR are rare in tumors in non-lung locations.11 higher response rates, longer times to progression, The exon 19 deletions and L858R substitution are and longer median OS rates when treated with TKIs activating mutations that result in increased phos- than those who do smoke.20,23 Although the results of phorylation of EGFR independent of ligand stimu- the phase 3 trial demonstrated that erlotinib provides lation.11,12,20,32 These mutant EGFRs have greater af- a survival benefit compared with placebo,8 another finity for TKIs and are susceptible to tyrosine kinase phase 3 trial found that gefitinib was noninferior to inhibition in vitro.16,32 Both Ji et al15 and Politi et al16 docetaxel and had no benefit over docetaxel even in demonstrated that expression of either the exon 19 subgroups of women, those with adenocarcinomas, deletion or the L858R substitution mutation in EGFR those who have never smoked, or Asians.24 causes adenocarcinoma to develop in mice and that tumor regression takes place when either expression Molecular Predictors of Response to Tyrosine of the mutant gene is blocked or erlotinib was admin- Kinase Inhibitors and Survival Rates istered. Greulich et al33 obtained similar results in cell The different mutations identified in EGFR are in ex- cultures and confirmed that these mutations result in ons 19 (58%), 21 (36%), and 18 (6%).25 Approximately constitutive ligand-independent receptor activation, 90% of the mutations are either small deletions from which is further enhanced by stimulating the ligand.

194 Cancer Control April 2015, Vol. 22, No. 2 The usefulness of the EGFR mutation status as a and PFS rates following first-line chemotherapy are predictive factor for response to TKIs is determined also significantly longer in patients with EGFR-mutant through the results of clinical trials and retrospec- tumors compared with those with wild-type EGFR tive studies. Lynch et al34 demonstrated that EGFR tumors. Moreover, use of gefitinib as first- or sec- mutations are associated with a clinical response to ond-line therapy resulted in significantly longer PFS gefitinib and showed that in vitro EGFR mutants were rates among patients with EGFR-mutant tumors than more sensitive to gefitinib. No mutations were present those treated with chemotherapy.48 In a series con- in patients who did not respond to gefitinib.34 Both ducted by Kosaka et al,30 397 individuals with EGFR in vitro exon 19 deletion mutants and L858R mutants mutations survived significantly longer than those were more sensitive to gefitinib than tumors with with wild-type mutations; no survival difference was wild-type EGFR.34 In addition, all 8 mutants Lynch seen among those with an exon 19 deletion and L858R et al34 identified were heterozygotes, indicating that substitution. these mutations act in a dominant manner, similar to other oncogenes. Paez et al23 confirmed these results, Molecular Alterations Conferring Resistance and Pao et al35 soon added that EGFR mutations can to Tyrosine Kinase Inhibitors also be found in tumors sensitive to erlotinib. Since Although many tumors with EGFR mutations initially then, other studies have confirmed the association respond to TKIs, disease progression will typically between EGFR mutation status and response to TKIs,36-41 occur within 6 to 12 months because of the selective and a review by Mitsudomi et al12 who compiled survival of tumor cells whose acquired mutations have data from 1,335 patients showed that EGFR TKI conferred resistance to these agents, as well as due response was observed in 70% of EGFR-mutant tumors to pre-existing–resistant subclones, the up regulation compared with 10% of EGFR wild-type tumors. In a of other parallel signaling pathways, or the transfor- phase 2 trial, EGFR-mutant tumors responded better mation of tumor cells to small cell carcinoma.12,18,49-51 to gefitinib than wild-type EGFR tumors.38 Kalikaki Even small clones of tumor cells containing a mutation et al42 demonstrated that patients with classic-acti- conferring resistance to TKIs may have a significant vating EGFR mutations treated with gefitinib had a clinical impact.50,52 The most common secondary EGFR significantly longer median OS rate than those with mutation leading to TKI resistance is the substitution wild-type EGFR; however, this relationship was not of methionine in place of threonine at codon 790 present in patients with other types of EGFR muta- (T790M) within exon 20.18,49,53-55 After identifying the tions. Significantly higher rates of objective tumor presence of the T790M point mutation in a patient response to TKIs in EGFR mutants compared with initially sensitive to gefitinib but who later became wild-type EGFR have been seen in several studies,19,29,40,43 resistant, Kobayashi et al54 demonstrated in vitro that whereas others have found no such relationship.44 introducing the T790M mutation into previously gefi- Differential response to treatment between the tinib-sensitive NSCLC cells harboring either wild-type, different types of EGFR mutations have been observed an exon 19 deletion, or L858R-substitution EGFR gen- by some authors. Mitsudomi et al12 reported that dele- otypes induced gefitinib resistance. Pao et al53 also tional mutations had higher response rates to gefitinib demonstrated in vitro that NSCLC cells harboring the than other mutations — in particular, L858R. In addi- T790M mutation with otherwise wild-type or mutated tion, Hirsch et al45 found that patients who had exon EGFR (either exon 19 deletion or L858R substitution) 19 deletions alone had significantly higher objective did not respond to either erlotinib or gefitinib. response rates than patients with exon 21 mutations. Structural analysis has also shown that the By contrast, Pao et al35 found that in vitro wild-type T790M point mutation of EGFR changes tyrosine EGFR and exon 19 deletion mutants similarly respond- in the catalytic domain to methionine, which is ed to gefitinib and erlotinib, whereas L858R mutants bulkier and prevents erlotinib — and presumably were approximately 10 times as sensitive. gefitinib — from binding.18,26,54,56 The T790M muta- Not every patient with an activating mutation of tion alters TKI binding and enhances the kinase ac- EGFR responds to TKIs.46 Alternatively, some patients tivity of EGFR when coupled with the L858R muta- without EGFR mutations still respond to TKIs.26,43,46 tion, thus offering cells with this mutation a survival Yoshioka et al47 published a phase 2 prospective study advantage.12,20,32,57 The T790M point mutation has and determined that EGFR wild-type tumors have a been identified in approximately 50% of patients modest response to erlotinib without irreversible who had activating EGFR mutations and later ac- toxicity. EGFR mutation status impacts treatment and quired resistance to EGFR TKIs.18,50,53,55,58,59 Engleman has a concomitant impact on survival. Longer median et al60 and Bonomi et al59 demonstrated in vitro that survival (but not PFS) rates after first-line chemother- malignant cells with activating mutations of EGFR apy have been observed in individuals with EGFR initially sensitive to gefitinib become resistant af- mutations.48 In addition, Hotta et al36 found that OS ter prolonged exposure to gefitinib, and they may

April 2015, Vol. 22, No. 2 Cancer Control 195 also acquire the T790M mutation in EGFR. Even growth factor binding, overexpression, or structural so, the question arises as to whether TKI treatment alterations) may be used in place of MET amplification induces the resistance mutation or if it results in to determine sensitivity to TKIs.66 The results of an in the selection of resistant clones.56 Depending on vitro study by Rho et al66 suggested that sensitivity to the sensitivity rate of the assay used, the T790M TKIs was not associated with MET activation in the mutation has been found in 0.5% to 3.6% of patients absence of MET amplification. who have never been treated with TKIs.12,50 Labo- ratory studies and clinical validation also indicate Methods to Overcome Acquired Resistance that the presence of pre-TKI EGFR T790M could to Tyrosine Kinase Inhibitors act as a negative predictor of PFS in patients with Although T790M makes cells resistant to gefitinib EGFR-activating mutations.12,50,58,61,62 and erlotinib (which are reversible TKIs), the ef- The exon 20 insertion mutation of EGFR fect of irreversible TKIs (which covalently bind the (D770insNPG) has been shown by Greulich et al33 EGFR kinase domain) is not changed by T790M. to confer erlotinib and gefitinib resistance in vitro; Irreversible TKIs include afatinib, dacomitinib, nera- however, the response was obtained from the ir- tinib, pelitinib, canertinib, and CL-387,785.54,56,58,60,67 reversible EGFR inhibitor CL-387,785. In another Irreversible EGFR inhibitors such as afatinib inhibit study, 2 women with adenocarcinoma harboring EGFR and downstream molecules more potently this mutation who never smoked both progressed than gefitinib or erlotinib in cells harboring the despite gefitinib treatment.41 Intermediate sensitiv- T790M mutation, although not to the degree ob- ity to erlotinib or gefitinib was observed in G719S served in cells without the T790M mutation.54,60,66 of exon 18, and that sensitivity may be related to Regales et al49 demonstrated in mice that tumors the specific mutation present.18,33,59 harboring the T790M TKI–resistance mutation Another missense mutation, D761Y in exon 19, had more shrinkage (and a complete response in has been reported to confer a milder degree of TKI most tumors) when treated with combination afa- resistance that may be clinically relevant.58 Greater tinib/cetuximab (an EGFR-specific antibody) when resistance to gefitinib is produced when the T790M compared with other antitumor agents. This combi- mutation occurs on the same DNA strand as the nation resulted in lower total and phosphorylated EGFR-activating mutation (L858R or exon 19 dele- EGFR.49 Neither agent alone induced a complete tion),60 which is the most common scenario.63 This response in any of the tumors in the study.49 An- can be explained by understanding that the mutant other study showed that CUDC-305, a heat-shock allele alone will code for the mutant, constitutive- protein 90 inhibitor, had inhibited tumor growth ly active EGFR; likewise, binding of TKIs to these in mouse xenograft models of NSCLC with EGFR mutant, constitutively active EGFRs will be altered T790M mutations.68 only if the resistance mutation has occurred on the Blocking signaling molecules downstream of mutant allele. EGFR has the potential to overcome the resistance Low PTEN expression is an important reason for induced by various mutations that alter the binding tumors becoming resistant to TKIs.64,65 PTEN normally of drugs to EGFR.69,70 Src is one such downstream functions to down-regulate the PI3K pathway (down- signaling molecule vital to maintaining the malignant stream of EGFR; it mediates growth, proliferation, and phenotype of EGFR-mutant cells.69 In vitro, Src in- survival). Loss of PTEN leads to persistent activation hibitors have been shown to prevent oncogenesis of the PI3K pathway independent of EGFR signaling, caused by EGFR mutations, suggesting that Src in- making this proliferation signal unresponsive to the hibitors may be of clinical benefit in TKI-resistant EGFR blockade.64,65 Among gefitinib-resistant cells EGFR-mutated NSCLC.69 Faber et al70 demonstrated lacking PTEN, Bianco et al65 demonstrated that the in vitro that blocking 2 major pathways downstream reconstitution of PTEN reversed EGFR-independent from EGFR resulted in tumor shrinkage, even in cell PI3K pathway activity and restored gefitinib sensi- lines with various TKI-resistant mutations, including tivity. Zhuang et al64 demonstrated that irradiating T790M. Afatinib and dacomitinib have shown sus- TKI-resistant tumors with low PTEN expression may tained control of disease progression in patients with help reverse the resistance by increasing PTEN ex- NSCLC who have primary or acquired TKI resistance.51 pression levels. Although low PTEN expression may The LUX-Lung 1 phase 2/2B trial investigated afatinib induce resistance to TKIs, a relationship may not exist compared with placebo in patients with NSCLC who between PTEN expression and rate of survival.10 were progressing following treatment with erlotinib or Some authors have reported acquired resistance gefitinib and combination platinum-based chemother- to EGFR-resistant TKIs in tumors with MET ampli- apy. No significant improvement in OS rate was seen fication.12,60,66 It has also been suggested that MET when compared with placebo.51 Individuals naive to activation (which may be induced by hepatocyte EGFR-TKI treatment with stage 3B and stage 4 NSCLC

196 Cancer Control April 2015, Vol. 22, No. 2 who harbored activating EGFR mutations were studied during the LUX-Lung 2 trial.51 These study volunteers were given once-daily dosing of afatinib (40 or 50 mg), and similar overall response rates were seen; however, more adverse events were seen among those given the higher dose.51 The phase 3 trial, LUX-Lung 3, randomized 345 individuals with EGFR mutation-positive adenocarcinoma to either once-daily 40 mg afatinib or combination cisplatin/pemetrexed.71 In that trial, 72% of study volunteers were of Asian descent.71 The afatinib group showed a prolonged PFS rate (11.1 vs 6.9 months) compared with the cisplatin/pemetrexed group.71 Patients with common EGFR mutations (eg, L858R, exon 19 deletion) had a maximum PFS rate of 13.6 months when treated with afatinib, whereas those assigned to chemotherapy had a maximum PFS Fig 2. — Image of adenocarcinoma showing micropapillary features. rate of 6.9 months.51,72 clinical outcome in which adenocarcinoma in situ and The effective afatinib dose has toxicity limitations minimally invasive adenocarcinoma had excellent that has necessitated continual studies, thus leading outcomes; lepidic, acinar, and papillary predominant to third-generation, irreversible TKIs, such as rocile- adenocarcinoma had intermediate outcomes, where- tinib, AZD9291, and HM61713, which have shown as solid and micropapillary predominant and inva- profound and sustained regression in T790M-mu- sive mucinous adenocarcinoma had poor outcomes.78 tant–selective resistance in NSCLC.73,74 Preliminary Fig 2 illustrates invasive adenocarcinoma of the lung reports have shown response rates of 58%, 64%, and with a micropapillary growth pattern. 29% for rociletinib, AZD9291, and HM61713, respec- Reclassification studies of adenocarcinoma based tively.75,76 AZD9291 does not have the same severe on the recommendations from the International adverse events as those seen with first-generation Association for the Study of Lung Cancer, the American TKIs because it does not specifically target wild-type Thoracic Society, and the European Respiratory Soci- EGFR, it is associated with reduced rates of skin rash ety International indicate that the lepidic pattern is and diarrhea, and it has an overall response rate of associated with 44% of EGFR-mutant lung cancers as 64% in patients positive for T790M.74 A recent trial compared with the acinar patern, which is associated with showed that no dose-limiting toxicities were present 69% of EGFR–wild-type lung cancers.77,79 at doses between 20 and 240 mg/day.75 Conclusions Correlation of Histological Subtype of Of all the molecular alterations that may have pre- Adenocarcinoma With EGFR-Mutation Status dictive value in non–small-cell lung cancer, testing NSCLC and small cell carcinoma are 2 major catego- for EGFR mutations is usually the first step in de- ries of lung cancer based on histology and response termining the course of adjuvant therapy. Activat- to chemotherapy. The most common cell type of ing EGFR mutations and amplification predict the NSCLC is adenocarcinoma. The International Asso- response of non–small-cell lung cancer to tyrosine ciation for the Study of Lung Cancer, the American kinase inhibitors. The use of these predictive factors Thoracic Society, and the European Respiratory So- help target therapy to individual tumors to achieve ciety International proposed a new classification of the best likelihood for long survival; they are also lung adenocarcinoma in 2011.77 This international, used to help avoid unnecessary adverse events relat- multidisciplinary classification divided adenocarcino- ed to therapies unlikely to have any effect. Studies ma into histological subtypes based on survival data showing post–tyrosine kinase inhibitor resistance that indicated 5-year survival rates of 100% among due to mutations not detected prior to treatment patients with adenocarcinoma in situ (previously impress the utility of reassessing tumors following known as bronchioloalveolar carcinoma) and mini- treatment. Additional, large cohort studies are re- mally invasive adenocarcinoma.77 Invasive adenocar- quired to accurately determine the association be- cinoma subtypes include lepidic, acinar, papillary, tween adenocarcinoma subtypes and EGFR-mutation micropapillary, and solid predominant patterns. Mu- status. Mutational heterogeneity within a single tu- cinous, colloid, fetal, and enteric histology have been mor and morphological associations will most likely described as variants of adenocarcinoma.77 Results mandate the continual assessment of each type of from clinicopathological studies have identified a cancer as it progresses from primary to metastatic correlation between histological subtypes and the and pretreatment to post-treatment status.

April 2015, Vol. 22, No. 2 Cancer Control 197 and in combination with erlotinib. J Clin Oncol. 2005;23(25):5900-5909. References 29. Zhu CQ, da Cunha Santos G, Ding K, et al; National Cancer Institute 1. Bepler G, Begum M, Simon GR. Molecular analysis-based treatment of Canada Clinical Trials Group Study BR.21. Role of KRAS and EGFR as strategies for non-small cell lung cancer. Cancer Control. 2008;15(2):130-139. biomarkers of response to erlotinib. J Clin Oncol. 2008;26(26):4268-4275. 2. Ikeda N, Nagase S, Ohira T. Individualized adjuvant chemotherapy 30. Kosaka T, Yatabe Y, Onozato R, et al. Prognostic implication of EGFR, for surgically resected lung cancer and the roles of biomarkers. Ann Thorac KRAS, and TP53 gene mutations in a large cohort of Japanese patients with Cardiovasc Surg. 2009;15(3):144-149. surgically treated lung adenocarcinoma. J Thorac Oncol. 2009;4(1):22-29. 3. Bepler G, Sommers KE, Cantor A, et al. Clinical efficacy and predictive 31. Lynch TJ, Bell DW, Sordella R, et al. Activating mutations in the epider- molecular markers of neoadjuvant gemcitabine and pemetrexed in resectable mal growth factor receptor underlying responsiveness of non-small-cell lung non-small cell lung cancer. J Thorac Oncol. 2008;3(10):1112-1118. cancer to gefitinib. N Engl J Med. 2004;350(21):2129-2139. 4. Schiller JH, Harrington D, Belani CP, et al; Eastern Cooperative On- 32. Mulloy R, Ferrand A, Kim Y, et al. Epidermal growth factor receptor cology Group. Comparison of four chemotherapy regimens for advanced non- mutants from human lung cancers exhibit enhanced catalytic activity and small-cell lung cancer. N Engl J Med. 2002;346(2):92-98. increased sensitivity to gefitinib. Cancer Res. 2007;67(5):2325-2330. 5. Kelly K, Crowley J, Bunn PA Jr, et al. Randomized phase III trial of 33. Greulich H, Chen TH, Feng W, et al. Oncogenic transformation by paclitaxel plus carboplatin versus vinorelbine plus cisplatin in the treatment inhibitor-sensitive and -resistant EGFR mutants. PLoS Med. 2005;2(11):e313. of patients with advanced non--small-cell lung cancer: a Southwest Oncology 34. Lynch TJ, Bell DW, Sordella R, et al. Activating mutations in the epider- Group trial. J Clin Oncol. 2001;19(13):3210-3218. mal growth factor receptor underlying responsiveness of non-small-cell lung 6. Emery IF, Battelli C, Auclair PL, et al. Response to gefitinib and erlotinib cancer to gefitinib. N Engl J Med. 2004;350(21):2129-2139. in non-small cell lung cancer: a restrospective study. BMC Cancer. 2009;9:333. 35. Pao W, Miller V, Zakowski M, et al. EGF receptor gene mutations 7. Modjtahedi H, Essapen S. Epidermal growth factor receptor inhibitors in are common in lung cancers from “never smokers” and are associated with cancer treatment: advances, challenges and opportunities. Anticancer Drugs. sensitivity of tumors to gefitinib and erlotinib. Proc Natl Acad Sci U S A. 2009;20(10):851-855. 2004;101(36):13306-13311. 8. Shepherd FA, Rodrigues Pereira J, Ciuleanu T, et al. Erlotinib in previ- 36. Hotta K, Kiura K, Toyooka S, et al. Clinical significance of epidermal ously treated non-small-cell lung cancer. N Engl J Med. 2005;353(2):123-132. growth factor receptor gene mutations on treatment outcome after first-line 9. Cohen S. Isolation of a mouse submaxillary gland protein accelerating cytotoxic chemotherapy in Japanese patients with non-small cell lung cancer. incisor eruption and eyelid opening in the new-born animal. J Biol Chem. J Thorac Oncol. 2007;2(7):632-637. 1962;237:1555-1562. 37. Ichihara S, Toyooka S, Fujiwara Y, et al. The impact of epidermal growth 10. Cetin Z, Ozbilim G, Erdogan A, et al. Evaluation of PTEN and Mcl-1 factor receptor gene status on gefitinib-treated Japanese patients with non- expressions in NSCLC expressing wild-type or mutated EGFR. Med Oncol. small-cell lung cancer. Int J Cancer. 2007;120(6):1239-1247. 2010;27(3):853-860. 38. Bell DW, Lynch TJ, Haserlat SM, et al. Epidermal growth factor receptor 11. Mitsudomi T, Yatabe Y. Epidermal growth factor receptor in relation to mutations and gene amplification in non-small-cell lung cancer: molecular analysis tumor development: EGFR gene and cancer. FEBS J. 2010;277(2):301-308. of the IDEAL/INTACT gefitinib trials. J Clin Oncol. 2005;23(31):8081-8092. 12. Mitsudomi T, Yatabe Y. Mutations of the epidermal growth factor re- 39. Cappuzzo F, Hirsch FR, Rossi E, et al. Epidermal growth factor receptor ceptor gene and related genes as determinants of epidermal growth fac- gene and protein and gefitinib sensitivity in non-small-cell lung cancer. J Natl tor receptor tyrosine kinase inhibitors sensitivity in lung cancer. Cancer Sci. Cancer Inst. 2005;97(9):643-655. 2007;98(12):1817-1824. 40. Adams VR, Harvey RD. Histological and genetic markers for non-small- 13. Lord RV, Brabender J, Gandara D, et al. Low ERCC1 expression cor- cell lung cancer: customizing treatment based on individual tumor biology. Am relates with prolonged survival after cisplatin plus gemcitabine chemotherapy J Health Syst Pharm. 2010;67(1 suppl 1):S3-S9. in non-small cell lung cancer. Clin Cancer Res. 2002;8(7):2286-2291. 41. Han SW, Kim TY, Jeon YK, et al. Optimization of patient selection for 14. Yarden Y, Sliwkowski MX. Untangling the ErbB signalling network. Nat gefitinib in non-small cell lung cancer by combined analysis of epidermal Rev Mol Cell Biol. 2001;2(2):127-137. growth factor receptor mutation, K-ras mutation, and Akt phosphorylation. Clin 15. Ji H, Li D, Chen L, et al. The impact of human EGFR kinase domain Cancer Res. 2006;12(8):2538-2544. mutations on lung tumorigenesis and in vivo sensitivity to EGFR-targeted 42. Kalikaki A, Koutsopoulos A, Hatzidaki D, et al. Clinical outcome of patients therapies. Cancer Cell. 2006;9(6):485-495. with non-small cell lung cancer receiving front-line chemotherapy according to 16. Politi K, Zakowski MF, Fan PD, et al. Lung adenocarcinomas induced EGFR and K-RAS mutation status. Lung Cancer. 2010;69(1):110-115. in mice by mutant EGF receptors found in human lung cancers respond to a 43. Takano T, Ohe Y, Sakamoto H, et al. Epidermal growth factor re- tyrosine kinase inhibitor or to down-regulation of the receptors. Genes Dev. ceptor gene mutations and increased copy numbers predict gefitinib sen- 2006;20(11):1496-1510. sitivity in patients with recurrent non-small-cell lung cancer. J Clin Oncol. 17. Gerber DE. EGFR inhibition in the treatment of non-small cell lung 2005;23(28):6829-6837. cancer. Drug Dev Res. 2008;69(6):359-372. 44. Tsao MS, Sakurada A, Cutz JC, et al. Erlotinib in lung cancer - molec- 18. Uramoto H, Mitsudomi T. Which biomarker predicts benefit from EGFR-TKI ular and clinical predictors of outcome. N Engl J Med. 2005;353(2):133-144. treatment for patients with lung cancer? Br J Cancer. 2007;96(6):857-863. 45. Hirsch FR, Varella-Garcia M, Cappuzzo F, et al. Combination of 19. Hirsch FR, Varella-Garcia M, Bunn PA Jr, et al. Molecular predictors EGFR gene copy number and protein expression predicts outcome for ad- of outcome with gefitinib in a phase III placebo-controlled study in advanced vanced non-small-cell lung cancer patients treated with gefitinib. Ann Oncol. non-small-cell lung cancer. J Clin Oncol. 2006;24(31):5034-5042. 2007;18(4):752-760. 20. Rudin CM, Avila-Tang E, Harris CC, et al. Lung cancer in never 46. Miller VA, Riely GJ, Zakowski MF, et al. Molecular characteristics of bron- smokers: molecular profiles and therapeutic implications. Clin Cancer Res. chioloalveolar carcinoma and adenocarcinoma, bronchioloalveolar carcinoma 2009;15(18):5646-5661. subtype, predict response to erlotinib. J Clin Oncol. 2008;26(9):1472-1478. 21. Parra HS, Cavina R, Latteri F, et al. Analysis of epidermal growth factor 47. Yoshioka H, Hotta K, Kiura K, et al; Okayama Lung Cancer Study Group. receptor expression as a predictive factor for response to gefitinib (‘Iressa’, A phase II trial of erlotinib monotherapy in pretreated patients with advanced ZD1839) in non-small-cell lung cancer. Br J Cancer. 2004;91(2):208-212. non-small cell lung cancer who do not possess active EGFR mutations: Okaya- 22. Fukuoka M, Yano S, Giaccone G, et al. Multi-institutional randomized ma Lung Cancer Study Group trial 0705. J Thorac Oncol. 2010;5(1):99-104. phase II trial of gefitinib for previously treated patients with advanced non- 48. Yoshida K, Yatabe Y, Park J, et al. Clinical outcomes of advanced non- small-cell lung cancer (the IDEAL 1 trial) [Erratum appears in J Clin Oncol. small cell lung cancer patients screened for epidermal growth factor receptor 2004;22(23):4863]. J Clin Oncol. 2003;21(12):2237-2246. gene mutations. J Cancer Res Clin Oncol. 2010;136(4):527-535. 23. Paez JG, Jänne PA, Lee JC, et al. EGFR mutations in lung can- 49. Regales L, Gong Y, Shen R, et al. Dual targeting of EGFR can over- cer: correlation with clinical response to gefitinib therapy. Science. come a major drug resistance mutation in mouse models of EGFR mutant 2004;304(5676):1497-1500. lung cancer. J Clin Invest. 2009;119(10):3000-3010. 24. Kim SO, Jeong JY, Kim MR, et al. Efficacy of gemcitabine in patients 50. Inukai M, Toyooka S, Ito S, et al. Presence of epidermal growth factor with non-small cell lung cancer according to promoter polymorphisms of the receptor gene T790M mutation as a minor clone in non-small cell lung cancer. ribonucleotide reductase M1 gene. Clin Cancer Res. 2008;14(10):3083-3088. Cancer Res. 2006;66(16):7854-7858. 25. Marchetti A, Martella C, Felicioni L, et al. EGFR mutations in non-small- 51. Köhler J, Schuler M. et al. Afatinib, erlotinib and gefitinib in the first-line cell lung cancer: analysis of a large series of cases and development of a rapid therapy of EGFR mutation-positive lung adenocarcinoma: a review. Onkologie. and sensitive method for diagnostic screening with potential implications on 2013;36(9):510-518. pharmacologic treatment. J Clin Oncol. 2005;23(4):857-865. 52. Marchetti A, Milella M, Felicioni L, et al. Clinical implications of KRAS mu- 26. Mountzios G, Dimopoulos MA, Soria JC, et al. Histopathologic and tations in lung cancer patients treated with tyrosine kinase inhibitors: an important genetic alterations as predictors of response to treatment and survival in lung role for mutations in minor clones. Neoplasia. 2009;11(10):1084-1092. cancer: a review of published data. Crit Rev Oncol Hematol. 2010;75(2):94-109. 53. Pao W, Miller VA, Politi KA, et al. Acquired resistance of lung adeno- 27. Lynch TJ, Bell DW, Sordella R, et al. Activating mutations in the epider- carcinomas to gefitinib or erlotinib is associated with a second mutation in the mal.growth factor receptor underlying responsiveness of non-small-cell lung EGFR kinase domain. PLoS Med. 2005;2(3):e73. cancer to gefitinib. N Engl J Med. 2004;350(21):2129-2139. 54. Kobayashi S, Boggon TJ, Dayaram T, et al. EGFR mutation and resistance 28. Eberhard DA, Johnson BE, Amler LC, et al. Mutations in the epidermal of non-small-cell lung cancer to gefitinib. N Engl J Med. 2005;352(8):786-792. growth factor receptor and in KRAS are predictive and prognostic indicators 55. Kosaka T, Yatabe Y, Endoh H, et al. Analysis of epidermal growth factor in patients with non-small-cell lung cancer treated with chemotherapy alone receptor gene mutation in patients with non-small cell lung cancer and acquired

198 Cancer Control April 2015, Vol. 22, No. 2 resistance to gefitinib. Clin Cancer Res. 2006;12(19):5764-5769. 56. Pao W, Wang TY, Riely GJ, et al. KRAS mutations and primary resistance of lung adenocarcinomas to gefitinib or erlotinib. PLoS Med. 2005;2(1):e17. 57. Vikis H, Sato M, James M, et al. EGFR-T790M is a rare lung cancer susceptibility allele with enhanced kinase activity. Cancer Res. 2007;67(10):4665-4670. 58. Balak MN, Gong Y, Riely GJ, et al. Novel D761Y and common secondary T790M mutations in epidermal growth factor receptor-mutant lung adenocarcinomas with acquired resistance to kinase inhibitors. Clin Cancer Res. 2006;12(21):6494-6501. 59. Bonomi PD, Buckingham L, Coon J. Selecting patients for treatment with epidermal growth factor tyrosine kinase inhibitors. Clin Cancer Res. 2007;13(15 pt 2):s4606-4612. 60. Engelman JA, Mukohara T, Zejnullahu K, et al. Allelic dilution obscures detection of a biologically significant resistance mutation in EGFR-amplified lung cancer. J Clin Invest. 2006;116(10):2695-2706. 61. Sequist LV, Waltman BA, Dias-Santagata D, et al. Genotypic and histological evolution of lung cancers acquiring resistance to EGFR inhibitors. Sci Transl Med. 2011;3(75):75ra26. 62. Su KY, Chen HY, Li KC, et al. Pretreatment epidermal growth factor receptor (EGFR) T790M mutation predicts shorter EGFR tyrosine kinase inhibitor response duration in patients with non-small-cell lung cancer. J Clin Oncol. 2012;30(4):433-440. 63. Soh J, Okumura N, Lockwood WW, et al. Oncogene mutations, copy number gains and mutant allele specific imbalance (MASI) frequently occur together in tumor cells. PLoS One. 2009;4(10):e7464. 64. Zhuang HQ, Wang J, Yuan ZY, et al. The drug-resistance to gefitinib in PTEN low expression cancer cells is reversed by irradiation in vitro. J Exp Clin Cancer Res. 2009;28:123. 65. Bianco R, Shin I, Ritter CA, et al. Loss of PTEN/MMAC1/TEP in EGF receptor-expressing tumor cells counteracts the antitumor action of EGFR tyrosine kinase inhibitors. Oncogene. 2003;22(18):2812-2822. 66. Rho JK, Choi YJ, Lee JK, et al. The role of MET activation in determining the sensitivity to epidermal growth factor receptor tyrosine kinase inhibitors. Mol Cancer Res. 2009;7(10):1736-1743. 67. Holm B, Mellemgaard A, Skov T, et al. Different impact of excision repair cross-complementation group 1 on survival in male and female patients with inoperable non-small-cell lung cancer treated with carboplatin and gemcitabine. J Clin Oncol. 2009;27(26):4254-4259. 68. Bao R, Lai CJ, Wang DG, et al. Targeting heat shock protein 90 with CUDC-305 overcomes erlotinib resistance in non-small cell lung cancer. Mol Cancer Ther. 2009;8(12):3296-3306. 69. Chung BM, Dimri M, George M, et al. The role of cooperativity with Src in oncogenic transformation mediated by non-small cell lung cancer-associated EGF receptor mutants. Oncogene. 2009;28(16):1821-1832. 70. Faber AC, Li D, Song Y, et al. Differential induction of apoptosis in HER2 and EGFR addicted cancers following PI3K inhibition. Proc Natl Acad Sci U S A. 2009;106(46):19503-19508. 71. Scagliotti GV, Parikh P, von Pawel J, et al. Phase III study comparing cisplatin plus gemcitabine with cisplatin plus pemetrexed in chemotherapy-naive patients with advanced-stage non-small-cell lung cancer. J Clin Oncol. 2008;26(21):3543-3551. 72. Li D, Ambrogio L, Shimamura T, et al. BIBW2992, an irreversible EGFR /HER2 inhibitor highly effective in preclinical lung cancer models. Oncogene. 2008;27(34):4702-4711. 73. Cross DA, Ashton SE, Ghiorghiu S, et al. AZD9291, an irreversible EGFR TKI, overcomes T790M-mediated resistance to EGFR inhibitors in lung cancer. Cancer Discov. 2014;4(9):1046-1061. 74. Luo YH, Chen YM. Recent advances in the development of mutant-selective EGFR inhibitors for non-small cell lung cancer patients with EGFR-TKI resistance. Transl Lung Cancer Res. 2014;3(6):368-369. 75. Jänne PA, Ramalingam SS, Yang CH, et al. Clinical activity of the mutant-selective EGFR inhibitor AZD9291 in patients (pts) with EGFR inhibitor-resistant non-small cell lung cancer (NSCLC). J Clin Oncol. 2014;32(5s):8009. 76. Sequist LV, Soria J-C, Gadgeel SM, et al. First-in-human evaluation of CO-1686, an irreversible, highly selective tyrosine kinase inhibitor of mutations of EGFR (activating and T790M). J Clin Oncol. 2014;32:8010. 77. Travis WD, Brambilla E, Noguchi M, et al. International Association for the Study of Lung Cancer/American Thoracic Society/European Respiratory Society International multidisciplinary classification of lung adenocarcinoma. J Thorac Oncol. 2011;6(2):244-285. 78. Solis LM, Behrens C, Raso MG, et al. Histologic patterns and molecular characteristics of lung adenocarcinoma associated with clinical outcome. Cancer. 2012;118(11):2889-2899. 79. Villa C, Cagle PT, Johnson M, et al. Correlation of EGFR mutation status with predominant histologic subtype of adenocarcinoma according to the new lung adenocarcinoma classification of the International Association for the Study of Lung Cancer/American Thoracic Society/European Respiratory Society. Arch Pathol Lab Med. 2014;138(10):1353-1357.

April 2015, Vol. 22, No. 2 Cancer Control 199 This article reviews the molecular

classification of diffuse gliomas

in the adult population.

Flight Deck Circa 1969. Photograph courtesy of Craig Damlo. www.soapboxrocket.com.

Impending Impact of Molecular Pathology on Classifying Adult Diffuse Gliomas Robert J. Macaulay, MD

Background: Progress in molecular oncology during the last decade has enabled investigators to more precisely define and group gliomas. The impacts of isocitrate dehydrogenase (IDH) mutation (mut) status and other molecular markers on the classification, prognostication, and management of diffuse gliomas are likely to be far-reaching. Methods: Clinical experience and the medical literature were used to assess the current status of glioma categorization and the likely impact of the pending revision of the classification scheme of the World Health Organization (WHO). Results: IDH-mut is a defining event in most adult fibrillary astrocytomas (FAs) and nearly all oligodendrogliomas (ODs). The IDH-mut status of most gliomas can be established by immunohistochemistry for the most common mutant of IDH1 (R132H). IDH wild-type (wt) diffuse gliomas include several familiar entities — in particular, glioblastoma (GBM) and most pediatric gliomas — as well as an assortment of less well-defined entities. The codeletion of 1p/19q distinguishes OD from FA, which, by contrast, shows frequent loss of the α thalassemia/mental retardation syndrome X-linked protein. Mixed oligoastrocytomas are typically classifiable as either OD or FA using molecular testing. Conclusions: The current practice of designating IDH-mut WHO grade 4 astrocytoma as secondary GBM will likely be discouraged, and primary or de novo GBM, which is always IDH-wt, may lose this qualification. Histologically, low- or intermediate-grade IDH-wt gliomas with molecular changes characteristic of GBM might justify the designation of GBM WHO grade 3. Mixed oligoastrocytoma is losing popularity as a diagnostic term because most cases will fall into either the FA or OD category. Distinguishing IDH-mut from IDH-wt tumors in clinical trials is likely to clarify sensitivity rates or tumor resistance among subgroups, thus suggesting opportunities for targeted therapy.

Introduction eries in molecular biology during the last decade. Insight into the pathobiology of primary brain tumors Important genetic, epigenetic, and proteomic alter- has accompanied technological and analytic discov- ations have been documented in a variety of low- and high-grade gliomas with clinical implications. Gliomas, which constitute the majority of tumors derived from From the Department of Anatomic Pathology, H. Lee Moffitt Cancer 1 Center & Research Institute, Tampa, Florida. intra-axial elements, may arise from undifferentiated Submitted February 2, 2015; accepted March 11, 2015. or partially differentiated astrocytes, oligodendrocytes, 2 Address correspondence to Robert J. Macaulay, MD, Department of or ependymal cells. Although histology remains a Anatomic Pathology, Moffitt Cancer Center, 12902 Magnolia Drive, mainstay of diagnosis, overlapping morphological Tampa, FL 33612. E-mail: [email protected] features can often be resolved without subjectivity; No significant relationship exists between the author and the companies/organizations whose products or services may be referenced for example, mixed oligoastrocytoma (MOA) vs ei- in this article. ther fibrillary astrocytoma (FA) or oligodendroglioma

200 Cancer Control April 2015, Vol. 22, No. 2 (OD), small cell glioblastoma (GBM) vs anaplastic OD, linked to tumorigenesis.21 Therefore, 2HG can be des- and ependymoma (EP) vs FA. ignated as an oncometabolite. Because IDH alterations Many gliomas exhibit characteristic genetic, chro- are fundamental to pathobiology, disease progres- mosomal, and/or biochemical signatures indicative of sion, diagnosis, response to adjuvant treatment, and distinct pathophysiology, presaging a reproducible clinical outcome, the classification of gliomas should and reliable molecular overlay to histopathological first acknowledge IDH-mut vs IDH-wt status. This diagnosis.3 Pertinent ancillary studies beyond histo- approach would solidify the analysis of subtypes of pathology may include immunohistochemical testing, IDH-mut tumors,22 allow more objective classification of chromosomal assessment, traditional and array-based IDH-wt entities, and highlight the molecular distinc- genetics,4 epigenetic analysis, and ribonucleic acid tions between pediatric and adult tumors exhibiting (RNA) profiling (both messenger RNA [mRNA]5 and similar morphological features.23,24 noncoding micro-RNA [miRNA]6-8). The World Health Organization (WHO) classification of central nervous Common Morphological Entities system tumors will likely incorporate these findings The 3 morphological entities that exhibit IDH-mut are in its soon-to-be released update.9 Such an improved FA (grades 2–4), OD (grades 2 and 3), and MOA (grade classification system may contribute to the improved 2 or 3).13 All other brain tumors are IDH-wt, includ- clinical investigation of new and targeted therapies ing most pediatric cases, pilocytic astrocytoma of any to combat the continued guarded outlook for most age, pleomorphic xanthoastrocytoma, subependymal primary brain neoplasms.1 Of particular interest are giant cell astrocytoma, EP, medulloblastoma, and me- the implications of isocitrate dehydrogenase (IDH) ningioma.25 IDH-wt diffuse astrocytomas, including mutation (mut) status. de novo GBM, gliomatosis cerebri,26 thalamic,27 and brainstem astrocytoma, harbor a variety of genetic Two Groups of Gliomas changes, and they should be regarded as distinct from The traditional categorization of gliomas into their their IDH-mut counterparts despite being morpholog- morphological subgroups (astrocytic, oligodendrog- ically indistinguishable. Conversely, cases of OD are lial, and ependymal) typically precedes textbook dis- almost always IDH-mut.22 Rare IDH-wt exceptions,28 cussions of particular entities. This is likely to change. including most pediatric cases,23 should be segregated In a seminal, genome-wide study, Parsons et al10 docu- within this diagnostic category until their pathogene- mented IDH1 mutations in subsets of glial neoplasms. sis is resolved. Morphologically MOAs are frequently It was soon recognized that mutations of IDH1 are IDH-mut and subsequently classifiable as either FA frequent in FAs, ODs, and MOAs,11,12 particularly at or OD following further molecular testing.29 The ap- amino acid residue 132 (arginine), which is replaced propriate pathological diagnosis for the rare IDH-wt by histidine in nearly 90% of tumors bearing an IDH1 MOA remains unsettled; however, such tumors should mutation.13 This residue may also be replaced by cys- still be segregated from the IDH-mut categories. teine,14 serine, or other amino acids,15 and mutations may rarely occur at other loci. IDH2 may also be Immunohistochemistry affected, albeit at a different locus (residue 172). Mu- Once assigned to the diffuse glioma category by rou- tations of other IDH isoforms have not been reported tine pathological assessment, testing for the IDH1 to accompany gliomagenesis. Thus, the finding of an R132H mutation may be easily conducted using a IDH1 or IDH2 mutation can be termed IDH-mut, and commercially available antibody test that recognizes the absence of a mutation in either gene is designat- the altered epitope.30 A positive result has high sen- ed as IDH wild type (wt). Acute myeloid leukemia, sitivity and specificity rates for IDH-mut R132H,22,31 chondrosarcoma,16 and giant cell tumor of the bone17 although reports exist of cross-reactivity with R132L may also exhibit IDH mutations. and R132M mutations.15,31 A total of 10% to 15% of IDH-mut cases are immunonegative, and such cases Accumulation of 2-Hydroxyglutarate can be sent for direct sequencing or hotspot analysis Early events in developing neoplasms involve the for both IDH1 and IDH2; IDH1 R132C is the most acquisition of aberrations in proliferation-related in- common immunonegative mutation.19 tracellular signaling pathways.18 IDH normally participates in the oxidative decarboxylation of isoci- IDH1/2 Mutations Account for Most Fibrillary trate to α ketoglutarate. The mutant form lacks this Astrocytomas important enzymatic property and instead generates A patient with low-grade astrocytoma typically pres- an abundance of 2-hydroxyglutarate (2HG).19 Excess ents with new-onset seizures, headaches, focal neu- 2HG interferes with chromatin-modifying activity, thus rological deficits, or all of these symptoms. Neuro- leading to global DNA hypermethylation and the gli- imaging typically shows a hemispheric hypointense, oma-CpG island methylation phenotype,20 which is nonenhancing mass that — if followed without sur-

April 2015, Vol. 22, No. 2 Cancer Control 201 gical intervention — may show minimal growth in the early stages. If magnetic resonance spectroscopy reveals a peak attributable to the accumulation of 2HG, then that result strongly predicts IDH-mut and is correlated with improved outcomes.32 His- topathology shows an astrocytic proliferation ex- A B hibiting coarse fibrillary processes and sufficient nuclear irregularity and hyperchromatism to justify designation as a neoplasm (Fig A–C). Approximately 70% of these tumors will exhibit IDH-mut (Fig D) as well as other changes.11 The gradual progression to high-grade astrocytoma is typical, although the time to progression is unpredicatable.12 When WHO grade 4 features are documented, the designation of C D secondary glioblastoma has found favor. (Whether this term is appropriate, even at first diagnosis,33 is discussed below.) Cases with features reminiscent of oligodendroglioma (small round monomorphic cells with perinuclear haloes) may prompt a preliminary diagnosis of MOA or the more generic diffuse glioma (not otherwise specified) until further testing E F 29 resolves the diagnosis. Fig A–F. — Representative images from a left parasagittal-enhancing mass in a man 37 years of age. The tumor was moderately cellular with (A) perinuclear TP53 Mutations Common in IDH-mut halos (× 20), (B) mitoses (× 40), vascular proliferation, and necrosis (not shown). (C) Neoplastic cells showed perinuclear and process positivity for Astrocytomas GFAP (× 20). The Ki67 index was 2% (not shown). (D) IDH1-mutant R132H Mutations in TP53 have been recognized in astrocyto- immunohistochemistry was strongly positive, and the mutation was con- ma, ranging from 30% to 50% of studied cases,34 and firmed with molecular testing (× 20). (E) Immunostaining for p53 protein showed overexpression in a significant minority of neoplastic cells (× 20). alterations in related genes are frequently detected (F) There was the conspicuous absence of ATRX expression in the neoplastic in all grades of diffuse astrocytomas.35 Interactions cells. Immunopositive, small background cells are likely normal oligodendro- between a p53 alteration and IDH-mut in the earliest cytes or lymphocytes (× 20). Codeletion of 1p/19q was not detected; MGMT promoter methylation was demonstrated. The diagnosis was astrocytoma stages of tumor initiation have been suggested by World Health Organization grade 4, isocitrate dehydrogenase mutation (also the predilection for the uncommon R132C mutation known as secondary glioblastoma). to occur in Li–Fraumeni syndrome (germline mutation of p53 with tumor predisposition).14 Nuclear p53 WHO Grade 4 IDH-mut Fibrillary overexpression is common in astrocytoma, is readily Astrocytoma detected with immunohistochemistry (Fig E), and of- WHO grade 4 FA with an IDH-mut accounts for fewer ten correlates with TP53 mutation; however, as a di- than 10% of WHO grade 4 gliomas,33 typically prompt- agnostic36 or prognostic37 marker, its utility is limited, ing the designation of secondary GBM11,22,33,42,43; and it has been supplanted by IDH, α thalassemia/ however, the weight of evidence, both molecular mental retardation syndrome X-linked (ATRX), and and clinical, suggests that this term is outdated. It 1p/19q testing. is likely that IDH-mut WHO grade 4 astrocytoma arises from a different precursor pool than IDH-wt ATRX Loss Characterizes IDH-mut GBM.44 IDH-mut gliomas follow a distinct molecular Astrocytomas trajectory from IDH-wt tumors.33 The ambiguity of Mutations in the telomere maintenance protein the term “primary GBM” has been emphasized by ATRX appear to distinguish IDH-mut FA from tu- the occasional de novo presentation of an IDH-mut mors that lack this change — in particular, OD.38 WHO grade 4 astrocytoma or conversely by IDH-wt Absence of ATRX protein (Fig F) by immunohis- low-grade glioma progressing quickly to GBM.45 In tochemistry is evident in 25% to 30% of adult and addition, compared with those with IDH-wt GBM, pediatric high- and low-grade astrocytomas.39 Loss patients with IDH-mut WHO grade 4 FA are younger of ATRX expression and IDH-mut typifies more and have longer survival times.33,45 Therefore, it is than 95% of adult diffuse astrocytomas, and it is clear that IDH1-mut FAs — whatever the grade — almost mutually exclusive with 1p/19q codeletion.40 and IDH1-wt gliomas are different diseases.18 Be- Classification based on IDH-mut status and ATRX cause GBM is a familiar term for pathologists and expression will likely be a key recommendation in clinicians, it will likely be retained, but this term may the next WHO classification update.40,41 be restricted to morphologically appropriate gliomas

202 Cancer Control April 2015, Vol. 22, No. 2 that lack IDH-mut. It will be interesting to determine therapy are improved. If IDH status is unknown, then whether the classic histopathological WHO grading array-based genetic testing for IDH1 and IDH2 can be system developed for diffuse astrocytoma when IDH considered in patients for whom an increased likeli- status was not being tested46 will require modifica- hood of a positive result exists,4 particularly for those tion for IDH-mut astrocytomas. tumors that exhibit features such as oligodendroglial morphology or loss of ATRX expression. Characterizing Oligodendroglioma The classic OD appearance of small round cells Complex Molecular Signature of IDH-wt GBM arranged in honeycomb-like nests with “chicken Among the myriad molecular alterations in GBM, wire”-like coarse, branching vasculature, microcalci- EGFR mutations, deletions, and protein overexpres- fications, and perinuclear haloes has withstood the sion are common. Clinical trials targeting the epi- scrutiny of advancing technologies. 1p/19q codele- dermal growth factor receptor vIII mutant form,55 in tion47 is prognostic,12 predictive,48,49 and diagnostic.50 which exons 2 to 7 are lost, have shown promise, but In addition, more than 90% of 1p-/19q- ODs are also detecting the mutation, either by immunohistochem- IDH-mut.11 Among IDH-mut gliomas, mutations in the istry, reverse transcription–polymerase chain reaction, promoter of TERT are confined to OD24; conversely, or multiplex ligation-dependent probe amplification, OD lacks ATRX mutations.38 For practical purposes, is of dubious significance in multivariate models.56 grading of OD is restricted to grades 2 and 3 because WT1 is variably expressed in gliomas, and although it patient outcomes are considerably better for OD than correlates with worse outcome, its effect is overshad- for IDH-wt gliomas.51 Surgical cases with ambiguous owed by age and IDH-mut status.57 The significance features can be initially designated as MOA or simply of the loss of PTEN is unsettled in GBM,58 but it may as diffuse glioma. Support is growing for the notion be a poor prognostic indicator in diffuse gliomas, par- that most MOAs can be placed into either the FA or ticularly in tumors lacking 1p/19q codeletion.59 TERT OD category using IDH status, ATRX expression, and promoter mutations are detectable in about one-half 1p/19q codeletion testing, and that a genuine mixture of IDH-wt GBMs and only those lacking ATRX mu- of the 2 is a rare occurrence.52 tation; TERT promoter mutations in IDH-mut tumors are restricted to OD.24 Other changes, such as p16, Diffuse IDH-wt Gliomas in Adults NF, Rb, or CDKN2A, have yet to achieve diagnostic About 30% of WHO low-grade adult diffuse astro- or therapeutic significance. Therefore, the molecu- cytomas and fewer than 10% of morphologically di- lar pathogenesis of sporadic GBM in adults remains agnosed OD are IDH-wt11; by contrast, the majority uncertain. Sophisticated expression profiling, which of newly encountered GBM is IDH-wt.10 Among the allows grouping into proneural, neural, mesenchymal, IDH-wt low-grade astrocytoma group, some of these and classical phenotypes, has yet to reach the point cases harbor chromosomal and molecular alterations, of diagnostic utility, possibly due to considerable in- thus aligning them with de novo GBM.3 The remaining tratumoral heterogeneity.60 cases possess a variety of genetic and biochemical alterations, some of which are more akin to pediatric Intraoperative Diagnosis: High-Grade Glioma gliomas, particularly BRAF alterations and histone A pathologist handling an aggressive neoplasm with modifications,27,53 facilitating further subclassification. ring enhancement and central necrosis may provide an intraoperative diagnosis of GBM if morpholog- IDH-wt Glioblastoma ically appropriate. However, IDH-mut FA may un- The most common primary brain tumor, GBM, is also dergo anaplastic transformation following disease the most lethal.1 Histologically, GBM is populated by progression, or it may have histological features sim- anaplastic astrocytes with mitotic activity, glomeru- ilar to IDH-wt GBM at presentation; alternatively, OD loid vascular proliferation, and geographic necrosis may acquire mitotic activity, vascular proliferation, with (or without) peripheral pseudopalisading. The and necrosis and yet still follow a clinical course molecular landscape of GBM is complex and likely more consistent with WHO grade 3. IDH-mut tu- accommodates several distinct entities,54 as well as mors maintain their molecular signatures over time considerable intratumoral heterogeneity5 and overlap. and subsequent resections, and they have improved Most GBMs arise rapidly and are clinically designated outcomes relative to IDH-wt GBM with similar grad- as being primary or de novo,42 whereas individuals ing features. Because these cases are morpholog- with a lower-grade IDH-mut astrocytoma may prog- ically indistinguishable on intraoperative smears ress to grade 4 (secondary GBM).33,51 Future classifica- and frozen sections, an intraoperative diagnosis of tion may eliminate the “secondary” GBM designation high-grade glioma for practical purposes would en- because progression-free and overall survival rates as sure the appropriate handling of diagnostic material. well as responses to adjuvant chemotherapy or radio- The final diagnosis could range from FA grades 3 or

April 2015, Vol. 22, No. 2 Cancer Control 203 4, OD grade 3, EP grade 3, GBM, or primitive neuro- dichotomy from the outset of study design risks a ectodermal tumor/medulloblastoma (depending on misinterpretation of results; for example, a compar- location). It remains to be determined whether the ison of micro-ribonucleic acid (RNA) in low- com- preoperative demonstration of a 2-hydroxyglutarate pared with high-grade gliomas may be better inter- peak on magnetic resonance spectroscopy is suffi- preted as comparing IDH-mut vs IDH-wt tumors.3,8 cient evidence of IDH-mut or conversely whether its By contrast, a trial of the intratumoral heterogene- absence is sufficient to imply IDH-wt.32 ity of messenger RNA expression identified IDH-wt status as a fundamental study inclusion criterion.5 Glioblastoma WHO Grade 3 Glioblastoma remains an essential diagnostic cat- Considerable interest exists in the biology and clinical egory and will likely be subdivided on a molecular behavior of low- or intermediate-grade diffuse astro- basis, but, ideally, it would exclude IDH-mut tumors. cytic IDH-wt neoplasms in adults. Some examples that The diagnostic term secondary glioblastoma might be share molecular features with GBM can be considered replaced with astrocytoma World Health Organization as being “missed” GBM due to undersampling. Other (WHO) grade 4 and restricted to IDH-mut tumors. patients may present during the early phase of tumor The diagnostic term de novo glioblastoma could then progression and before florid mitotic activity, vascular be shortened to glioblastoma WHO grade 4 and re- proliferation, and necrosis have developed; biopsy stricted to IDH-wt tumors. Histologically lower-grade findings may indicate FA WHO grade 2 or 3.40 The neoplasms of evidently similar biology may qualify tumors with molecular findings of GBM will more for the designation of glioblastoma WHO grade 3 if rapidly advance than their IDH-mut counterparts, al- they possess the molecular signature of glioblastoma though their overall outcome is still better than GBM.51 but lack vascular proliferation and necrosis. In the future, such examples may be categorized as GBM WHO grade 3, and the designation of FA WHO References grade 3 might be reserved for IDH-mut cases. 1. Ostrom QT, Gittleman H, Liao P, et al. CBTRUS statistical report: primary brain and central nervous system tumors diagnosed in the United States in 2007-2011. Neuro Oncol. 2014;16(suppl 4):iv1-63. IDH-mut Is Uncommon in Pediatric Diffuse 2. Taylor MD, Poppleton H, Fuller C, et al. Radial glia cells are candidate stem cells of ependymoma. Cancer Cell. 2005;8(4):323-335. Glioma 3. Appin CL, Brat DJ. Molecular pathways in gliomagenesis and their IDH-mut glial neoplasms do not manifest before ad- relevance to neuropathologic diagnosis. Adv Anat Pathol. 2015;22(1):50-58. 61 4. Idbaih A, Crinière E, Ligon KL, et al. Array-based genomics in glioma olescence ; conceptually, this can be attributed to a research. Brain Pathol. 2010;20(1):28-38. slow accumulation of 2HG and a delayed development 5. Patel AP, Tirosh I, Trombetta JJ, et al. Single-cell RNA-seq highlights intratumoral heterogeneity in primary glioblastoma. Science. of the glioma-CpG island methylation phenotype. Mo- 2014;344(6190):1396-1401. lecular alterations predominantly observed in children 6. Haapa-Paananen S, Chen P, Hellström K, et al. Functional profiling of precursor MicroRNAs identifies MicroRNAs essential for glioma proliferation. and young adults correlate with patient age and tu- PLoS One. 2013;8(4):e60930. mor location.27,62 Recent studies have emphasized the 7. Li S, Chowdhury R, Liu F, et al. Tumor-suppressive miR148a is silenced by CpG island hypermethylation in IDH1-mutant gliomas. Clin Cancer Res. prognostic significance of detecting the mutated form 2014;20(22):5808-5822. of histone 3 and will likely lead to its routine testing 8. Barbano R, Palumbo O, Pasculli B, et al. A miRNA signature for 63,64 defining aggressive phenotype and prognosis in gliomas. PLoS One. in midline pediatric high-grade astrocytoma. By 2014;9(10):e108950. contrast to the tumors seen in adults, pediatric OD 9. Louis DN, Perry A, Burger P, et al. International society of neuropathology-haarlem consensus guidelines for nervous system tumor classification is rarely accompanied by IDH-mut or 1p/19q codele- and grading. Brain Pathol. 2014;24(5):429-435. tion.23,65 This indicates at least 2 forms of OD: the 10. Parsons DW, Jones S, Zhang X, et al. An integrated genomic analysis of human glioblastoma multiforme. Science. 2008;321(5897):1807-1812. adult IDH-mut and pediatric IDH-wt types. Whether 11. Yan H, Parsons DW, Jin G, et al. IDH1 and IDH2 mutations in gliomas. rare IDH-mut pediatric diffuse gliomas represent an N Engl J Med. 2009;360(8):765-773. 12. Houillier C, Wang X, Kaloshi G, et al. IDH1 or IDH2 mutations predict early occurrence of their adult counterparts has not longer survival and response to temozolomide in low-grade gliomas. Neurology. yet been established.61,66 2010;75(17):1560-1566. 13. Sanson M, Marie Y, Paris S, et al. Isocitrate dehydrogenase 1 codon 132 mutation is an important prognostic biomarker in gliomas. J Clin Oncol. Conclusions 2009;27(25):4150-4154. 14. Watanabe T, Vital A, Nobusawa S, et al. Selective acquisition of IDH1 Isocitrate dehydrogenase (IDH) mutation (mut) is R132C mutations in astrocytomas associated with Li-Fraumeni syndrome. an early and likely initiating event in the develop- Acta Neuropathol. 2009;117(6):653-656. 15. Ogasawara S, Kaneko MK, Tsujimoto Y, et al. Multi-specific monoclo- ment of many gliomas. This provides an opportuni- nal antibody MsMab-2 recognizes IDH1-R132L and IDH2-R172M mutations. ty to more precisely classify the 2 IDH-mut tumor Monoclon Antib Immunodiagn Immunother. 2013;32(6):377-381. 16. Amary MF, Bacsi K, Maggiani F, et al. IDH1 and IDH2 mutations are types, fibrillary astrocytoma and oligodendroglioma. frequent events in central chondrosarcoma and central and periosteal chon- Distinguishing IDH-mut from IDH wild-type (wt) dromas but not in other mesenchymal tumours. J Pathol. 2011;224(3):334-343. 17. Kato Kaneko M, Liu X, Oki H, et al. Isocitrate dehydrogenase mutation is tumors may improve our understanding of tumor frequently observed in giant cell tumor of bone. Cancer Sci. 2014;105(6):744-748. biology, prognostication, patient stratification in clin- 18. Weller M, Wick W, von Deimling A. Isocitrate dehydrogenase mutations: a challenge to traditional views on the genesis and malignant progression of ical trials, and the selection of potential therapeu- gliomas. Glia. 2011;59(8):1200-1204. tic targets.67 Failure to appreciate this fundamental 19. Pusch S, Schweizer L, Beck AC, et al. D-2-Hydroxyglutarate producing

204 Cancer Control April 2015, Vol. 22, No. 2 neo-enzymatic activity inversely correlates with frequency of the type of of oligodendroglial tumors shows preferential allelic deletions on 19q and 1p. isocitrate dehydrogenase 1 mutations found in glioma. Acta Neuropathol Am J Pathol. 1994;145(5):1175-1190. Commun. 2014;2:19. 48. Cairncross JG, Ueki K, Zlatescu MC, et al. Specific genetic predictors 20. Noushmehr H, Weisenberger DJ, Diefes K, et al. Identification of a of chemotherapeutic response and survival in patients with anaplastic oligo- CpG island methylator phenotype that defines a distinct subgroup of glioma. dendrogliomas. J Natl Cancer Inst. 1998;90(19):1473-1479. Cancer Cell. 2010;17(5):510-522. 49. Cairncross G, Wang M, Shaw E, et al. Phase III trial of chemoradio- 21. Rakheja D, Medeiros LJ, Bevan S, et al. The emerging role of d-2-hy- therapy for anaplastic oligodendroglioma: long-term results of RTOG 9402. droxyglutarate as an oncometabolite in hematolymphoid and central nervous J Clin Oncol. 2013;31(3):337-343. system neoplasms. Front Oncol. 2013;3:169. 50. Jiang H, Zhang Z, Ren X, et al. 1p/19q-driven prognostic molecu- 22. Reuss DE, Sahm F, Schrimpf D, et al. ATRX and IDH1-R132H immu- lar classification for high-grade oligodendroglial tumors. J Neurooncol. nohistochemistry with subsequent copy number analysis and IDH sequenc- 2014;120(3):607-614. ing as a basis for an “integrated” diagnostic approach for adult astrocytoma, 51. Agnihotri S, Aldape KD, Zadeh G. Isocitrate dehydrogenase status oligodendroglioma and glioblastoma. Acta Neuropathol. 2015;129(1):133-146. and molecular subclasses of glioma and glioblastoma. Neurosurg Focus. 23. Suri V, Jha P, Agarwal S, et al. Molecular profile of oligodendrogliomas 2014;37(6):E13. in young patients. Neuro Oncol. 2011;13(10):1099-1106. 52. Huse JT, Diamond EL, Wang L, et al. Mixed glioma with molecular 24. Koelsche C, Sahm F, Capper D, et al. Distribution of TERT promoter features of composite oligodendroglioma and astrocytoma: a true “oligoas- mutations in pediatric and adult tumors of the nervous system. Acta Neuro- trocytoma”? Acta Neuropathol. 2015;129(1):151-153. pathol. 2013;126(6):907-915. 53. Sturm D, Witt H, Hovestadt V, et al. Hotspot mutations in H3F3A and 25. Balss J, Meyer J, Mueller W, et al. Analysis of the IDH1 codon 132 IDH1 define distinct epigenetic and biological subgroups of glioblastoma. mutation in brain tumors. Acta Neuropathol. 2008;116(6):597-602. Cancer cell. 2012;22(4):425-437. 26. Desestret V, Ciccarino P, Ducray F, et al. Prognostic stratification 54. Mendez G, Ozpinar A, Raskin J, et al. Case comparison and literature of gliomatosis cerebri by IDH1 R132H and INA expression. J Neurooncol. review of glioblastoma: a tale of two tumors. Surg Neurol Int. 2014;5:121. 2011;105(2):219-224. 55. Mellinghoff IK, Wang MY, Vivanco I, et al. Molecular determinants 27. Aihara K, Mukasa A, Gotoh K, et al. H3F3A K27M mutations in thalamic of the response of glioblastomas to EGFR kinase inhibitors. N Engl J Med. gliomas from young adult patients. Neuro Oncol. 2014;16(1):140-146. 2005;353(19):2012-2024. 28. Hewer E, Beck J, Murek M, et al. Polymorphous oligodendroglioma 56. Weller M, Kaulich K, Hentschel B, et al. Assessment and prognostic of Zulch revisited: a genetically heterogeneous group of anaplastic gliomas significance of the epidermal growth factor receptor vIII mutation in glioblas- including tumors of bona fide oligodendroglial differentiation. Neuropathology. toma patients treated with concurrent and adjuvant temozolomide radioche- 2014;34(4):323-332. motherapy. Int J Cancer. 2014;134(10):2437-2447. 29. Sahm F, Reuss D, Koelsche C, et al. Farewell to oligoastrocytoma: in 57. Rauscher J, Beschorner R, Gierke M, et al. WT1 expression increases situ molecular genetics favor classification as either oligodendroglioma or with malignancy and indicates unfavourable outcome in astrocytoma. J Clin astrocytoma. Acta Neuropathol. 2014;128(4):551-559. Pathol. 2014;67(7):556-561. 30. Capper D, Weissert S, Balss J, et al. Characterization of R132H 58. Carico C, Nuño M, Mukherjee D, et al. Loss of PTEN is not associated mutation-specific IDH1 antibody binding in brain tumors. Brain Pathol. with poor survival in newly diagnosed glioblastoma patients of the temozolo- 2010;20(1):245-254. mide era. PLoS One. 2012;7(3):e33684. 31. Agarwal S, Sharma MC, Jha P, et al. Comparative study of IDH1 muta- 59. Sabha N, Knobbe CB, Maganti M, et al. Analysis of IDH mutation, tions in gliomas by immunohistochemistry and DNA sequencing. Neuro Oncol. 1p/19q deletion, and PTEN loss delineates prognosis in clinical low-grade 2013;15(6):718-726. diffuse gliomas. Neuro Oncol. 2014;16(7):914-923. 32. Natsumeda M, Igarashi H, Nomura T, et al. Accumulation of 2-hydrox- 60. Verhaak RG, Hoadley KA, Purdom E, et al. Integrated genomic analysis yglutarate in gliomas correlates with survival: a study by 3.0-tesla magnetic identifies clinically relevant subtypes of glioblastoma characterized by abnor- resonance spectroscopy. Acta Neuropathol Commun. 2014;2(1):158. malities in PDGFRA, IDH1, EGFR, and NF1. Cancer Cell. 2010;17(1):98-110. 33. Nobusawa S, Watanabe T, Kleihues P, et al. IDH1 mutations as molec- 61. Pollack IF, Hamilton RL, Sobol RW, et al. IDH1 mutations are common ular signature and predictive factor of secondary glioblastomas. Clin Cancer in malignant gliomas arising in adolescents: a report from the Children’s On- Res. 2009;15(19):6002-6007. cology Group. Childs Nerv Syst. 2011;27(1):87-94. 34. Cavenee WK, Scrable HJ, James CD. Molecular genetics of human 62. Gielen GH, Gessi M, Buttarelli FR, et al. Genetic analysis of diffuse cancer predisposition and progression. Mutat Res. 1991;247(2):199-202. high-grade astrocytomas in infancy defines a novel molecular entity. Brain 35. Ichimura K, Bolin MB, Goike HM, et al. Deregulation of the p14ARF/ Pathol. 2014. [Epub ahead of print]. MDM2/p53 pathway is a prerequisite for human astrocytic gliomas with G1-S 63. Bechet D, Gielen GG, Korshunov A, et al. Specific detection of methi- transition control gene abnormalities. Cancer Res. 2000;60(2):417-424. onine 27 mutation in histone 3 variants (H3K27M) in fixed tissue from high- 36. Takami H, Yoshida A, Fukushima S, et al. Revisiting TP53 mutations grade astrocytomas. Acta Neuropathol. 2014;128(5):733-741. and immunohistochemistry: a comparative study in 157 diffuse gliomas. Brain 64. Venneti S, Santi M, Felicella MM, et al. A sensitive and specific histo- Pathol. 2014. [Epub ahead of print]. pathologic prognostic marker for H3F3A K27M mutant pediatric glioblastomas. 37. Jaros E, Perry RH, Adam L, et al. Prognostic implications of p53 protein, Acta Neuropathol. 2014;128(5):743-753. epidermal growth factor receptor, and Ki-67 labelling in brain tumours. Br J 65. Rodriguez FJ, Tihan T, Lin D, et al. Clinicopathologic features of Cancer. 1992;66(2):373-385. pediatric oligodendrogliomas: a series of 50 patients. Am J Surg Pathol. 38. Kannan K, Inagaki A, Silber J, et al. Whole-exome sequencing iden- 2014;38(8):1058-1070. tifies ATRX mutation as a key molecular determinant in lower-grade glioma. 66. Byeon SJ, Myung JK, Kim SH, et al. Distinct genetic alterations in Oncotarget. 2012;3(10):1194-1203. pediatric glioblastomas. Childs Nerv Syst. 2012;28(7):1025-1032. 39. Abedalthagafi M, Phillips JJ, Kim GE, et al. The alternative lengthening 67. Elhammali A, Ippolito JE, Collins L, et al. A high-throughput fluorimetric of telomere phenotype is significantly associated with loss of ATRX expression assay for 2-hydroxyglutarate identifies Zaprinast as a glutaminase inhibitor. in high-grade pediatric and adult astrocytomas: a multi-institutional study of Cancer Discov. 2014;4(7):828-839. 214 astrocytomas. Mod Pathol. 2013;26(11):1425-1432. 40. Reuss DE, Sahm F, Schrimpf D, et al. ATRX and IDH1-R132H immunohistochemistry with subsequent copy number analysis and IDH sequencing as a basis for an “integrated” diagnostic approach for adult astrocytoma, oligodendroglioma and glioblastoma. Acta Neuropathol. 2015;129(1):133-146. 41. Mur P, Mollejo M, Hernández-Iglesias T, et al. Molecular classification defines 4 prognostically distinct glioma groups irrespective of diagnosis and grade. J Neuropathol Exp Neurol. 2015;74(3):241-249. 42. Watanabe K, Tachibana O, Sata K, et al. Overexpression of the EGF receptor and p53 mutations are mutually exclusive in the evolution of primary and secondary glioblastomas. Brain Pathol. 1996;6(3):217-223. 43. von Deimling A, von Ammon K, Schoenfeld D, et al. Subsets of glioblastoma multiforme defined by molecular genetic analysis. Brain Pathol. 1993;3(1):19-26. 44. Gorovets D, Kannan K, Shen R, et al. IDH mutation and neuroglial developmental features define clinically distinct subclasses of lower grade diffuse astrocytic glioma. Clin Cancer Res.. 2012;18(9):2490-2501. 45. Popov S, Jury A, Laxton R, et al. IDH1-associated primary glioblastoma in young adults displays differential patterns of tumour and vascular morphology. PLoS One. 2013;8(2):e56328. 46. Daumas-Duport C, Scheithauer B, O’Fallon J, et al. Grading of astrocytomas. A simple and reproducible method. Cancer. 1988;62(10):2152-2165. 47. Reifenberger J, Reifenberger G, Liu L, et al. Molecular genetic analysis

April 2015, Vol. 22, No. 2 Cancer Control 205 The coexpression of VEGF-A and

CD31 may play a prominent role

in the neoangiogenesis of IMPC.

This Way to Gas. Photograph courtesy of Craig Damlo. www.soapboxrocket.com.

Overexpression of Vascular Endothelial Growth Factor A in Invasive Micropapillary Colorectal Carcinoma Marilin Rosa, MD, Maisoun Abdelbaqi, MD, Katherine M. Bui, Aejaz Nasir, MD, MPhil, Marilyn M. Bui, MD, PhD, David Shibata, MD, and Domenico Coppola, MD

Background: Invasive micropapillary carcinoma (IMPC) is a rare variant of colorectal cancer with an adverse prognosis. “Retraction artifact” around tumor cells is a feature of IMPC. The aim of this study was to assess the nature of the retractions around the tumor cells and to describe the histopathological features of a group of 18 cases of IMPC. Methods: A pathology review of 128 consecutive colorectal cancers identified 18 cases of histologically proven IMPC using 5% of the total tumor volume comprised of a micropapillary component as the diagnostic criterion. Immunostains for D2-40, CD31, CD34, vascular endothelial growth factor A (VEGF-A), and mucin 1 (MUC-1) were performed using the avidin-biotin complex method. Results: Cases of IMPC were characterized by pseudomicropapillae surrounded by lacunar-like clear spaces. These structures exhibited the inside-out growth pattern as highlighted by MUC-1 staining. The lining of the lacunar spaces was immunoreactive to CD31 but not CD34 or D2-40, indicating that they are neovascular structures. Furthermore, the tumor cells strongly and diffusely expressed VEGF-A. Conclusions: The strong coexpression of VEGF-A and CD31 suggests a prominent role of neoangiogenesis in these tumors.

Introduction From the Departments of Anatomic Pathology (MR, MA, MMB, DC) and Gastrointestinal Surgical Oncology (DS) and the Chemical Colorectal cancer (CRC) is the third most commonly Biology & Molecular Medicine Program (MMB, DC), H. Lee Moffitt diagnosed cancer and the third leading cause of cancer Cancer Center & Research Institute, the Departments of Pathology death in both men and women in the United States.1 (MR, MA, MMB, DC), Oncologic Sciences (MR, MMB, DS, DC), and Honors College (KMB), University of South Florida Morsani College Invasive micropapillary carcinoma (IMPC) has recently of Medicine, Tampa, Florida, and Eli Lilly and Company (AN), been described as a rare variant of CRC with a poor Indianapolis, Indiana. prognosis independently of stage.2-5 Although a focal mi- Submitted June 30, 2014; accepted October 14, 2014. cropapillary pattern can be seen in conventional cases of Address correspondence to Domenico Coppola, MD, Department of CRC, the diagnosis of the micropapillary variant of CRC Anatomic Pathology, Moffitt Cancer Center, 12902 Magnolia Drive, Tampa, FL 33612. E-mail: [email protected] requires that at least 5% of the total tumor volume has a 2,5 No significant relationships exist between the authors and micropapillary component. The proportion of the micro- the companies/organizations whose products or services may be papillary component has also been suggested to have an referenced in this article. important predictive factor for nodal metastasis, but this We thank Angela Reagan and Jacki Hattle for the technical assistance finding has not been consistent.4 during the preparation and submission of the manuscript as well as the Pathology Histology Laboratory at Moffitt Cancer Center for IMPC has been recognized and studied at other preparing the stains. anatomical locations such as the ovary, breast, urinary

206 Cancer Control April 2015, Vol. 22, No. 2 bladder, salivary glands, and lung.6-10 In its breast from formalin-fixed, paraffin-embedded representa- counterpart, IMPC shows a tendency to invade lym- tive tumor blocks with an autostainer using monoclo- phatic vessels and to spread to regional lymph nodes. nal antibodies against CD31, CD34, D2-40, vascular Whether the lacunar-like spaces surrounding the endothelial growth factor A (VEGF-A), and mucin 1 micropapillary structures are true vascular channels (MUC-1). All stains were run with appropriate controls or “retraction artifacts,” as is commonly considered, and stained in accordance with the manufacturer’s rec- remains a subject of debate.11,12 Some published stud- ommended protocol. Results were considered positive ies suggest that retraction clefts are likely the result when cytoplasmic staining, membranous staining, or of an altered tumor–stromal interaction.11,13 In addi- both were present in any percentage of the cells of tion, these retraction clefts may be potential spaces interest and considered negative when there was a that represent prelymphatic spaces involved in the complete lack of staining. facilitation of the initial lymphatic invasion and may undergo endothelialization under the influence of Results growth factors secreted by tumor cells.11 Our review identified a total of 18 (14% of consecu- The histopathological, immunohistochemical, tive colorectal carcinomas) bona fide cases of IMPC. and molecular features of colorectal IMPC have been The age of the patients ranged between 34 and 93 years previously defined.1-5 In this study, we review a single (mean, 52.3 years). Twelve patients were men and 6 pa- institution series of colorectal IMPC cases and char- tients were women. Eight tumors were located in the right acterized their vascular immunohistochemical pheno- colon (2 in the cecum), 4 in the left colon, 4 in the rectum, type to understand whether the presence of retraction 1 at the splenic flexure, and 1 in the terminal ileum. The clefts correlates with lymphangiogenesis and, thus, is tumor size ranged from 2.0 to 6.1 cm (mean, 4.0 cm). a factor facilitating metastatic tumor spread. Twelve tumors were diagnosed as moderately differentiated and 6 were poorly differentiated. The pathological stages Methods of the tumors were as follows: 4 carcinomas were stage 1, Participants 3 were stage 2A, 2 were stage 3B, 8 were stage 3C, and This study was carried out in accordance with a re- 1 was stage 4. Three patients received preoperative search protocol approved by the Institutional Review chemotherapy (Table). Board at the H. Lee Moffitt Cancer Center & Research Institute and the University of South Florida in Tampa, Florida. A total Table. — Pathological Features of 18 Cases of Invasive Micropapillary Carcinoma of 128 consecutive colon resections per- Location Size Grade % of No. of Morphology of LN formed at the Moffitt Cancer Center during (cm) Micropapillary LNs Metastasis a 5-year period from 2005 to 2010 were Component retrospectively selected from the anatom- Ascending colon 4.5 3 90 0/15 NA ical pathology files. Two pathologists (DC Ileocecal valve 4.3 2 10 1/11 Micropapillary and MA) reviewed the pathology reports Right colon 3.5 3 10 3/26 1/3 micropapillary and the hematoxylin and eosin–stained microscopic slides of each tumor to con- Sigmoid colon 5.0 2 80 14/20 9/14 micropapillary firm the diagnosis and to select a rep- Rectum 3.3 2 10 2/31 Micropapillary resentative block for immunohistochem- Rectum 6.0 2 20 7/16 Micropapillary ical studies. As in previously published Terminal ileum 3.5 2 30 13/34 Slides not available studies, tumors were classified as IPMC Rectum 3.2 3 10 0/23 NA when at least 5% of the tumor was com- Right colon 4.0 3 10 18/18 3/18 micropapillary prised of a micropapillary component.2,5 Rectum 2.6 2 10 0/24 NA In case of a discrepancy, the slides were reviewed again by both pathologists at Cecum 3.5 2 10 4/5 Micropapillary a multiheaded microscope to reach a Cecum 4.5 2 10 7/46 5/12 micropapillary diagnostic consensus. Collected clinico- Ileocecal valve 6.5 3 50 15/49 13/15 micropapillary pathological variables included tumor Sigmoid colon 4.0 2 30 0/18 NA size, grade, stage, histological type, and Sigmoid colon 2.5 2 60 3/13 No micropapillary the presence of nodal metastases, distant Splenic flexure 0.6 3 90 0/15 NA metastases, or both. Right colon 2.1 2 40 7/13 Micropapillary Immunohistochemistry Left colon 2.0 2 20 1/15 Micropapillary Immunohistochemical analysis was LN = lymph node, NA = not applicable. performed on 5-µm unstained sections

April 2015, Vol. 22, No. 2 Cancer Control 207 Histologically, all micropapillary areas exhibited A B the characteristically features of IMPC as previously described.2-5 Tumors were composed of clusters of cells growing in a micropapillary (or pseudomi- cropapillary) pattern surrounded by lacunar-like clear spaces lined by elongated endothelial-like cells (Fig 1A). Fibrovascular cores were not identified (Fig 1B). Tumor cells were columnar to polygonal and had a moderate amount of eosinophilic cytoplasm C D and conspicuous nucleoli. Characteristically, the micropapillae exhibited a rotation of cell polarization known as an “inside-out” growth pattern, or “reverse cell polarity,” which was also highlighted by MUC-1 immunostaining (Fig 1C). Collections of neutrophils were seen in some cases to infiltrate the micropapillae and occasionally to spill into the lacunar-like spaces Fig 1A–D. — (A) Histologically, invasive micropapillary carcinoma is characterized by micropapillae surrounded by lacunar-like clear spaces (Fig 1D). In addition, the micropapillary features rep- (H & E × 100). (B) The micropapillae are composed of columnar, resented at least 10% of the tumors in all of our cases. round, or oval cells with a moderate amount of eosinophilic cytoplasm, The transition between conventional adenocarcinoma vesicular nuclei with conspicuous nucleoli, and the absence of fibrovascular cores (H & E × 200). (C) The MUC-1 immunostain highlights the and IMPC was abrupt, and, in general, the IMPC com- “inside-out” growth pattern (immunohistochemistry × 200). (D) Numerous ponent was identified at the advancing, infiltrating neutrophils infiltrate the micropapillae and spill into the lacunar-like spaces edge of the tumors. (H & E × 200). H & E = hematoxylin and eosin, MUC-1 = mucin 1. Immunohistochemically, we observed a strong and diffuse expression of VEGF-A in the tumor cells A B within the lacunar-like spaces (Fig 2A). With regard to the vascular markers, CD31 — but not CD34 — positivity highlighted the cells lining the clear spaces surrounding the micropapillae, thus suggesting that they are immature vessels (Fig 2B and C). D2-40 also failed to stain this lining, suggesting that the peritumoral spaces are not lymphatic (Fig 2D). C D Discussion IMPC is an unusual and aggressive variant of colorectal adenocarcinoma commonly associated with lymphovascular invasion, lymph node metastasis, and poor clinical outcome.2-5 In our retrospective review of 128 cases of CRC, we identified 18 cases of IMPC (approximately 14%), similar to what was reported by Fig 2A–D. — (A) Immunostain showing the tumor cells are strongly and 4 diffusely positive for vascular endothelial growth factor A. (B) The CD31 Haupt et al (19% of 178 cases of CRC with IMPC-like decorates the cells lining the lacunar spaces, indicating the vascular nature features). Histologically, our cases also exhibited sim- of these structures. (C) The same cells fail to stain with CD34, indicating ilar features to those previously described — in par- that they are neovascular structures. (D) They also fail to stain with D2-40, ticular, the presence of micropapillae without a fibro- indicating that the spaces are not lymphatic (immunohistochemistry × 200). vascular core surrounded by a clear lacunar space.2-5 ic vessels is a common step in the spread of solid The “micropapillary” appearance, which was tumors, including CRC. In addition, angiogenesis is first described in tumors of the breast,10,11 has been a well-known, necessary step for tumor progression attributed to a peculiar rotation of cell polarization and metastasis. However, the mechanisms behind seen in these tumors and is described as an inside-out tumor spread are not completely understood. Stud- growth pattern. This characteristic feature has been ies of micropapillary carcinomas of the breast have supported by electron microscopy studies, which have suggested that the lacunae characteristic of these shown that the microvilli of the cellular surface are tumors may represent prelymphatic spaces that un- at the stromal–epithelial interface and not toward the dergo endothelialization and promote metastasis.11 lumen of the neoplastic gland.2 This inside-out pattern This theory would explain why nodal metastases are is better observed with MUC-1 immunostain2 and has more common in IMPC. Because lymphatic invasion been linked to stromal and vascular invasion.4,5 typically occurs at the infiltrative edges of the tumor, Metastasis to local lymph nodes via the lymphat- our findings that the IMPC component was usual-

208 Cancer Control April 2015, Vol. 22, No. 2 ly identified in this tumoral area may explain this and CD31 are related biomarkers in neoangiogenesis early metastatic and local progression. In our series, and may partially contribute to the biological aggressive of the removed 392 lymph nodes, 95 were positive behavior of IMPC. The coexpression of VEGF and CD31 for metastatic disease; the metastasis was of micro- in IMPC is a novel finding that warrants further inves- papillary morphology in 53 lymph nodes. Although tigation, particularly with respect to the possibility that the percentage of the micropapillary component did targeted antiangiogenic therapy may be more efficacious not appear to correlate with the presence or number in this subset of CRC. of positive lymph nodes, a slight increase of papil- In certain cases of IMPC, we observed collections lary morphology was seen in the metastatic tumors of neutrophils infiltrating the micropapillae and occa- (56% of lymph node metastases were of micropapil- sionally spilling into the lacunar-like spaces, thus mim- lary morphology; see Table). icking microabscesses. Neutrophils are a critical source To explore the role of peritumoral spaces in an- of cytokines, such as interleukin 8, VEGF, and matrix giogenesis, tumor progression, and metastatic spread metalloproteinases 2 and 9.21 Therefore, it is possible of IMPC of a colorectal origin, we studied a panel of that they may also have a role in the neoangiogen- vascular markers (CD31, CD34, D2-40) and VEGF-A. esis associated with IMPC; however, further studies CD31, also known as platelet-endothelial cell comparing IMPC with nonmicropapillary colorectal adhesion molecule 1 (PECAM-1), is a 130k-Da trans- carcinomas are necessary to draw specific conclusions. membrane glycoprotein expressed at high levels on early and mature endothelial cells, platelets, and most Conclusions leukocyte subpopulations. In addition to its role in The results of this study demonstrate that the lacu- the adhesion and transmigration of inflammatory nar spaces surrounding tumor cells in colonic micro- cells, CD31 has an important role in angiogenesis.14-17 papillary carcinoma represent vascular channels and DeLisser et al15 showed that PECAM-1 is involved in vascular endothelial growth factor A is produced by angiogenesis and suggested that endothelial cell–cell tumor cells. Our findings of strong CD31 and vascular adhesion molecules are important in the formation endothelial growth factor A positivity in the lining of new vessels. PECAM-1 also plays a complex role tumor cells of the retraction clefts of invasive micro- in tumor-induced angiogenesis in which neovascu- papillary colorectal carcinoma may represent neoan- larization arises from the interactions between mul- giogenesis around tumor clusters. This also suggests tiple stimulatory factors (eg, VEGF, basic fibroblast a prominent role of neoangiogenesis in the charac- growth factor, interleukin 8, angiogenin).16 CD31 is teristics of early tumor progression, local invasion, involved in the regulation of hematopoietic progeni- and metastatic spread seen in invasive micropapillary tor cell compartmentalization between the peripheral colorectal carcinoma. blood and bone marrow as well as in maintaining the levels of the matrix-degrading enzyme matrix References metalloproteinase 9 in the bone marrow vascular 1. American Cancer Society. Cancer Facts & Figures 2015. Atlanta, GA: niche.15 In all of our IMPC cases, the cells lining the American Cancer Society; 2015. 2. Verdú M, Román R, Calvo M, et al. Clinicopathological and mo- lacunar-like spaces surrounding the micropapillae lecular characterization of colorectal micropapillary carcinoma. Mod Path. strongly stained for CD31 but not for CD34, sug- 2011;24(5):729-738. 3. Kuroda N, Oonishi K, Ohara M, et al. Invasive micropapillary car- gesting that they are newly formed immature blood cinoma of the colon: an immunohistochemical study. Med Mol Morphol. vessels. The same cells were negative for D2-40, which 2007;40(4):226-230. 4. Haupt B, Ro JY, Schwartz MR, et al. Colorectal adenocarcinoma with is a sensitive and relatively specific marker for lym- micropapillary pattern and its association with lymph node metastasis. Mod phatic endothelium in all parenchymatous organs.18 Pathol. 2007;20(7):729-733. 5. Lino-Silva LS, Salcedo-Hernández RA, Caro-Sánchez CH. Colonic VEGF-A is a vascular-related protein widely regard- micropapillary carcinoma, a recently recognized subtype associated with his- ed as a classic angiogenic cytokine because of its role in tological adverse factors: clinicopathological analysis of 15 cases. Colorectal Dis. 2012;14(9):e567-e572. stimulating various endothelial cell responses necessary 6. Guo X, Chen L, Lang R, et al. Invasive micropapillary carcinoma of the for angiogenesis. It belongs to the VEGF–platelet-de- breast: association of pathologic features with lymph node metastasis. Am J Clin Pathol. 2006;126(5):740-746. rived growth factor gene family, which also includes 7. Amin MB, Ro JY, el-Sharkawy T, et al. Micropapillary variant of tran- VEGF-B, VEGF-C, VEGF-D, and VEGF-E. While VEGF-C sitional cell carcinoma of the urinary bladder. Histologic pattern resembling 19 ovarian papillary serous carcinoma. Am J Surg Pathol. 1994;18(12):1224-1232. and VEGF-D are involved in lymphangiogenesis, the 8. Nagao T, Gaffey TA, Visscher DW, et al. Invasive micropapillary salivary expression of VEGF-A has been identified as an early duct carcinoma: a distinct histological variant with biologic significance. Am J 20 Surg Pathol. 2004;28(3):319-326. change in the formation of new blood vessels. In our 9. Amin MB, Tamboli P, Merchant SH, et al. Micropapillary component in study, we used a monoclonal antibody that recognized lung adenocarcinoma: a distinctive histological feature with possible prognostic significance. Am J Surg Pathol. 2002;26(3):358-364. the 189, 165, and 121 isoforms of VEGF. While the 10. Luna-Moré S, Gonzales B, Acedo C, et al. Invasive micropapillary 2 smaller isoforms VEGF-165 and VEGF-121 are secreted carcinoma of the breast. A new special type of invasive mammary carcinoma. Pathol Res Pract. 1994;190(7):668-674. proteins that act as diffusible agents, the larger isoform 11. Acs G, Paragh G, Rakosy Z, et al. The extent of retraction clefts cor- VEGF-189 remains cell associated. Presumably, VEGF relates with lymphatic vessel density and VEGF-C expression and predicts

April 2015, Vol. 22, No. 2 Cancer Control 209 nodal metastasis and poor prognosis in early-stage breast carcinoma. Mod Pathol. 2012;25(2):163-77. 12. Nassar H. Carcinomas with micropapillary morphology: clinical significance and current concepts. Adv Anat Pathol. 2004;11(6):297-303. 13. Nassar H, Pansare V, Zhang H, et al. Pathogenesis of invasive micropapillary carcinoma: role of MUC1 glycoprotein. Mod Pathol. 2004;17(9):1045-1050. 14. Newman PJ, Berndt MC, Gorski J, et al. PECAM-1 (CD31) cloning and relation to adhesion molecules of the immunoglobulin gene superfamily. Science. 1990;247(4947):1219-1222. 15. DeLisser HM, Christofidou-Solomidou M, Strieter RM, et al. Involvement of endothelial PECAM-1/CD31 in angiogenesis. Am J Pathol. 1997;151(3):671-677. 16. Ross EA, Freeman S, Zhao Y, et al. A novel role for PECAM-1 (CD31) in regulating haematopoietic progenitor cell compartmentalization between the peripheral blood and bone marrow. PLoS One. 2008;3(6):e2338. 17. Pusztaszeri MP, Seelentag W, Bosman FT. Immunohistochemical expression of endothelial markers CD31, CD34, von Willebrand factor, and Fli-1 in normal human tissues. J Histochem Cytochem. 2006;54(4):385-395. 18. Fukunaga M. Expression of D2-40 in lymphatic endothelium of normal tissues and in vascular tumors. Histopathology. 2005;46(4):396-402. 19. Gu Y, Qi X, Guo S. Lymphangiogenesis induced by VEGF-C and VEGF-D promotes metastasis and a poor outcome in breast carcinoma: a retrospective study of 61 cases. Clin Exp Metastasis. 2008;25(7):717-725. 20. Chang SH, Kanasaki K, Gocheva V, et al. VEGF-A induces angiogenesis by perturbing the cathepsin-cysteine protease inhibitor balance in venules, causing basement membrane degradation and mother vessel formation. Can- cer Res. 2009;69(10):4537-4544. 21. Jablonska J, Leschner S, Westphal K, et al. Neutrophils responsive to endogenous IFN-beta regulate tumor angiogenesis and growth in a mouse tumor model. J Clin Invest. 2010;120(4):1151-1164.

210 Cancer Control April 2015, Vol. 22, No. 2 Molecularly targeted therapy may offer

more tailored, personalized treatments

for patients with breast cancer.

The Broken Hall. Photograph courtesy of Craig Damlo. www.soapboxrocket.com.

Advances in the Molecular Analysis of Breast Cancer: Pathway Toward Personalized Medicine Marilin Rosa, MD

Background: Breast cancer is a heterogeneous disease that encompasses a wide range of clinical behaviors and histological and molecular variants. It is the most common type of cancer affecting women worldwide and is the second leading cause of cancer death. Methods: A comprehensive literature search was performed to explore the advances in molecular medicine related to the diagnosis and treatment of breast cancer. Results: During the last few decades, advances in molecular medicine have changed the landscape of cancer treatment as new molecular tests complement and, in many instances, exceed traditional methods for determining patient prognosis and response to treatment options. Personalized medicine is becoming the standard of care around the world. Developments in molecular profiling, genomic analysis, and the discovery of targeted drug therapies have significantly improved patient survival rates and quality of life. Conclusions: This review highlights what pathologists need to know about current molecular tests for classification and prognostic/ predictive assessment of breast carcinoma as well as their role as part of the medical team.

Introduction to treatment, presentation, and biology. For example, For years the diagnosis and classification of breast histologically similar tumors may have different clin- cancer has been based on clinicopathological features ical behavior and responses to treatment. Significant such as tumor type and size, lymph node status, and advances in personalized treatment during the last histological grade. However, during the last few de- decade have been due to genomic analysis, which cades, significant advances have taken place in this allows for the molecular study and classification of area as we approach the era of personalized medicine. tumors and gives rise to the availability of biologi- Breast cancer comprises a heterogeneous group cal targeted drugs such as trastuzumab for human of tumors that significantly vary in their responses epidermal growth factor receptor 2 (HER2)–positive breast cancer.1 However, the biological heterogeneity From the Departments of Anatomic Pathology and Women’s Oncology, of tumors continues to be problematic because only a H. Lee Moffitt Cancer Center & Research Institute, Tampa, Florida. subset of patients with a particular type of tumor will Submitted August 19, 2014; accepted February 2, 2015. benefit or respond to targeted treatments. In addition, Address correspondence to Marilin Rosa, MD, Moffitt Cancer Center, most commercially available assays are designed to 12902 Magnolia Drive, Tampa, FL 33612. E-mail: Marilin.Rosa@ determine prognostic and predictive information in Moffitt.org early-stage carcinoma and, thus, offer little clinical No significant relationship exists between the author and the companies/organizations whose products or services may be referenced value to patients with advanced stage or aggressive in this article. variants of breast cancer. Therefore, it is important

April 2015, Vol. 22, No. 2 Cancer Control 211 to continue developing tests that can predict the risk optimal ER and PR testing, tissue must be sectioned of cancer recurrence as well as which patients will at 5-mm intervals and fixed in a sufficient amount of respond to specific therapeutic measures. 10% buffered formalin for at least 6 hours but no more than 72 hours. Cold ischemic time should be for no Evaluation Using Traditional Methods more than 1 hour. In addition, if nuclear staining is Historically, breast cancer has been classified using present in 1% or more tumor cells, then the results are histological grade. The first attempts to grade breast positive. Every PR and ER IHC assay should include cancer based on histology were initiated by Gree- positive and negative controls.17 nough in 1925,2 followed by Patey and Scarff in 1928,3 Haagensen in 1933,4 and Bloom and Richardson in Human Epidermal Growth Factor Receptor 2 1957.5 Elston’s modified Bloom–Richardson grading The HER2 oncogenic protein is a transmembrane system (also called the Nottingham system) is now glycoprotein member of the HER family encoded by used to grade breast cancer. However, interobserver ERBB2.19,20 HER2 is expressed at low levels in several variability and less-than-optimal reproducibility exist normal epithelia, including the breast. HER2 amplifi- among pathologists regarding tumor grading.6 Nev- cation and the accompanying protein overexpression ertheless, tumor grading plays a role in the diagnosis occur in approximately 15% to 20% of breast can- and management of breast cancer because its value as cers.14,21 HER2 overexpression, gene amplification, or an independent prognostic factor for overall survival both are independent prognostic markers of clinical rates in spite of tumor size and nodal status has been outcomes. HER2 is also a predictive factor of tumor proven in numerous studies.7-11 response to chemotherapeutic agents. Its status is In addition to tumor grade, pathologists have tra- typically evaluated to determine patient eligibility ditionally classified breast carcinoma into histological for anti-HER2 therapy. When used as monotherapy types. The general consensus is that select types of or in combination with other drugs or chemothera- carcinoma of the breast are associated with distinctive, peutic agents, HER2-targeted drugs have significantly biological characteristics related to clinical behavior. improved survival and response rates.22-24 However, the classification and diagnostic criteria HER2 status can be determined in formalin-fixed, for each tumor type have also varied throughout the paraffin-embedded (FFPE) tissue by assessing protein years. Select special-type carcinomas are uncommon; expression on the membrane of the tumor cells using thus, they cannot always be included in long-term or IHC or by assessing the number of HER2 copies using molecular studies.12 in situ hybridization. For adequate testing, the fixation process should be initiated within 1 hour of tissue Estrogen and Progesterone Receptors removal and the total fixation time should range from The estrogen receptor (ER) and progesterone recep- a minimum of 6 hours to a maximum of 72 hours. By tor (PR) are prototypical tumor markers that have an IHC, scores of 0 and 1+ are considered to be negative, a immediate impact on systemic treatment decisions score of 3+ is considered to be positive, and a score of for patients with breast cancer. ER and PR are more 2+ constitutes a “gray zone” in which testing with an commonly positive in low- to intermediate-grade tu- alternative methodology is necessary.21,25 mors and in postmenopausal women. They are con- In situ hybridization methods include fluorescence sidered weak prognostic markers but strong predic- in situ hybridization, chromogenic in situ hybridiza- tive factors of tumor response to hormonal therapy tion, dual in situ hybridization, and silver-enhanced (eg, tamoxifen).13,14 PR is also an independent predic- in situ hybridization. Some assays use a single probe tor of response because ER- and PR-positive patients to determine the number of HER2 copies; however, have better responses to treatment than ER-positive, most assays also include a chromosome enumeration PR-negative patients.13-16 probe (CEP17) chromosome 17. Results are reported ER and PR are measured using immunohisto- as follows21,25: chemistry (IHC), which is the preferred method for • Not amplified; ratio < 2.0 with an average measurement due to its wide availability and ease of number of HER2 copies < 4.0 signals/cell use.14 However, it is estimated that nearly 20% of all • Amplified ratio ≥ 2.0 or an average HER2 ER and PR testing may not be accurate,17 and concerns copy number ≥ 6.0 signals/cell regarding current testing, interlaboratory reproducibil- • Equivocal < 2.0 with an average HER2 copy ity, and possible false-negative or false-positive results number ≥ 4.0 and < 6.0 signals/cell exist.13,14 Several factors influence the accurate testing Several factors influence the accuracy of HER2 of ER and PR, including specimen type, fixation type testing and tumor response to therapeutic agents. and time, tissue decalcification, automated compared In the metastatic setting, response rates to trastu- with manual procedure, antibody selection, threshold zumab are below 50%, and many patients initially for positivity, and quality assurance and control.18 For responding to trastuzumab therapy go on to develop

212 Cancer Control April 2015, Vol. 22, No. 2 tumor recurrence and drug resistance.26-30 Although Other Prognostic and Predictive Methods variable responses to trastuzumab may not all be re- Many women diagnosed with early-stage, ER-positive, lated to inaccuracies in testing, the standardization of HER2-negative breast cancer receive systemic chemo- testing methodologies across laboratories is problem- therapy in addition to hormonal therapy following atic; moreover, approximately 20% of HER2 testing surgery to increase the likelihood of cure and reduce may be inaccurate.26-30 the risk of recurrence.38 However, it is possible that The most common factors that affect results many of these women could be spared from the toxic include prolonged fixation in formalin, fixation adverse side effects of systemic chemotherapy if they in nonformalin fixatives, and tissue decalcifica- were accurately stratified into groups that may or may tion that may degrade DNA.21,25 Coamplification not benefit from adjuvant chemotherapy. of CEP17 and HER2 can occur in breast cancer, Although it is not a molecular diagnostic test, causing miscalculation of the HER2:CEP17 ratio Adjuvant! Online (www.adjuvantonline.com) is an and, thus, underreporting of HER2 amplification.31 Internet-based tool whose purpose is to help health In addition, some tumors may demonstrate intra- care professionals and patients with early breast can- tumoral heterogeneity for HER2 amplification. cer review the benefits of adjuvant therapy follow- Although this definition has been challenged,19 ing surgery, whether it is hormonal therapy, chemo- HER2 tumor heterogeneity has been defined as a therapy, or both. This online tool uses information tumor that has at least 5% but no more than 50% of such as estimates of comorbidity, tumor staging and nonclustered tumor nuclei and a HER2:CEP17 ratio characteristics (eg, ER status, size of tumor, num- higher than 2.2.19,32 HER2 tumor heterogeneity may ber of positive axillary nodes), menopausal status, range between 5% and 15% of total cases tested and age to provide baseline prognostic estimates of and tumor heterogeneity may be more frequent 10-year outcomes (with and without adjuvant systemic (up to 27%) in breast carcinomas with an equivocal therapy).39 Most of the prognostic information is based (2+) HER2 score.32,33 on data from the Surveillance, Epidemiology, and End Concordance between HER2 IHC (protein Results program, and projections of the efficacy rates expression) and in situ hybridization (gene am- of adjuvant therapy are based on data from the Ear- plification) is required in at least 95% of cas- ly Breast Cancer Trialists’ Collaborative Group.38,40-42 es.29 Among more than 1,500 patients whom The Adjuvant! Online tool estimates the efficacy of they centrally screened, Perez et al34 found dis- endocrine therapy when used as monotherapy or cordant results in approximately 4%. These find- in combination with systemic chemotherapy to de- ings may be due to amplification without over- termine likely outcomes.43 The tool can also model expression, marked intratumoral heterogeneity, overall survival and disease-free survival rates as well or protein overexpression without amplification. as improvements seen in clinical trials. However, the Response to anti-HER2 therapy in these cases is Adjuvant! Online tool tends to overestimate the risk unknown.21,25,29,35 of recurrence in some patients.38,44 The IHC prognostic model IHC4 uses the quanti- Human Epidermal Growth Factor Receptor 2 tative values of 4 standard IHC assays (ER, PR, HER2, Testing Using Other Methodologies and Ki67). The test was developed using data from The HERmark (Monogram Biosciences, South San a cohort of 1,125 ER-positive patients who did not Francisco, California) assay uses a dual antibody ap- receive adjuvant chemotherapy, had a genomic health proach to make quantitative measurements of HER2 Oncotype (Genomic Health, Redwood City, California) content in FFPE tissue based on the VeraTag Technolo- Recurrence Score, and had adequate tissue for the gy platform (Monogram Biosciences). Compared with 4 IHC measurements (Arimidex, Tamoxifen, Alone single antibody-based IHC methods, the dual antibody or in Combination (ATAC) trial).45 The test was fur- approach increases rates of specificity and sensitivity. ther validated using data from another cohort of 786 HER2:HER2 homodimers and HER2 total protein are ER-positive patients.45-47 To strengthen the IHC4 val- measured by the assay. Current testing methods iden- ue, researchers combined it with clinicopathological tify patients eligible for trastuzumab.36,37 Alternatively, parameters such as tumor grade, size, nodal burden, HERmark can distinguish subpopulations of patients patient age, and type of endocrine treatment (aro- likely to have different clinical outcomes; for example, matase inhibitors or tamoxifen), thus creating the those with higher levels of expression may have better IHC4 + clinical (IHC4+C) score.38,46 The IHC4+C score outcomes than those with lower expression levels. helps predict the residual risk of distant recurrence at Although the data is limited, studies suggested that 9 years in postmenopausal women with node-nega- compared with other marketed tests, HERmark can tive, hormone receptor–positive disease treated with more accurately identify patients likely to respond to 5 years of adjuvant endocrine therapy independent trastuzumab.36,37 of the type of endocrine therapy38; however, the test

April 2015, Vol. 22, No. 2 Cancer Control 213 cannot provide predictive information regarding drug other genes of the ERBB2 amplicon, characterize of choice.46,47 Other researchers have concluded that HER2-enriched cancers. Most (but not all) basal-like the IHC4+C score provides additional prognostic in- tumors correspond to ER-, PR-, and HER2-negative formation in this population that is at least as infor- tumors (Table 1). The normal breast-like subtype has mative as Oncotype DX (Genomic Health).46 It pro- not been reproducibly defined and is thought to be vides a less costly and readily available methodology an artifact of having too few tumor cells and a back- because 3 of the included IHC measurements are ground of normal breast tissue in the sample.48-51 routinely performed in all patients diagnosed with Although this classification gained acceptance, the breast cancer.46 However, variation in tissue handling, initial testing methodology used messenger ribonucle- laboratory testing, and reporting of IHC tests hinders ic acid (mRNA) expression analysis in fresh frozen tis- its generalized clinical use.38,46 sue, thus hampering its introduction into clinical prac- Barton et al38 compared the IHC4+C score tice.48,49,52 To identify subtypes using standard FFPE with the Adjuvant! Online tool and found that the tissue specimens, other alternatives have been sought. IHC4+C score additionally stratified patients Although intrinsic subtypes can be approximated using according their the residual risk of distant re- IHC stains, such as ER, PR, HER2, Ki67, epidermal currence among those already determined by growth factor receptor, and cytokeratins 5/6, 7, 8, 17, Adjuvant! Online as being of intermediate risk. 18, and 19, the use of IHC for this purpose has not The IHC4+C score decreased the level of risk in more gained wide acceptance because the concordance with than 50% of individuals defined as having interme- other molecular methods is not perfect and because of diate risk per the Adjuvant! Online to low risk, thus the complexity introduced when using multiple IHC sparing some patients from adjuvant chemothera- markers on limited amounts of tissue.51,53,54 py.38 In addition, Barton et al38 inferred that the IH- Gene-expression profiling helps to identify genes C4+C score may have clinical utility in individuals with potential to be used as a molecular signature in labeled as high risk by Adjuvant! Online because the guiding therapy and predicting patient prognosis.55 Pop- IHC4+C score downgraded almost one-half of these ular breast cancer multigene predictor test platforms patients to the intermediate risk group and a select include IHC, fluorescence in situ hybridization, reverse few were downgraded to the low-risk group. No ad- transcriptase–polymerase chain reaction (RT-PCR), ditional usefulness of the IHC4+C score was seen in patients classified as low risk Table 1. — Summary of Clinical and Pathological Features by the Adjuvant! Online tool38; thus, this of Main Intrinsic and Molecular Breast Cancer Subtypes online tool may be sufficient for treatment Intrinsic Luminal A Luminal B HER2 Basal Type decisions in these patients. Type Enriched Histological Low to inter- Intermediate to High High Gene-Expression Profiling Grade mediate high Gene-expression profiling has been used for Breast 40 20 20–30 ~ 15 more than 10 years to develop tests so that Carcinomas, % accurate and personalized clinical outcomes Most Common ER positive ER (weaker) ER negative ER negative can be better predicted when compared Marker Results PR positive positive PR negative PR negative with traditional pathological and clinical HER negative PR positive or HER2 positive HER2 negative 48 negative standards. In 2000, Perou et al described Low Ki67 CK5/6 positive for the first time a molecular classification HER2 positive or negative EGFR positive system for breast carcinoma, identifying Higher Ki67 4 major molecular subtypes: ER-positive/ luminal-like, basal-like, ERRB2-positive Prognosis Good Intermediate Poor Poor Mutations in High frequency (HER2-enriched), and normal breast-like. TP53 of BRCA1 Subsequent studies redefined the intrinsic High risk of mutations molecular classification, resulting in a sub- relapse division of the luminal type into types A Targeted Hormonal Hormonal HER2-targeted No targeted and B.49 Luminal type A cancers have a low Treatment therapy therapy therapies (eg, treatment morphological grade and are predominant- trastuzumab) options ly positive for ER, whereas luminal type B Tumor cancers are also predominantly positive for Histology ER, often have a high morphological grade, and may sometimes express low levels of hormone receptors. Amplification and high CK = cytokeratin, EGFR = epidermal growth factor receptor, HER2 = human epidermal growth factor expression of ERBB2, along with several receptor 2.

214 Cancer Control April 2015, Vol. 22, No. 2 and microarray technology. Several tests are commer- identify biological drivers of disease relapse that cially available and many more are in development complement traditional pathological findings, such (Table 2). Some of the most clinically used method- as lymph node status, tumor grade and size, and ologies will be discussed. other biological markers (eg, HER2).59

Immunohistochemistry-Based Multigene Assay Reverse Transcriptase–Polymerase Chain Mammostrat (Clarient, Aliso Viejo, California) can Reaction–Based Multigene Assays help health care professionals make decisions Oncotype DX is a prognostic, predictive, 21-gene pro- regarding the use of chemotherapy as well as file, real-time RT-PCR assay performed by a central endocrine therapy in patients with ER-positive, laboratory using FFPE samples of breast cancer. In early-stage breast cancer. Mammostrat uses IHC individuals with ER-positive, HER2-negative, lymph to evaluate the gene expression of p53, SLC7A5, node–negative carcinoma, the assay is used to deter- CEACAM5, NDRG1, and HTF9C, which have been mine their 10-year risk for disease recurrence, thus selected from 700 gene targets in 3 individual co- assigning them a recurrence score of low risk (< 18), horts, and has been comprehensively and clini- intermediate risk (18–30), or high risk (≥ 31).58 On- cally validated.52,56,57 The biomarkers are then cotype DX was developed after researchers identified analyzed using an algorithm that assesses risk and analyzed 250 candidate genes from 447 partici- for cancer recurrence, which is reported as low pants from 3 separate studies, eventually leading to (7.6% likelihood of distant recurrence during a its 21-gene profile that includes 5 reference genes as 10-year period), moderate (16.3% likelihood of dis- internal controls and 16 genes related to cancer.55 For tant recurrence during a 10-year period), or high the assay, the proliferation and ER pathways followed (20.9% likelihood of distant recurrence during a by the HER2 pathway play the most influential role 10-year period).52,58 This calculated risk of recur- in the calculation of the recurrence score.55 rence is independent of grade, lymph node sta- Current guidelines advise oncologists to withhold tus, and stage.56 In addition, the test appears to chemotherapy in patients with low recurrence scores but to offer treatment to Table 2. — Select Molecular Tests for Predicting Clinical Outcomes those with high recurrence HERmark Mammostrat Oncotype DX PAM50 MammaPrint scores.60 Patients with an Manufacturer Monogram Clarient Diag- Genomic Health NanoString Agendia (Irvine, intermediate recurrence Biosciences nostic Services (Redwood City, Technologies CA) score constitute a “gray (South San (Aliso Viejo, CA) (Seattle, WA) zone.” To further clarify Francisco, CA) CA) the risk of recurrence and Tissue FFPE FFPE FFPE FFPE Fresh, frozen, or benefit from chemother- Sample FFPE apy in such patients, the Type of Quantitative 5-gene profile 21-gene recur- 50 classifier 70-gene profile Trial Assigning Individual- Assay measure- rence score genes and 5 ized Options for Treatment ments of control genes (TAILORx trial) was initiat- ERBB2 ed and is ongoing.53,61 Technology 2 distinct epi- Immunohisto- Real-time Quantitative Microarrays tope-specific chemistry RT-PCR RT-PCR In addition to risk of monoclonal recurrence, Oncotype DX antibodies also appears to predict Proprietary, chemotherapy benefits.53 proximi- ty-based In 2 large studies, lack of technology benefit from chemother- platform apy was associated with Laboratory Centralized Centralized Centralized Centralized Centralized lower recurrence scores, Cost, $ 3,350 ~ 500–600 ~ 4,000 Estimated to be ~ 4,000 and greater benefit from 2,000–3,000 adjuvant therapy was as- Clinical Use Prognostic Prognostic Prognostic Prognostic Prognostic sociated with higher re- Response to Response to currence scores.53,62-64 therapy therapy Oncotype DX is included in treatment guidelines FDA No No No Yes Yes from the American Society Approval? of Clinical Oncology60 and FDA = US Food and Drug Administration, FFPE = formalin-fixed, paraffin-embedded, RT-PCR = reverse transcriptase–poly- National Comprehensive merase chain reaction. Cancer Network.65 How-

April 2015, Vol. 22, No. 2 Cancer Control 215 ever, the assay does have limitations. For example, classifying subtypes of breast cancer, the test yields it is validated in hormone receptor–positive breast a risk of recurrence score, taking into consideration cancer alone. It also has a high false-negative rate the pathological tumor size and a subset of quantita- for HER2 that could lead to the underestimation of tive values for proliferation, luminal gene expression, recurrence risk because such results influence the ESR1 (ER), PGR (PR), and ERBB2.52,55,72,73 recurrence score.55,66 In some cases, the recurrence The PAM50 breast cancer intrinsic classifier test score does not correlate with histological features of is recommended for patients diagnosed with inva- the tumor (ie, low-grade tumors with high recurrence sive breast cancer, regardless of their stage or ER scores). In such cases, inflammatory cells or a cellular status.50,55,73,74 PAM50 has been shown in multivariate stroma may artificially increase the recurrence score analyses to be an independent predictor of survival in low-grade invasive breast cancers.67 In addition rate in breast cancer, as well as independent and to these issues, the strength of this genomic assay superior to ER status, tumor grade, lymph node status, has been challenged by some researchers who sug- and other variables.50,55,73,74 gest that similar results and prognostic information However, PAM50 does not entirely correlate with can be obtained using less costly and more widely IHC results because the test may classify some tumors available testing methods (eg, IHC).46,55 However, be- that are HER2-positive by standard techniques, such cause Oncotype DX is performed in a centralized set- as IHC or in situ hybridization, as luminal types A or ting, problems with test reproducibility and issues of B. Conversely, up to 30% of HER2-enriched tumors by interpretation are minimized.68 PAM50 are clinically negative for HER2.52,75 In an effort to predict the risk of recurrence of ductal carcinoma in situ (DCIS), the Oncotype DX Microarray-Based Multigene Assays DCIS score was introduced.69 Approximately 30% of The Symphony Genomic Breast Cancer Profile (Agen- women with DCIS treated with breast-conservation dia, Irvine, California) is a comprehensive analysis of surgery alone have local recurrence, and adjuvant gene expression that includes MammaPrint (Agendia), radiotherapy has been shown to decrease the rate of BluePrint (Agendia), and TargetPrint (Agendia). recurrence by 50% in such individuals.70 The Oncotype MammaPrint is approved by the US Food and DX DCIS score is a 12-gene profile (7 cancer-related Drug Administration to determine whether pa- and 5 reference genes) assay that estimates individual tients can safely avoid chemotherapy and its toxic 10-year risk of local recurrence (DCIS or invasive car- adverse effects. BluePrint is a molecular subtyping cinoma) to determine whether adjuvant radiotherapy profile that discriminates among 3 distinct molec- is likely to be beneficial in women with DCIS treated ular subtypes (luminal, basal-like, and HER2-en- by local excision with or without tamoxifen. The test riched) and also guides treatment choices and has been validated for69: combination therapies to optimize the treatment • Low- to intermediate-grade DCIS with tumor of breast cancer. TargetPrint tests for ER, PR, and size ≤ 2.5 cm HER2; therefore, it is used to determine whether • High-grade DCIS with tumor size ≤ 1.0 cm patients with breast cancer are candidates for with a minimum negative margin width hormonal therapy or other targeted therapies. ≥ 3 mm or no tumor on re-excision For brevity, this article will focus its review on The risk is stratified into low (< 39), intermediate MammaPrint alone. (39–54), or high (≥ 55) categories.69 When used in this context, the Oncotype DX DCIS can predict which MammaPrint women might be spared from adjuvant radiothera- MammaPrint is a microarray, in vitro, 70-gene ex- py. However, the score does not account for other pression profile that includes genes considered to important clinical and pathological features that may be the hallmarks of local invasion, cancer-related influence risk of recurrence, including tumor size, biology, regulators of cell cycle, proliferation, me- margin width, tumor grade, and necrosis. Therefore, tastasis, extravasation, survival in circulation, angio- if Oncotype DX DCIS is ordered outside the clinically genesis, and adaptation to the microenvironment.55 validated variables, its results may be inaccurate and MammaPrint provides prognostic information, and potentially risky to patients.71 its value is independent of conventional pathologi- The Prediction Analysis of Microarrays (PAM50; cal and clinical factors (eg, HER2 status, tumor size, NanoString Technologies, Seattle, Washington) was hormone receptor status).55 This test can also identify introduced in 2009 by Parker et al.72 It uses a quan- groups of patients at low risk within a node-posi- titative RT-PCR assay on FFPE tissue, measuring the tive population traditionally viewed as having a high gene expression of 50 classifier and 5 control genes risk for recurrence; thus, this assay has prognostic to identify luminal type A, luminal type B, HER2-en- value in these individual patients.76 The test results riched, and basal-like breast cancer.72 In addition to are reported as low or high risk. Its prognostic use

216 Cancer Control April 2015, Vol. 22, No. 2 is approved for women under the age of 61 who sented with new and unique opportunities to directly have ER-positive or ER-negative, lymph node-negative engage in research and patient care. The practice of breast cancer, and whose tumors are smaller than pathology has shifted from being purely diagnostic 5 cm in size. Those considered to be low risk per the to having a central role on the medical team. Tradi- results of the test should be advised to avoid adjuvant tionally, pathology departments have acted as “safe chemotherapy and instead receive endocrine therapy keepers” or “custodians” of tissue samples removed alone, whereas those at high risk are generally advised from patients for treatment reasons, diagnostic rea- to receive chemotherapy. The test is highly accurate sons, or both.83 New challenges regarding the ethical, for patients at low risk because the likelihood of pro- regulatory, and legal aspects involved in tissue man- gression to metastatic disease in these cases is low. agement require pathologists to be informed about However, for patients at high risk, the prediction of the technical aspects of appropriate specimen man- metastatic progression is not as precise because only agement and suitability.84 approximately 25% of these patients will progress at Molecular testing is performed for diagnostic, clin- 5 years, which may be partly due to the use of ical (to obtain predictive or prognostic information), adjuvant therapies in these patients.77 In addition, the or research purposes. The ethical, legal, and technical majority of ER-negative patients will be classified as issues involved in each of these applications are, to being high risk based on the results of the assay.77 some extent, different. Molecular tests developed for To determine whether individuals classified to clinical uses are subject to stringent regulations and be at low risk by MammaPrint but at high risk via can only be performed in laboratories certified by the Adjuvant! Online tool could safely avoid chemo- the Clinical Laboratory Improvement Amendments therapy, the Microarray In Node negative and 1-3 (CLIA).85,86 Typically, these tests do not require in- positive lymph node Disease may Avoid Chemother- formed consent unless they are performed to detect apy (MINDACT) and MammaPrint trial with Adjuvant heritable genetic mutations such as BRCA1/2.85,86 Tests Online! was designed to randomize individuals with developed for research purposes alone do not need breast cancer and discordant results to receive either to undergo the same degree of federal and state regu- hormonal therapy alone or hormonal therapy in com- lations and can be performed in smaller, noncertified bination with chemotherapy.53,55 At the time of writing, laboratories.85,86 Tests performed in laboratories not the results of this trial were not available.54,55 certified by CLIA cannot be formally reported or used The main limitation for the clinical use of Mam- for clinical management decisions.85,86 maPrint is that it could not be performed on FFPE The success of molecular medicine depends to tissue, and tissue collected into an RNA preservative a great extent on the quality of tissues to be tested. solution or fresh-frozen tumor samples was required.77 Preanalytic variables, such as tissue and biological RNA retrieval from FFPE material is challenging due sample collection, conservation, and transportation to the partial degradation of RNA during processing.78 requirements, vary according to the type of test re- Several investigators have tried to resolve this issue quested. Many molecular tests can be performed on using archival FFPE tissue for clinical and research FFPE tissue; few tests require fresh or frozen material. purposes, and their studies have demonstrated that Fixation type, duration, and adequate tissue process- applying DNA microarray analysis to FFPE tissue is ing in the laboratory are crucial to ensure that genetic possible, mainly due to improvements in technolo- material is properly preserved. However, the role of gy and stringent protocols for tissue processing.78-81 the pathologist is not limited to supervision and guar- A study by Sapino et al82 validated the MammaPrint antees of adequate tissue processing. Because of the assay on FFPE tissue blocks. The authors compared broad spectrum of currently available molecular tests, the results between fresh tissue samples and FFPE no single laboratory is likely to be capable of offering tissue in an independent series of matched tissue from them all.85 Pathology departments must have policies 5 hospitals and found an overall equivalence rate of to establish how to handle “send-outs” for specialized 91.5%, a precision rate of 97.3%, a repeatability rate tests, including clinical validity (medical necessity) of 97.8%, and highly reproducible results between of the test, adequate block and tissue selection, the replicate samples of the same tumor and between release of tissue blocks or slides, covering the cost of 2 laboratories (concordance rate, 96%).82 These re- supplies, shipping, and the test itself, among others. sults may open the door for the wide clinical use of The pathologist must also be aware of the amount MammaPrint. of tissue needed for diagnostic purposes and make sure enough tissue remains to perform molecular Role of Pathologists in the Era of Molecular studies, if needed, and when possible. The pathologist and Personalized Medicine is also responsible for ensuring that enough tissue is Because specialized molecular tests are aimed to de- available in the block to perform the requested test liver personalized cancer care, pathologists are pre- and to inform the testing facility about any deviation

April 2015, Vol. 22, No. 2 Cancer Control 217 breast cancer population. Breast Cancer Res Treat. 1999;53(1):1-8. to specimen collection or processing (eg, prolonged 10. Elston CW, Ellis IO. Pathologic prognostic factors in breast cancer. I. fixation time). In addition, the pathologist should The value of histological grades in breast cancer: experience from a large study with long-term follow-up. Histopathology. 1991;19(5):403-410. select the tissue block with the greatest amount of 11. Schwartz AM, Henson DE, Chen D, et al. Histologic grade remains a invasive tumor, highest tumor grade (unless otherwise prognostic factor for breast cancer regardless of the number of positive lymph nodes and tumor size: a study of 161 708 cases of breast cancer from the specified by the oncologist), and the least amount of SEER program. Arch Pathol Lab Med. 2014;138(8):1048-1052. in situ carcinoma, stromal inflammation, and biop- 12. Weigelt B, Reis-Filho JS. Histological and molecular types of breast cancer: is there a unifying taxonomy? Nat Rev Clin Oncol. 2009;6(12):718-730. sy changes. Testing results should also be correlated 13. Diaz LK, Sneige N. Estrogen receptor analysis for breast cancer: current with histological findings to identify and resolve any issues and keys to increasing testing accuracy. Adv Anat Pathol. 2005;12(1):10-9. 87 14. Gown AM. Current issues in ER and HER2 testing by IHC in breast discordant results. cancer. Mod Pathol. 2008;21(suppl 2):S8-S15. The pathologist also has a duty to ensure that, 15. Sommer S, Fuqua SA. Estrogen receptor and breast cancer. Semin Cancer Biol. 2001;11(5):339-352. when material collected for diagnostic purposes, treat- 16. American Society of Clinical Oncology. Clinical practice guidelines ment purposes, or both types of purposes is then used for the use of tumor markers in breast and colorectal cancer. Adopted on May 17, 1996 by the American Society of Clinical Oncology. J Clin Oncol. for research, enough tissue remains for clinical tests 1996;14(10):2843-2877. that may be needed or developed in the future and 17. Hammond ME, Hayes DF, Dowsett M, et al; American Society of Clinical Oncology; College of American Pathologists. Guideline recommendations for that such research has been reviewed and approved immunohistochemical testing of estrogen and progesterone receptors in breast by the corresponding Institutional Review Board. cancer (unabridged version). Arch Pathol Lab Med. 2010;134(7):e48-e72. 18. Rhodes A, Jasani B, Balaton AJ, et al. Frequency of oestrogen and progesterone receptor positivity by immunohistochemical analysis in 7016 Conclusions breast carcinomas: correlation with patient age, assay sensitivity, threshold value, and mammographic screening. J Clin Pathol. 2000;53(9):688-696. Developments in molecular medicine help to iden- 19. Chang MC, Malowany JI, Mazurkiewicz J, et al. ‘Genetic heterogeneity’ tify biologically different subtypes of breast cancer in HER2/neu testing by fluorescence in situ hybridization: a study of 2,522 cases. Mod Pathol. 2012;25(5):683-688. and characterize the risk of recurrence as well as 20. Sapino A, Goia M, Recupero D, et al. Current challenges for HER2 to predict treatment responses among patients with testing in diagnostic pathology: state of the art and controversial issues. Front Oncol. 2013;3:129. breast cancer. In addition to traditional prognostic and 21. Wolff AC, Hammond ME, Hicks DG, et al; American Society of Clinical predictive factors, such as estrogen and progesterone Oncology; College of American Pathologists. Recommendations for human epidermal growth factor receptor 2 testing in breast cancer: clinical practice receptors, several assays that support breast cancer guideline update. Arch Pathol Lab Med. 2014;138(2):241-256. prognostication in clinical practice are available. These 22. Geyer CE, Forster J, Lindquist D, et al. Lapatinib plus capecitabine for HER2-positive advanced breast cancer [Erratum appears in N Engl J Med. tests have been designed to help oncologists and pa- 2007;356(14):1487]. N Engl J Med. 2006;355(26):2733-2743. tients make appropriate decisions regarding the use 23. Baselga J, Cortés J, Kim SB, et al; CLEOPATRA Study Group. Pertu- zumab plus trastuzumab plus docetaxel for metastatic breast cancer. N Engl of adjuvant systemic therapy in addition to surgery. J Med. 2012;366(2):109-119. However, molecular tests should be used together 24. Verma S, Miles D, Gianni L, et al; EMILIA Study Group. Trastuzumab emtansine for HER2- positive advanced breast cancer [Erratum appears in and not as a substitute to well-established patholog- N Engl J Med. 2013;368(25):2442]. N Engl J Med. 2012;367(19):1783-1791. ical and clinical variables used in routine practice to 25. College of American Pathologists. Template for reporting results of biomarker testing of specimens from patients with carcinoma of the breast. evaluate patient prognosis (eg, tumor grade, lymph Published 2012. http://www.cap.org/apps/docs/committees/cancer/cancer node status). Technical advances in molecular med- _protocols/breast_cancer_biomarker_template.pdf. Accessed March 8, 2015. 26. Larson JS, Goodman LJ, Tan Y, et al. Analytical validation of a highly icine and the increasing number of molecularly tai- quantitative, sensitive, accurate, and reproducible assay (HERmark) for the lored treatments have made it possible to potentially measurement of HER2 total protein and HER2 homodimers in FFPE breast cancer tumor specimens. Patholog Res Int. 2010;2010:814176. offer individualized treatment options to patients in 27. de Alava E, Ocaña A, Abad M, et al. Neuregulin expression modulates the near future. clinical response to trastuzumab in patients with metastatic breast cancer. J Clin Oncol. 2007;25(19):2656-2663. 28. Piccart M, Lohrisch C, Di Leo A, et al. The predictive value of HER2 References in breast cancer. Oncology. 2001;61(suppl 2):73-82. 29. Wolff AC, Hammond ME, Schwartz JN, et al; American Society of Clin- 1. Kalia M. Personalized oncology: recent advances and future challenges. ical Oncology; College of American Pathologists. Guideline recommendations Metabolism. 2013;62(suppl 1):S11-S14. for human epidermal growth factor receptor 2 testing in breast cancer. Arch 2. Greenough RB. Varying degrees of malignancy in cancer of the breast. Pathol Lab Med. 2007;131(1):18-43. J Cancer Res. 1925;9:453-463. 30. Paik S, Bryant J, Tan-Chiu E, et al. Real-world performance of HER2 3. Patey DH, Scarff RW. The position of histology in the prognosis carci- testing--National Surgical Adjuvant Breast and Bowel Project experience. noma of the breast. Lancet. 1928;1:801-804. J Natl Cancer Inst. 2002;94(11)852-854. 4. Haagensen CD. The bases for the histologic grading of carcinoma of 31. Varga Z, Tubbs RR, Wang Z, et al. Co-amplification of the HER2 gene the breast. Am J Cancer. 1933;19:285-327. and chromosome 17 centromere: a potential diagnostic pitfall in HER2 testing 5. Bloom HJ, Richardson WW. Histological grading and prognosis in breast in breast cancer. Breast Cancer Res Treat. 2012;132(3):925-935. cancer; a study of 1409 cases of which 359 have been followed for 15 years. 32. Vance GH, Barry TS, Bloom KJ, et al; College of American Pathologists. Br J Cancer. 1957;11(3):359-377. Genetic heterogeneity in HER2 testing in breast cancer: panel summary and 6. Robbins P, Pinder S, de Klerk N, et al. Histological grading of breast guidelines. Arch Pathol Lab Med. 2009;133(4):611-612. carcinomas: a study of interobserver agreement. Hum Pathol. 1995;26(8): 33. Ohlschlegel C, Zahel K, Kradolfer D, et al. HER2 genetic heterogeneity 873-879. in breast carcinoma. Clin Pathol. 2011;64(12):1112-1116. 7. Rakha EA, El-Sayed ME, Lee AH, et al. Prognostic significance of 34. Perez EA, Suman VJ, Davidson NE, et al. HER2 testing by local, central, Nottingham histologic grade in invasive breast carcinoma. J Clin Oncol. and reference laboratories in specimens from the North Central Cancer Treatment 2008;26(19):3153-3158. Group N9831 intergroup adjuvant trial. J Clin Oncol. 2006;24(19):3032-3038. 8. Pereira H, Pinder SE, Sibbering DM, et al. Pathological prognostic 35. Reddy JC, Reimann JD, Anderson SM, et al. Concordance between factors in breast cancer. IV: Should you be a typer or a grader? A comparative central and local laboratory HER2 testing from a community-based clinical study of two histological prognostic features in operable breast carcinoma. study. Clin Breast Cancer. 2006;7(2):153-157. Histopathology. 1995;27(3):219-226. 36. Lipton A, Ali S. HER2 protein expression and homodimer levels predict 9. Sundquist M, Thorstenson S, Brudin L, et al; South East Swedish Breast response to trastuzumab in centrally tested FISH-positive metastatic breast Cancer Study Group. Applying the Nottingham Prognostic Index to a Swedish cancer patients. Presented at: Proceedings of the 31st CTRC-AACR San

218 Cancer Control April 2015, Vol. 22, No. 2 Antonio Breast Cancer Symposium; San Antonio, Texas; 2008. assay in postmenopausal, node-positive, oestrogen-receptor-positive breast 37. Bates M, Sperinde J, Köstler WJ, et al Sperinde J, Ali S, Leitzel K, et cancer on chemotherapy: a retrospective analysis of a randomised trial. Lancet al. Identification of a subpopulation of metastatic breast cancer patients with Oncol. 2010;11(1):55-65. very high HER2 expression levels and possible resistance to trastuzumab. 64. Marchionni L, Wilson RF, Marinopoulos SS, et al. Impact of Gene Ex- Ann Oncol. 2011;22(9):2014-2020. pression Profiling Tests on Breast Cancer Outcomes. Rockville, MD: Agency 38. Barton S, Zabaglo L, A’Hern R, et al. Assessment of the contribution for Healthcare Research and Quality; 2008. of the IHC4+C score to decision making in clinical practice in early breast 65. National Comprehensive Cancer Network. NCCN clinical prac- cancer. Br J Cancer. 2012;106(11):1760-1765. tice guidelines in oncology: breast cancer, v2.2015. http://www.nccn.org 39. Olivotto IA, Bajdik CD, Ravdin PM, et al. Population-based validation /professionals/physician_gls/pdf/breast.pdf. Accessed April 29, 2015. of the prognostic model ADJUVANT! for early breast cancer. J Clin Oncol. 66. Park MM, Ebel JJ, Zhao W, et al. ER and PR immunohistochemistry 2005;23(12):2716-2725. and HER2 FISH versus oncotype DX: implications for breast cancer treatment. 40. Early Breast Cancer Trialists’ Collaborative Group. Polychemother- Breast J. 2014;20(1):37-45. apy for early breast cancer: an overview of the randomised trials. Lancet. 67. Acs G, Esposito NN, Kiluk J, et al. A mitotically active, cellular tumor 1998;352(9132):930-942. stroma and/or inflammatory cells associated with tumor cells may contribute 41. Early Breast Cancer Trialists’ Collaborative Group. Tamoxifen to intermediate or high Oncotype DX Recurrence Scores in low-grade invasive for early breast cancer: an overview of the randomised trials. Lancet. breast carcinomas. Mod Pathol. 2012;25(4):556-566. 1998;351(9114):1451-1467. 68. Ross JS. Multigene predictors in early-stage breast cancer: moving in 42. Early Breast Cancer Trialists’ Collaborative Group (EBCTCG). Effects or moving out? [Erratum appears in Expert Rev Mol Diagn. 2008;8(5):669]. of chemotherapy and hormonal therapy for early breast cancer on recur- Expert Rev Mol Diagn. 2008;8(2):129-135. rence and 15-year survival: an overview of the randomised trials. Lancet. 69. Solin LJ, Gray R, Baehner FL, et al. A multigene expression assay to 2005;365(9472):1687-1717. predict local recurrence risk for ductal carcinoma in situ of the breast. J Natl 43. Ravdin PM, Siminoff LA, Davis GJ, et al. Computer program to assist in Cancer Inst. 2013;105(10):701-710. making decisions about adjuvant therapy for women with early breast cancer. 70. Printz C. CancerScope: new method to predict DCIS recurrence. J Clin Oncol. 2001;19(4):980-991. Cancer. 2013;119(22):3899. http://onlinelibrary.wiley.com/doi/10.1002 44. Dowsett M, Cuzick J, Wale C, et al. Prediction of risk of distant recur- /cncr.28458/epdf. Accessed April 28, 2015. rence using the 21-gene recurrence score in node-negative and node-posi- 71. Lagios MD, Silverstein MJ. Risk of recurrence of ductal carcinoma in tive postmenopausal patients with breast cancer treated with anastrozole or situ by oncotype Dx technology: some concerns. Cancer. 2014;120(7):1085. tamoxifen: a TransATAC study. J Clin Oncol. 2010;28(11):1829-1834. 72. Parker JS, Mullins M, Cheang MC, et al. Supervised risk predictor of 45. Dowsett M, Allred C, Knox J, et al. Relationship between quantitative breast cancer based on intrinsic subtypes. J Clin Oncol. 2009;27(8):1160-1167. estrogen and progesterone receptor expression and human epidermal growth 73. Nielsen T, Wallden B, Schaper C, et al. Analytical validation of the factor receptor 2 (HER-2) status with recurrence in the arimidex, tamoxifen, PAM50-based Prosigna Breast Cancer Prognostic Gene Signature Assay alone or in combination trial. J Clin Oncol. 2008;26(7):1059-1065. and nCounter Analysis System using formalin-fixed paraffin-embedded breast 46. Cuzick J, Dowsett M, Pineda S, et al. Prognostic value of a com- tumor specimens. BMC Cancer. 2014;14:177. bined estrogen receptor, progesterone receptor, Ki-67, and human epidermal 74. Chia SK, Bramwell VH, Tu D, et al. A 50-gene intrinsic subtype classifier growth factor receptor 2 immunohistochemical score and comparison with for prognosis and prediction of benefit from adjuvant tamoxifen. Clin Cancer the Genomic Health recurrence score in early breast cancer. J Clin Oncol. Res. 2012;18(16):4465-4472. 2011;29(32):4273-4278. 75. Cancer Genome Atlas Network. Comprehensive molecular portraits 47. Dowsett M, Salter J, Zabaglo L, et al. Predictive algorithms for adjuvant of human breast tumours. Nature. 2012;490(7418):61-70. therapy: TransATAC. Steroids. 2011;76(8):777-780. 76. Mook S, Schmidt MK, Weigelt B, et al. The 70-gene prognosis signature 48. Perou CM, Sørlie T, Eisen MB, et al. Molecular portraits of human predicts early metastasis in breast cancer patients between 55 and 70 years breast tumours. Nature. 2000;406(6797):747-752. of age. Ann Oncol. 2010;21(4):717-722. 49. Sørlie T, Perou CM, Tibshirani R, et al. Gene expression patterns of 77. Slodkowska EA, Ross JS. MammaPrint 70-gene signature: another breast carcinomas distinguish tumor subclasses with clinical implications. Proc milestone in personalized medical care for breast cancer patients. Expert Rev Natl Acad Sci U S A. 2001;98(19):10869-10874. Mol Diagn. 2009;9(5):417-422. 50. Wu Y, Sahin A. Molecular classification and testing of breast carcinoma. 78. Mittempergher L, de Ronde JJ, Nieuwland M, et al. Gene expression In: Hoda S, Brogi E, Koerner F, et al, ed. Rosen’s Breast Pathology. 4th ed. profiles from formalin fixed paraffin embedded breast cancer tissue are largely Philadelphia: LWW; 2014:1337-1342. comparable to fresh frozen matched tissue. PLoS One. 2011;6(2):e17163. 51. Bastien RR, Rodríguez-Lescure Á, Ebbert MT, et al. PAM50 breast 79. Sadi AM, Wang DY, Youngson BJ, et al. Clinical relevance of DNA cancer subtyping by RT-qPCR and concordance with standard clinical mo- microarray analyses using archival formalin-fixed paraffin-embedded breast lecular markers. BMC Med Genomics. 2012;5:44. cancer specimens. BMC Cancer. 2011;11:253:1-13. 52. Sinn P, Aulmann S, Wirtz R, et al. Multigene assays for classification, 80. Fedorowicz G, Guerrero S, Wu TD, et al. Microarray analysis of RNA prognosis, and prediction in breast cancer: a critical review on the background extracted from formalin-fixed, paraffin-embedded and matched fresh-frozen and clinical utility. Geburtshilfe Frauenheilkd. 2013;73(9):932-940. ovarian adenocarcinomas. BMC Med Genomics. 2009;2:23. 53. Sotiriou C, Wirapati P, Loi S, et al. Gene expression profiling in breast 81. Liu A, Tetzlaff MT, Vanbelle P, et al. MicroRNA expression profiling cancer: understanding the molecular basis of histologic grade to improve outperforms mRNA expression profiling in formalin-fixed paraffin-embedded prognosis. J Natl Cancer Inst. 2006;98(4):262-272. tissues. Int J Clin Exp Pathol. 2009;2(6):519-527. 54. Nielsen TO, Parker JS, Leung S, et al. A comparison of PAM50 intrin- 82. Sapino A, Roepman P, Linn SC, et al. MammaPrint molecular diagnostics sic subtyping with immunohistochemistry and clinical prognostic factors in on formalin-fixed, paraffin-embedded tissue. J Mol Diagn. 2014;16(2):190-197. tamoxifen-treated estrogen receptor-positive breast cancer. Clin Cancer Res. 83. Dry S. Who owns diagnostic tissue blocks? Lab Med. 2009;40:69-73. 2010;16(21):5222-5232. http://labmed.ascpjournals.org/content/40/2/69.full. Accessed March 8, 2015. 55. Kittaneh M, Montero AJ, Glück S. Molecular profiling for breast cancer: 84. LiVolsi V, Upton MP. The anatomic pathologist meets molecular pathol- a comprehensive review. Biomark Cancer. 2013;5:61-70. ogy. Am J Clin Pathol. 2012;137(3):341-342. 56. Ring BZ, Seitz RS, Beck R, et al. Novel prognostic immunohistochemi- 85. Dry S, Grody WW, Papagni P. Stuck between a scalpel and a rock, cal biomarker panel for estrogen receptor-positive breast cancer. J Clin Oncol. or molecular pathology and legal-ethical issues in use of tissues for clin- 2006;24(19):3039-3047. ical care and research: what must a pathologist know? Am J Clin Pathol. 57. Ross DT, Kim CY, Tang G, et al. Chemosensitivity and stratification by 2012;137(3):346-355. a five monoclonal antibody immunohistochemistry test in the NSABP B14 and 86. Grody WW, Richards CS. New quality assurance standards for rare B20 trials. Clin Cancer Res. 2008;14(20):6602-6609. disease testing. Genet Med. 2008;10(5):320-324. 58. Ross JS, Hatzis C, Symmans WF, et al. Commercialized multigene 87. Allison KH. Molecular pathology of breast cancer: what a pathologist predictors of clinical outcome for breast cancer [Erratum appears in Oncologist. needs to know. Am J Clin Pathol. 2012;138(6):770-780. 2008;13(8)]. Oncologist. 2008;13(5):477-493. 59. Bartlett JM, Thomas J, Ross DT, et al. Mammostrat as a tool to stratify breast cancer patients at risk of recurrence during endocrine therapy. Breast Cancer Res. 2010;12(4):R47. 60. Harris L, Fritsche H, Mennel R, et al; American Society of Clinical Oncol- ogy. American Society of Clinical Oncology 2007 update of recommendations for the use of tumor markers in breast cancer. J Clin Oncol. 2007;25(33):5287-312. 61. Sparano JA. TAILORx: trial assigning individualized options for treatment (Rx). Clin Breast Cancer. 2006;7(4):347-350. 62. Paik S, Tang G, Shak S, et al. Gene expression and benefit of chemotherapy in women with node-negative, estrogen receptor-positive breast cancer. J Clin Oncol. 2006;24(23):3726-3734. 63. Albain KS, Barlow WE, Shak S, et al; Breast Caner Intergroup of North America. Prognostic and predictive value of the 21-gene recurrence score

April 2015, Vol. 22, No. 2 Cancer Control 219 IHC stains help guide the differential

diagnosis in challenging epithelial

lesions of the breast.

The Oven Vent. Photograph courtesy of Craig Damlo. www.soapboxrocket.com.

Use of Immunohistochemical Stains in Epithelial Lesions of the Breast Laila Khazai, MD, and Marilin Rosa, MD

Background: During the last few decades, immunohistochemistry (IHC) has become an integral part of pathology. Although hematoxylin and eosin (H & E) stain remains the fundamental basis for diagnostic pathology of the breast, IHC stains provide useful and sometimes vital information. Moreover, considering the role of hormonal therapy in hormone receptor–positive breast tumors, as well as the availability of targeted chemotherapeutic agents for HER2-positive cases, IHC studies represent a major part of workups. Methods: A literature search was performed to explore the uses of IHC stains related to the diagnoses of breast lesions and prognostic/predictive information. Results: Selective use of IHC stains in conjunction with H & E examination helps resolve most diagnostic issues encountered by surgical pathologists during their day-to-day practice. Pathologists should be familiar with the use of each immunostain and its limitations to avoid interpretative errors. Conclusions: IHC stains help guide the differential diagnosis of challenging epithelial lesions of the breast. They should be selectively and judiciously used and their findings must be interpreted with the differential diagnoses in mind and with an understanding of possible pitfalls.

Introduction tinction can be problematic. In addition, interobserver In the majority of cases, diagnoses of benign and ma- variability exists among pathologists when interpret- lignant epithelial lesions of the breast are achievable ing difficult and borderline lesions of the breast.1,2 using hematoxylin and eosin (H & E) microscopic Therefore, using immunohistochemical (IHC) stains sections alone. However, in some circumstances, such can be of help when dealing with these challenging as in atypical and borderline breast lesions, as well as lesions, particularly in cases for which the diagnosis in core needle biopsy (CNB), this morphological dis- carries a significant impact on management and prognosis. Pathologists involved in the diagnosis of breast From the Departments of Anatomic Pathology and Women’s Oncology, cases should be familiar with diagnostic challenges H. Lee Moffitt Cancer Center & Research Institute, Tampa, Florida. they may encounter during their daily practice. Submitted December 2, 2014; accepted February 18, 2015. The purpose of this article is to describe the Address correspondence to Marilin Rosa, MD, Moffitt Can- ancillary studies used for the diagnosis of benign, bor- cer Center, 12902 Magnolia Drive, Tampa, FL 33612. E-mail: derline, and malignant epithelial lesions of the breast [email protected] as well as to describe studies used for the prognostic No significant relationships exist between the authors and the companies/organizations whose products or services may be referenced assessment of cancer. For the purpose of discussion, in this article. IHC stains used in breast pathology are divided into

220 Cancer Control April 2015, Vol. 22, No. 2 2 groups: those used for diagnostic purposes and those used to obtain prognostic/predictive information.

Diagnostic Immunohistochemical Stains Benign and Atypical Ductal Proliferative Lesions Diagnosing benign (usual and florid) ductal hyper- A B plasia from atypical ductal hyperplasia (ADH) can be difficult on H & E examination. In addition to providing information regarding long-term cancer risk stratification, distinguishing the 2 groups in CNB specimens helps the clinical team plan the next step in the workup process. The likelihood of finding a more severe lesion (ie, in situ, invasive carcinoma) on excision after C D the diagnosis of ADH ranges from 12% to 36%.3 Therefore, in most centers, the diagnosis of ADH is followed by excisional biopsy; consequently, making the correct diagnosis in borderline cases has great clinical implications. For several years, IHC has been used by pathologists for this diagnostic purpose. High-molecular-weight keratins (cytoker- E F atins [CK] 5 and 6 and 34βE12) in conjunction with estrogen receptor (ER) IHC may be helpful in these borderline cases. Atypical proliferative lesions and in situ carcinoma lack staining with such keratins or they show occasional staining alone. By contrast, a strong and diffuse staining pattern indicates a benign process (Fig 1A).4,5 G H When using these stains for the diagnosis of Fig 1A–H. — (A) Breast duct partially filled by a proliferative process. atypia in our practice, we have found that a poten- The differential diagnosis includes usual and atypical ductal hyperpla- tial pitfall is that cells with apocrine differentiation sia. Negative staining for CK5/6 IHC favors a diagnosis of atypical ductal hyperplasia (× 100). (B) CK5/6 IHC is negative in apocrine metaplasia are generally negative. This finding is important involving papilloma. Because apocrine cells are typically negative for CK5/6, when diagnosing atypia in columnar cell chang- a negative result should not be interpreted as atypia. This constitutes a es, which characteristically have apocrine features possible pitfall when interpreting CK5/6 stain in apocrine lesions (× 50). 3 (C) ER stain shows scattered positivity in benign breast tissue (× 200). (Fig 1B). In this setting, the pathologist should rely (D) Conversely, ER is strongly and diffusely positive in low-nuclear-grade on cytologic and architectural features on H & E DCIS (× 50). (E) Myoepithelial cell marker calponin is positive at the periph- examination alone. ER is sometimes used because, ery of this cribriform carcinoma, confirming it as an in situ process (× 50). (F) Small glands of tubular carcinoma lack myoepithelial cells as demon- in the setting of benign breast tissue, ER shows strated by this negative calponin IHC. Note the presence of benign glands in scattered positivity alone in lobules and ducts. Con- the nearby vicinity (× 200). (G) Low-grade solid carcinoma in situ is strongly versely, in atypical and low-grade neoplastic pro- positive for E-cadherin. This stain result is consistent with ductal type (× 100). (H) p120 shows cytoplasmic positivity in invasive lobular carcinoma. A be- cesses such as ADH or low-grade ductal carcinoma nign duct at the center reveals positive membranous staining and serves in situ (DCIS), this stain tends to be diffusely and as an internal control (× 200). CK = cytokeratin, DCIS = ductal carcinoma in strongly positive (Fig 1C and 1D). situ, ER = estrogen receptor, IHC = immunohistochemistry. It is important to highlight that these IHC stains DCIS or prominent inflammatory infiltrate associ- help classify a proliferative lesion into benign or atyp- ated with DCIS. In addition, certain entities are ical, and routine histological criteria must be followed known to cause diagnostic challenges due to their to distinguish between ADH and DCIS. complex architectural features. Invasive carcinoma or DCIS involving adenosis, invasive cribriform car- In Situ and Invasive Carcinoma cinoma or cribriform DCIS, and invasive carcinoma The presence or absence of a myoepithelial cell or solid papillary carcinoma are select examples layer around carcinoma cells is the basis for divid- (Fig 1E and 1F). ing tumors into in situ and invasive types. In most Any single myoepithelial cell marker is imperfect cases, distinguishing in situ from invasive carcino- due to cross-reactivity with stromal myofibroblasts, ma is straightforward. However, myoepithelial cell vascular smooth muscle cells, or luminal epithelial markers can be useful in diagnosing microinvasion, cells. Therefore, at least 2 markers should be used particularly in the setting of extensive high-grade in combination. The most appropriate combinations

April 2015, Vol. 22, No. 2 Cancer Control 221 recommended in the literature have been a nuclear Microglandular adenosis is a benign lesion that stain (p63) together with a cytoplasmic stain, such as defies the general rule of benign glands being sur- smooth muscle myosin heavy chain, smooth muscle rounded by a layer of myoepithelial cells. Absence actin, or calponin.4,6,7 Among the cytoplasmic mark- of a myoepithelial cell layer surrounding the glands, ers, smooth muscle myosin heavy chain appears to together with their tendency for haphazard growth be more specific but is less sensitive than calponin. into adipose tissue, makes this entity prone to being P63 has the advantage of being a clean and easy-to- misdiagnosed as invasive carcinoma, specifically the read stain and it has good specificity rates in the set- tubular type.6 ting of breast cancer.8 However, its staining pattern Paying careful attention to subtle morphological can be discontinuous, particularly around distended features such as the commonly present, colloid-like ducts with carcinoma in situ. In addition, most met- secretions within the glandular lumina, as well as aplastic carcinomas and, rarely, ductal carcinomas positivity of the epithelial cells for S100 can help in stain positively for p63.4 Despite this cross-reactivity, making the correct diagnosis. Of note, these lesions even when breast carcinomas stain positive for p63, are negative for ER and progesterone receptor (PR), staining tends to be patchy and of less intensity than which would be unexpected in low-grade tubular car- adjacent myoepithelial cells. cinoma.6 This “unexpected” immunoprofile should Recent studies have also shown that qualitative raise questions in cases misdiagnosed as invasive differences may exist with regard to the myoepi- carcinoma. thelial cell markers expressed in different lesions; specifically, myoepithelial cells surrounding DCIS as Classification of Tumor Type well as benign sclerosing lesions have phenotypic Ductal and Lobular Carcinoma alterations compared with myoepithelial cells sur- Invasive and in situ carcinomas in the breast are di- rounding normal ducts and lobules.9,10 P63, smooth vided into 2 main groups: ductal and lobular car- muscle myosin heavy chain, and calponin all show cinomas. These 2 carcinoma types have major dif- reduced expression in these cases, potentially leading ferences in anatomical distributions, long-term risk to the erroneous diagnosis of invasive carcinoma. In stratifications, and metastatic patterns. Consequent- particular, the myoepithelial cells layer in small-sized ly, to perform an appropriate preoperative workup, biopsy specimens may not be present in the plane patient counseling, and treatment planning, an ac- of the section.8 In addition, expansion of the ducts curate diagnosis on CNB specimens is necessary.14-17 associated with DCIS could result in “attenuation” of IHC staining for E-cadherin, which is an adhe- the myoepithelial cell layer, and, in such a scenario, sion molecule expressed in carcinomas of the ductal reactivity with myoepithelial markers may be absent.7 type, has long been used by pathologists. Ductal car- Paying careful attention to the type of lesion in ques- cinomas show a strong membranous staining pattern, tion and using a panel that includes several myoep- whereas lobular carcinomas are either negative or ithelial cell markers could help avoid such potential rarely display an aberrant staining pattern (Fig 1G diagnostic errors. and 1H). A small subset of lobular carcinomas (in Myoepithelial cell markers have also been particular, the pleomorphic type) can be positive for used to classify papillary lesions. Benign papillary E-cadherin, and E-cadherin can show granular cyto- lesions are characterized by a continuous layer plasmic staining in some lobular carcinomas. This of myoepithelial cells in the fibrovascular cores is a potential pitfall in tumor classification because and at the periphery of the lesion. Although pap- the finding is sometimes misinterpreted as positive illary carcinomas lack myoepithelial cells inside staining. Careful attention to H & E findings as well the lesion, the hallmark of in situ carcinomas is as the location and the relative intensity of staining their presence at the periphery. Unusual entities, between tumor cells and benign ducts are all helpful such as solid papillary carcinoma and encapsu- in making the correct diagnosis. lated papillary carcinoma, can be devoid of “de- The combination of E-cadherin and p120 (a cat- monstrable” myoepithelial cells at the periphery. enin that binds to E-cadherin on the cell membrane Therefore, discussion regarding the nature of and is essential for the formation of tight junctions) these tumors has been taking place for some time, is superior to using E-cadherin alone when facing with some researchers noting that solid papillary ambiguous cases.4 Ductal carcinomas will have a carcinoma and encapsulated papillary carcinoma membranous pattern of staining, whereas lobular tend to behave as in situ carcinomas and, thus, carcinomas show cytoplasmic staining. Because duc- are classified as such.11-13 Paying attention to mor- tal and lobular carcinomas can be distinguished on phological features on H & E and interpreting IHC H & E alone in the majority of cases or with the use stains in the correct context are key to making of E-cadherin alone, we recommend using p120 for the correct diagnosis in such cases. difficult-to-diagnose cases alone.

222 Cancer Control April 2015, Vol. 22, No. 2 Special-Type Carcinomas and Spindle Cell Tumors Typically, special-type carcinomas are diagnosed based on their morphological features on H & E examination. However, occasionally, IHC stains can be helpful in this setting, particularly for spindle cell tumors (Fig 2A–C). Recognizing metaplastic carci- A B noma without an obvious epithelial or heterologous element is difficult — and sometimes impossible — on H & E alone, and this is especially true for limited samples such as CNB. The differential diagnosis of spindle cell metaplastic carcinoma includes other spindle cell lesions, such as phyllodes tumor, fibromatosis, angiosarcoma, and spindle cell melanoma, among others. Keratin expression in met- C D aplastic tumors can be heterogeneous and focal; therefore, a battery of CK stains (both pan-keratins and high-molecular-weight keratins), epithelial membrane antigen, p63, melanoma markers (eg, S100, melan A, HMB45), and additional markers to rule out other sarcomas (as needed) should be used in difficult cases.18-21 E F The identification of basal-like carcinomas is important in premenopausal young women because this type of tumor may be associated with hereditary breast and ovarian cancers and carries a poor prognosis.4 Most of basal-like carcinomas are triple negative (negative for ER, PR, and HER2), and variably express basal-type keratins (CK5/6, G H CK14, CK17), luminal-type keratin CK8/18, epider- Fig 2A–H. — (A) Spindle cell lesion without an evident epithelial compo- mal growth factor receptor, vimentin, and p53.18,19,22 nent (H & E, × 100). (B) Cytokeratin AE1/3 is positive in areas of the tumor This panel of immunostains can be used in clinical (× 100). (C) IHC for p63 reveals scattered positive cells (× 100). The combination of positive AE1/3 and p63 is confirmatory of spindle cell meta- practice and represents a less costly, more readily plastic carcinoma. (D) Metastatic, poorly differentiated adenocarcinoma available methodology in which to classify these to the liver. Without a positive clinical history, the differential diagnosis tumors when compared with other commercially includes lung, colon, breast, and hepatobiliary primaries, among others (H & E, × 100). (E) Positive immunostain for GATA-binding protein 3, together available assays. with the patient’s history of breast malignancy, is consistent with metastatic carcinoma from breast primary (× 50). (F) Immunostain for CK7 in this Paget Disease of the Nipple intestinal biopsy specimen from a patient with a history of lobular carcinoma highlights the presence of malignant cells involving the submucosa Most cases of Paget disease of the nipple are as- (× 100). (G) ER is strongly positive in this grade 1 invasive ductal carci- sociated with underlying breast carcinoma (main- noma (× 200). (H) HER2-positive (3+) staining on IHC is characterized by ly DCIS); therefore, the diagnosis can be made on strong and diffuse staining in this grade 3 invasive ductal carcinoma (× 100). CK = cytokeratin, ER = estrogen receptor, H & E = hematoxylin and eosin, H & E. However, biopsy specimens in patients without IHC = immunohistochemistry. a detectable underlying mass can be diagnostically challenging. The main differential diagnosis in such of this entity from nonmammary lesions. a rare setting includes melanoma and squamous cell Select pitfalls must be kept in mind when evalu- carcinoma (Bowen disease). ating CK7 immunostain. Both Toker and Merkel cells CK7 immunostain is positive in almost all cases may be positive for CK7.25,26 Moreover, intraepider- of Paget disease and is negative in squamous cell mal CK7-positive cells may represent an extension carcinoma. Typically, Paget cells express the same of benign lactiferous ducts into the nipple.4 Careful IHC characteristics of the underlying carcinoma.3 Be- attention to the distribution and volume of positive cause the underlying malignancy is frequently of a cells may help distinguishing these benign entities high-grade type, ER, PR negative, HER2-positive DCIS, from Paget disease. Melanoma can be ruled out by HER2 immunostain is useful and positive in up to 90% a panel of at least 2 immunostains, with the under- of Paget cases. Therefore, the use of CK7 and HER2 standing that S100 is sometimes positive in Paget dis- is considered the most appropriate combination for ease.27 Squamous cell carcinomas are negative for CK7 confirming the diagnosis.23,24 Conversely, ER and PR and positive for pan-keratin, high-molecular-weight immunostains are of no value in the differentiation keratins, and p63.

April 2015, Vol. 22, No. 2 Cancer Control 223 Metastatic Setting Prognostic/Predictive Immunohistochemical Although breast cancer commonly metastasizes to Stains other organs, metastases to the breast are rare and Hormone Receptors and HER2 Status account for 0.5% to 2.0% of all breast malignancies.28 Hormone receptor and HER2 testing should be per- In most cases, a history of malignancy is known. Mel- formed on all primary invasive breast carcinomas, as anoma is one of the most common tumors metasta- well as in recurrent or metastatic tumors. Multiple sizing to the breast, followed by carcinomas of lung preanalytic and analytic factors can affect test results; and those of gynecological origin.29 Absence of an in however, this review only briefly summarizes the an- situ component as well as clinical history of anoth- alytic component of these tests based on guidelines er malignancy should raise suspicion for metastasis from the American Society of Clinical Oncology and rather than primary breast carcinoma. The distinction the College of American Pathologists.34,35 is crucial because therapeutic management differs in All cases with at least 1% positive tumor cells these cases. for ER and PR have been associated with clinical A commonly encountered situation is metastatic response; therefore, they are classified as positive carcinoma of unknown origin to a nonbreast site. In (Fig 2G).34 The College of American Pathologists re- these cases, metastatic breast carcinoma is usually quires providing quantitative information regarding in the differential diagnosis (Fig 2D). Breast carci- the extent of positivity in all pathology reports.34 nomas are likely to be CK7 positive and CK20 neg- Quantification can be performed by simply providing ative. However, a similar profile is seen in lung and the proportion of positive cells or using the Allred gynecological tract carcinomas. Positivity for ER and score or H score systems, both of which use the in- PR can be helpful; however, these stains can also be tensity and percentage of positive cells.35 positive in other tumors — in particular, carcinomas Staining patterns/intensities of HER2 immunostain of a gynecological origin are positive for ER and PR have been categorized into 3 groups. Positive (3+) (in which case, Wilms tumor 1 and paired box 8 can staining is defined by complete, intense, circumfer- be helpful). When the differential diagnosis includes ential membrane staining (Fig 2H). Negative results lung adenocarcinoma, thyroid transcription factor 1 include cases categorized as 1+ characterized as in- and napsin-A, which are both positive in lung ade- complete membrane staining that is faint/barely visi- nocarcinomas, can be used.4 ble and in more than 10% of the invasive tumor cells, No single stain is unique to the breast. Gross and 0 characterized by no staining or membranous cystic disease fluid protein 15 (GCDFP15) is rela- staining that is incomplete and is faint/barely per- tively specific for the breast (also positive in sali- ceptible and within 10% or less of the invasive tumor vary gland tumors and adnexal tumors of the skin), cells.35 Equivocal cases (2+) are those that display but it lacks sensitivity and shows a patchy staining, circumferential membrane staining that is incomplete, which becomes problematic in small-sized biopsy weak/moderate, or both and within less than 10% of specimens. Mammaglobin is more sensitive than the invasive tumor cells or complete and circumfer- GCDFP15 but it lacks specificity. A combination of ential membrane staining that is intense and within the 2 stains may be superior to either one alone; 10% or less of invasive tumor cells.35 If results are however, even when used in combination, up to 30% equivocal, then reflex testing should be performed of tumors can be negative for both.30 Of note, neither using an alternate assay (ie, in situ hybridization if stain can be used to differentiate breast carcinoma IHC was used as the initial test).35 from adnexal tumors of the skin (sweat gland carcinomas) or those of salivary gland duct origin. To Conclusions distinguish between primary breast carcinoma and Immunohistochemical (IHC) stains provide informa- sweat gland carcinoma, a panel of IHC that includes tion and aid in the differential diagnosis of challeng- mammaglobin, GCDFP15, p63 (strong expression in ing epithelial lesions of the breast. In addition, IHC the latter group), and basal cytokeratins (CK5, CK14, can be used to obtain information relating to prog- and CK17) has been suggested as having highly sen- nostic/predictive markers crucial for the treatment sitivity and specificity rates.31 and prognostic assessment of patients with breast GATA-binding protein 3 (GATA3) IHC is highly carcinoma. sensitive and fairly specific for breast carcinoma if IHC stains should be used selectively and judi- urothelial carcinoma is not in the differential diagno- ciously and interpreted with the differential diagnoses sis. Rarely, staining for GATA3 has been reported in and pitfalls in mind. Hematoxylin and eosin continues endometrioid adenocarcinoma.32,33 Of note, the rate to be the most appropriate tool for diagnosing breast of positive staining in breast carcinomas is highest epithelial lesions. in well-differentiated, ER-positive tumors, including lobular carcinomas (Fig 2E and 2F).

224 Cancer Control April 2015, Vol. 22, No. 2 differential diagnosis from primary breast carcinoma. J Surg Oncol.1991; References 48(2):112-116. 1. Rosai J. Borderline epithelial lesions of the breast. Am J Surg Pathol. 29. Di Bonito L, Luchi M, Giarelli L, et al. Metastatic tumors to the female 1991;15(3):209-221. breast. An autopsy study of 12 cases. Pathol Res Pract. 1991;187(4):432-436. 2. Schnitt SJ, Connolly JL, Tavassoli FA, et al. Interobserver reproducibility 30. Bhargava R, Dabbs DJ. Diagnostic immunohistology of the breast. In: in the diagnosis of ductal proliferative breast lesions using standardized criteria. Dabbs DJ. Breast Pathology. Philadelphia: Elsevier Saunders; 2012:189-232. Am J Surg Pathol. 1992;16(12):1133-1143. 31. Rollins-Raval M, Chivukula M, Tseng GC, et al. An immunohistochem- 3. Rosen PP. Ductal hyperplasia: usual and atypical. In: Rosen PP, ed. ical panel to differentiate metastatic breast carcinoma to skin from primary Rosen’s Breast Pathology. 3rd ed. Philadelphia, PA: LWW; 2009:244-262. sweat gland carcinomas with a review of the literature. Arch Pathol Lab Med. 4. Bhargava R, Dabbs DJ. Use of immunohistochemistry in diagnosis of 2011;135(8):975-983. breast epithelial lesions. Adv Anat Pathol. 2007;14(2):93-107. 32. Liu H, Shi J, Wilkerson ML, et al. Immunohistochemical evaluation of 5. Moinfar F, Man YG, Lininger RA, et al. Use of keratin 34betaE12 GATA-3 expression in tumors and normal tissue; a useful biomarker for breast as an adjunct in the diagnosis of mammary intraepithelial neoplasia-duc- and urothelial carcinomas. Am J Clin Pathol. 2012;138(1):57-64. tal type-benign and malignant intraductal proliferations. Am J Surg Pathol. 33. Ciocca V, Daskalakis C, Ciocca RM, et al. The significance of GATA-3 in 1999;23(9):1048-1058. breast cancer: a 10-year follow-up study. Human Pathol. 2009;40(4):489-495. 6. Zhao L, Yang X, Khan A, et al. Diagnostic role of immunohistochem- 34. Fitzgibbons P, Dilon D, Alsabeh R, et al. Templates for reporting results istry in the evaluation of breast pathology specimens. Arch Pathol Lab Med. of biomarker testing of specimens from patients with carcinoma of the breast. 2014;138(1):16-24. Arch Pathol Lab Med. 2014;138(5):595-601. 7. Werling RW, Hwang H, Yaziji H, et al. Immunohistochemical distinc- 35. Wolff A, Hammond E, Hicks DG, et al; American Society of Clinical tion of invasive from noninvasive breast lesions: a comparative study of p63 Oncology; College of American Pathologists. Recommendations for human versus calponin and smooth muscle myosin heavy chain. Am J Surg Pathol. epidermal growth factor receptor 2 testing in breast cancer: clinical practice 2003;27(1):82-90. guidelines update. J Clin Oncol. 2013;31(31):3997-4013. 8. Lerwill MF. Current practical applications of diagnostic immunohistochemistry in breast pathology. Am J Surg Pathol. 2004;28(8):1076-1091. 9. Hilston JB, Schnitt SJ, Collins LC. Phenotypic alterations in ductal carcinoma in situ-associated myoepithelial cells: biologic and diagnostic implications. Am J Surg Pathol. 2009;33(2):227-232. 10. Hilston JB, Schnitt SJ, Collins LC. Phenotypic alteration in myoepithelial cells associated with benign sclerosing lesions of the breast. Am J Surg Pathol. 2010;34(6):896-900. 11. Saremian J, Rosa M. Solid papillary carcinoma of the breast: a pathologically and clinically distinct breast tumor. Arch Pathol Lab Med. 2012;136(10):1308-1311. 12. Collins LC, Carlo VP, Hwang H, et al. Intracystic papillary carcinomas of the breast: a reevaluation using a panel of myoepithelial cell markers. Am J Surg Pathol. 2006;30(8):1002-1007. 13. Solorzano CC, Middleton LP, Hunt KK, et al. Treatment and outcome of patients with intracystic papillary carcinoma of the breast. Am J Surg. 2002;184(4):364-368. 14. Cristofanili M, Gonzalez-Angulo A, Sneige N, et al. Invasive lobular carcinoma classic type: response to primary chemotherapy and survival outcomes. J Clin Oncol. 2005;23(1):41-48. 15. Tubiana-Hulin M, Stevens D, Lasry S, et al. Response to neoadjuvant chemotherapy in lobular and ductal breast carcinomas: a retrospective study on 860 patients from one institution. Ann Oncol. 2006;17(8):1228-1233. 16. Cocquyt VF, Blondeel PN, Depypere HT, et al. Different response to preoperative chemotherapy for invasive lobular and invasive ductal breast carcinoma. Euro J Surg Oncol. 2003;29(4):361-367. 17. Rosa M, Mohammadi A, Masood S. Lobular neoplasia displaying central necrosis: a potential diagnostic pitfall. Pathol Res Pract. 2010;206(8):544-549. 18. Dunne B, Lee A, Pinder SE, et al. An immunohistochemical study of metaplastic spindle cell carcinoma, phyllodes tumor and fibromatosis of the breast. Hum Pathol. 2003;34(10):1009-1015. 19. Leibl S, Gogg-Kammerer, Sommersacher A, et al. Metaplastic breast carcinoma: are they of myoepithelial differentiation?: immunohistochemical profile of the sarcomatoid subtype using novel myoepithelial markers. Am J Surg Pathol. 2005;29(3):347-353. 20. Carter MR, Hornick JL, Lester S, et al. Spindle cell (sarcomatoid) carcinoma of the breast: a clinicopathologic and immunohistochemical analysis of 29 cases. Am J Pathol. 2006;30(3):300-309. 21. Adem C, Reynolds C, Adlakha H, et al. Wide spectrum screening keratin as a marker of metaplastic spindle cell carcinoma of the breast: an immunohistochemical study of 24 patients. Histopathol. 2002;40(6):556-562. 22. Livasy CA, Karaca G, Nanda R, et al. Phenotypic evaluation of the basal-like subtype of invasive breast carcinoma. Mod Pathol. 2006;19(2): 264-271. 23. Anderson JM, Ariga R, Govl H, et al. Assessment of Her-2/Neu status by immunohistochemistry and fluorescence in situ hybridization in mammary Paget disease and underlying carcinoma. Appl Immunohistochem Mol Mor- phol. 2003;11(2):120-124. 24. Lammie GA, Barnes DM, Millis RR, et al. An immunohistochemical study of the presence of c-erbB-2 protein in Paget’s disease of the nipple. Histopathology. 1989;15(5):505-514. 25. Lundquist K, Kohler S, Rouse RV. Intraepidermal cytokeratin 7 expression is not restricted to Paget cells but is also seen in Toker cells and Merkel cells. Am J Surg Pathol. 1999;23(2):212-219. 26. Di Tommaso L, Franchi G, Destro A, et al. Toker cells of the breast. Morphological and immunohistochemical characterization of 40 cases. Hum Pathol. 2008;39(9):1295-1300. 27. Gillet CE, Bobrow LG, Milis RR. S-100 protein in human mammary tissue--immunoreactivity in breast carcinoma, including Paget’s disease of the nipple, and the value as a marker of myoepithelial cells. J Pathol. 1990;160(1):19-24. 28. Vergier B, Trojani M, de Mascarel I, et al. Metastases to the breast:

April 2015, Vol. 22, No. 2 Cancer Control 225 Reliable predictive assays will provide

information to help inform clinical

treatment decisions.

Past Flights. Photograph courtesy of Craig Damlo. www.soapboxrocket.com.

Special Technologies for Ex Vivo Analysis of Cancer Jenny M. Kreahling, PhD, and Soner Altiok, MD, PhD

Background: Predictive assays for cancer treatment are not new technology, but they have failed to meet the criteria necessary for standardized use in clinical decision-making. Methods: The authors summarize the use of predictive assays and the challenges and values associated with these assays in the clinical setting. Results: Predictive assays commercially available in the clinical setting are not standardized, have significant obstacles to overcome, and cannot be relied upon by health care professionals due to the limited value these assays provide to the decision-making process for the treatment of patients. Conclusions: A method that more closely recapitulates the human tumor microenvironment and accurately predicts response with high reproducibility would be beneficial to patient outcomes and quality of life.

Introduction events endured by patients. Cancer is the second leading cause of death in the The treatment of cancer currently relies on clas- United States, despite the many chemotherapy agents sical chemotherapy regimens based on histology and and hundreds of combinations available for treat- targeted agents for patients who carry specific genom- ment.1 However, health care professionals generally ic alterations.1 In the era of genomics, we are learning make treatment decisions based on standard protocols that cancer is an individual disease and, with a push developed in clinical trials to successfully match a toward personalized medicine, treatment is shifting to patient to a treatment regimen.2 This generalized ap- a more tailored approach. However, genomic testing proach can lead to a response, but oftentimes patients still fails to capture the factors that ultimately deter- fail to show improvement. A predictive assay to help mine how tumor cells will behave inside the body. guide treatment decisions has the potential to lead to One of these factors is intratumoral heterogeneity. improved outcomes and fewer unnecessary adverse Large-scale sequencing of solid cancers has revealed that different cells within a tumor can have distinct profiles, including in gene expression, proliferation, From the Departments of Experimental Therapeutics (JMK) and Anatomic Pathology (SA), H. Lee Moffitt Cancer Center & Research and metastatic potential, among others. These differ- Institute, Tampa, Florida. ences can contribute to treatment failures and have Submitted December 15, 2014; accepted April 1, 2015. consequences related to personalized medicine.3 The Address correspondence to Soner Altiok, MD, PhD, Department of impact on predictive assays lies in the sampling of Anatomic Pathology, Moffitt Cancer Center, 12902 Magnolia Drive, tumors that commonly rely on biopsy specimens or Tampa, FL 33612. E-mail: [email protected] small resections that do not fully represent the com- No significant relationships exist between the authors and the companies/organizations whose products or services may be plexity of the tumor, thus resulting in the failure of referenced in this article. drugs selected through personalized screening based

226 Cancer Control April 2015, Vol. 22, No. 2 on a small population of biologically identical cells.4-6 several fold below physiological relevance to several The information gained from studying tumor hetero- fold above physiological relevance. Some cell viability geneity will be invaluable in optimizing predictive assays exist to examine chemosensitivity or chemo- assays, such as serial biopsies, to gain a broader pic- resistance; however, few are commercially available ture of individual tumor response to drug therapy.4 and periodically used in the clinic (Table).13 Functional predictive assays providing informa- The differential staining cytotoxicity assay in- tion on personalized responses to drugs are need- volves mechanical disaggregation of cells from sur- ed to help guide treatment decisions and improve gical or biopsy specimens.14 Three-dimensional fresh outcomes. Chemosensitivity assays offer potential in tumor cell clusters are cultured in anchorage-indepen- predicting treatment response; however, after more dent conditions and treated with the drugs of interest than 20 years, controversy still exists on the predictive at 3 dose levels; the middle dose is that which could value of such assays.7 be achieved in therapy: 10-fold lower than the phys- iologically relevant dose and 10-fold higher from 4 to Assay-Guided Therapy 6 days. At the conclusion of the culture period, fast Chemosensitivity assays are described as any in vitro green dye is added to the microwells, the contents are laboratory analysis that tests whether tumor growth then sedimented onto permanent Cytospin (Thermo has been inhibited by a known chemotherapy drug or Fisher Scientific, Waltham, Massachusetts) centrifuge a panel of drugs, including both classical and targeted slides, and then they are counterstained with hema- agents and their combinations. All chemosensitivity toxylin and eosin (H & E). The dead cells take up assays share the same basic steps, including isolating the fast green dye, and H & E allows the tumor cells tumor cells, incubating cells with drugs, assessing to be differentiated from normal cells. Rates of drug cell growth or survival, and interpreting the results. sensitivity are measured by the ratio of live cells in These assays are often referred to as chemore- the treated samples to the number of live cells in the sistance assays because ineffective drugs (resistant to untreated controls.8,13,14 This assay is capable of mea- treatment) are identified with significantly more accu- suring both apoptotic- and nonapoptotic-mediated cell racy more frequently than effective drugs (sensitive to death in a population of cells; in addition, the assay treatment).8 This leads to questions regarding distin- can be applied to solid and hematological neoplasms, guishing drug-resistance assays from chemosensitivity does not require a pure population, and can be used assays and the reliability of these types of assays.2,9 with a wide variety of drugs.15,16 For any type of available chemopredictive assay, The ex vivo analysis of programmed cell death cancer tissue samples are obtained from biopsy speci- assay also measures apoptotic and nonapoptotic cell mens or during surgery. These samples are then disso- death markers in tumor samples exposed to che- ciated, grown in a laboratory, and exposed to a range motherapeutic agents. Tumor specimens obtained of concentrations of chemotherapeutic drugs, usually through biopsy specimens or surgical resections are chosen by the ordering physician and regularly take at least 3 weeks. The intent is that cancer cells treated Table. — Commercially Available Assays in the laboratory will reflect the same response in the patient; however, these results are often ambiguous Assay Method at best because the cancer cells show unexpected be- Differential Drug sensitivity measured by the ratio of live cells havior, such as resistance to the drugs typically used staining in the treated samples to number of live cells in cytotoxicity untreated controls to treat that particular cancer and sensitivity to drugs 10,11 Ex vivo analysis End point of interest is cell death as assessed by not generally used. These and other obstacles have of programmed number of cells differentially stained due to slowed the use of chemopredictive assays in routine cell death changes in the integrity of the cellular membrane 7 clinical testing. 3H-thymidine Measures uptake of tritiated thymidine by DNA of viable cells Cell Viability Assays Histoculture drug Drug sensitivity evaluated by quantification of cell Tissue culture–based chemopredictive assays have resistance growth in the 3-dimensional collagen matrix. An inverse relationship exists between drug sensitivi- been used since chemotherapy was first used in can- ty of the tumor and cell growth cer treatment. Growth inhibition or cell death has ATP Measures ATP to quantify the viable cells in a been used in previous iterations of assays of sensi- bioluminescence culture using a luminometer. Decrease in ATP tivity to conventional chemotherapeutic agents.12 All indicates drug sensitivity, whereas no loss of ATP assays use characteristics of cell physiology to distin- suggests drug resistance guish between viable and nonviable cells to quantify Chemosensitivity Utilizing ATP quantification with a luminometer, cells are grown in a monolayer rather than a cell death following exposure to a particular drug of 3-dimensional matrix interest. Drug doses used in the assays are variable, but all assays require drug exposures ranging from ATP = adenosine triphosphate.

April 2015, Vol. 22, No. 2 Cancer Control 227 disaggregated using DNAse and type 4 collagenase to 2 mm in size placed on a collagen matrix so the frag- yield tumor clusters of 50 to 100 cell spheroids. These ments can grow 3-dimensionally and maintain their microaggregates are thought to more closely approx- signaling pathways. After 24 hours, explants are incu- imate the human tumor microenvironment because bated with a particular drug for 3 days. Subsequently, they are not proliferated. Aggregates are then treated they are fixed in formalin and embedded in paraf- with dilutions of drugs and incubated for 3 days. After fin. Radioactivity is quantified in slide sections using drug exposure is completed, a mixture of nigrosin B autoradiography.13 Extreme drug resistance assays and fast green dye with glutaraldehyde-fixed avian have been shown to have a high negative predictive erythrocytes is added to the cellular suspensions.17 value for identifying ineffective drugs. In this way, The samples are then agitated and Cytospin-centri- by process of elimination, results of chemoresistance fuged and then counterstained with H & E. The end dictate the positive selection of potentially effective point of interest for this assay is cell death, which is drugs. However, avoiding potentially unnecessary tox- assessed by observing the number of cells differen- icity is not a straightforward outcome and, although tially stained due to changes in the integrity of the the eliminated drug will not be used, health care pro- cellular membrane.13,17,18 According to the results of a fessionals may be faced with making a choice to sub- single phase 2 clinical trial, functional profiling using stitute it with another drug. This convoluted outcome ex vivo analysis of programmed cell death doubles the suggests that these assays may still be inadequate response rate and improves time-to-progression and for routine oncology care until further randomized survival rates in patients with advanced lung cancer.17 controlled testing is performed.13,18 Several methods measure the incorporation of Drug sensitivity assays are evaluated by quantify- radioactive precursors by macromolecules in viable ing cell growth in the 3-dimensional collagen matrix cells. Incorporating tritiated thymine measures the evaluated by an inverse relationship between the drug uptake of tritiated thymidine by the DNA of viable sensitivity of the tumor and cell growth. However, cells.19 Using proteases and DNAse to disaggregate the concentrations of drug and incubation times are not tissue, samples are seeded into single-cell suspension standardized and vary depending on drug combi- cultures on soft agar and treated with a particular drug nation and tumor type. The adenosine triphosphate for 4 days and, after 3 days, tritiated thymidine is add- (ATP) bioluminescence assay relies on measuring the ed. After 24 hours of additional incubation, cells are level of ATP after single cells or small aggregates are lysed, and radioactivity is quantified and compared cultured and then exposed to particular drugs. Fol- with a blank control. Only viable and proliferating lowing incubation with the drug, the cells are lysed, cells will take up the radioactive thymidine.29 There- cytoplasmic components are solubilized, and then fore, an inverse relationship exists between the uptake luciferin and firefly luciferase are added to the cell of radioactivity and the sensitivity rate of the cells and lysis product, catalyzing the conversion of ATP to the agent of interest.13 adenosine diphosphate and monophosphate and light The extreme drug resistance assay is method- is proportionally emitted to metabolic activity and is ologically similar to the thymidine incorporation as- quantified with a luminometer.13 The number of cells say.20 By incorporating metabolic tritiated thymidine can be calculated from the measurement of light. A to measure cell viability, small tissue samples, rather decrease in ATP indicates drug sensitivity, whereas no than single cell suspensions, are incubated with a loss of ATP suggests that the tumor is resistant to the particular drug for 5 days at doses ranging from 5-fold agent of interest. ChemoFX (Precision Therapeutics, below to 80-fold above concentrations that would Pittsburgh, Pennsylvania) uses this type of technology; reflect physiological relevance. Tritiated thymidine is however, cells are grown in a monolayer rather than subsequently added to the culture, and the uptake rate a 3-dimensional matrix.18 is quantified after various incubation times. Only live The rationale for chemosensitivity assays is stron- (resistant) cells will incorporate tritiated thymidine, gest where a variety of therapeutic options exists but and the level of incorporation is directly related to the no clear selection criteria have been chosen for any rate of chemoresistance, rather than responsiveness particular drug regimen in an individual patient.13 (chemosensitivity). Extreme resistance is identified Some studies have reported a correlation between when the incorporation of thymidine is inhibited in in vitro prediction or response and clinical response, the presence of the particular drug by less than 1 stan- and, although these studies may have internal validi- dard deviation of the median cell inhibition measured ty, they cannot determine whether patients given as- for several hundred reference tumor samples. In this say-guided therapy or empirical therapy have different test, a positive resistance to a specific drug would outcomes.18,22-26 To determine whether assay-guided prevent the selection of an “ineffective” drug.13,18,21 treatment results in overall different outcomes than Another resistance assay is the histoculture empirical treatment, it is important to take into ac- drug-resistance assay that tests tissue fragments of 1 to count response rates, survival rates, the presence of

228 Cancer Control April 2015, Vol. 22, No. 2 adverse events, and patient quality of life. Chemosen- data had not yet been published.2,31 Previously, reports sitivity and chemoresistance assays are used in some in soft-tissue sarcoma xenograft models showed low centers for decisions related to future chemotherapy responses to conventional agents and were unable in situations with multiple equivalent treatment op- to find a strong correlation between selected genetic tions available and when “the current level of evi- markers and chemotherapeutic effect.32 Furthermore, dence is not sufficient to supplant current standard xenograft models of non–small-cell lung cancer were of care chemotherapy.”27 A need exists to further val- treated with 3 hemotherapeutic combinations.33 A 90% idate these assays with direct evidence gathered from engraftment rate was achieved, and approximately prospective trials comparing individuals empirically one-half of the tumors were tested against all 3 che- treated with those given assay-directed therapy. In motherapeutic regimens.33 The xenografts histologi- this way, response rates, survival rates, the incidence cally resembled the primary tumor. Nearly one-third of adverse events, and patient quality of life can all of the tumors were not sensitive to any of the tested be taken into consideration.13,18,21,28-30 combinations.33 In addition, a series of 11 patients received one of the tested combinations (vinorelbine Concerns and cisplatin), and 7 of the patients had recurrences, Concern exists as to whether disaggregated tumor 6 of whom had corresponding nonresponsive xeno- cells in culture represent the 3-dimensional structure grafts.33 One xenograft was sensitive to the combi- of an intact tumor, including the tumor microenviron- nation and also had a recurrence. Results from this ment, and their effect on drug treatment.2 Further- assay took 6 to 8 weeks.33 The authors concluded that more, cells exposed to a particular drug in a culture a correlation was present between the assay results do not take into account the pharmacokinetics of and clinical data in the recurrence group and that drug metabolism and delivery in a human being. The xenograft-based testing may help identify new active reliability and consistency of chemopredictive test- combinations in non–small-cell lung cancer.33 ing are also concerning. Many studies have produced In another study, 29 advance sarcoma xenografts conflicting results, some of which have suggested a were collected and 76% were engrafted.31 Of these 22, small benefit in assay-directed treatments, whereas a total of 16 were correlated with patient outcomes, others have demonstrated no difference at all.12 Spe- 13 of which had a positive correlation, including cific problems with previous assays include the low 6 tumors prospectively tested and 7 tumors tested percentage of successfully completed assays that yield after the study volunteer had already received chemo- clinically useful results, with rare interpretations that therapy.31 According to the authors, all 7 prospectively differed from standard of care, technical complexity determined regimens, including 2 novel, nonstandard that made their application beyond a single labora- approaches, resulted in disease control or response tory/institution unlikely, and the long time needed in the xenograft system and were matched with clin- to study completion, thus delaying therapy. Reviews ical benefit to the 5 individuals who received these of the medical literature, government studies, and agents.31 Although this study had a small data set, its health care insurance evaluations concluded that the results are suggestive of a positive predictive value, evidence is insufficient to support the use of che- and further investigation might identify tumor- and mopredictive assays in routine clinical practice and patient-specific treatments.2,31 that such assays are not recommended, nor covered, by most insurance companies.18,21,28-30 Still, support Conclusions continues for a predictive assay in the context of a The pursuit of reliable predictive assays is an im- clinical trial. Proposals for a successful assay include portant task, and the information these assays could standardizing materials, testing a range of concen- provide would help inform clinical treatment deci- trations to provide a dose-response curve, applying sions. However, after many years, these assays are simple techniques with the possibility of automation, still experimental and cannot be recommended for and measuring cell survival using clear interpretation routine clinical use.7 Many still rely on traditional techniques. approaches of dissociating tumors and establishing cell lines maintained in vitro in serum-based growth Patient-Derived Xenograft Models media. Cultured cells are not completely representa- An advance over cell culture assay systems is the tive of the parent tumor and such differences are of patient-derived mouse xenograft model that allows concern when predicting drug response as well as to for the recapitulation of 3-dimensional tumor archi- the basic study of cancer. In particular, culture selec- tecture and incorporates aspects of tumor stroma.12 tion in cell lines may disturb the in vitro relationship Two major limitations of animal studies are their cost between the cancer stem cell and its progeny, and it and intensity of labor required. Until recently, a com- also removes tumor–stromal interactions essential to parison of xenograft responses with clinical outcome the 3-dimensional biology of solid tumors in vivo.12

April 2015, Vol. 22, No. 2 Cancer Control 229 The field is lacking studies published with ran- 2014;30C:1-6. 5. Gerlinger M, Swanton C. How Darwinian models inform therapeu- domized, prospective designs to evaluate the clinical tic failure initiated by clonal heterogeneity in cancer medicine. Br J Cancer. utility of chemoresistance and chemosensitivity as- 2010;103(8):1139-1143. 6. Gillies RJ, Verduzco D, Gatenby RA. Evolutionary dynamics of car- says. The data are insufficient to determine whether cinogenesis and why targeted therapy does not work. Nat Rev Cancer. use of a particular test to select chemotherapy regi- 2012;12(7):487-493. 7. Jenks S. Chemosensitivity assays: still eyeing the clinic. J Natl Cancer mens for individual patients will improve outcomes. Inst. 2012;104(23):1775-1777. Limitations exist and include sampling bias due to 8. Weisenthal LM, Kern DH. Prediction of drug resistance in cancer chemotherapy: the Kern and DiSC assays. Oncology (Williston Park). 1991;5(9):93- heterogeneity of tumors and insufficient biospecimen 104, 111-114, 117, 118. processing resulting in nonevaluable data; therefore, 9. Fruehauf JP, Alberts DS. In vitro drug resistance versus chemosensitivity: two sides of different coins. J Clin Oncol. 2005;23(15):3641-3643. all chemosensitivity assays are currently considered 10. Health Net. National medical policy. Chemotherapy sensitivity and re- to be investigational.13,18 sistance assays. Policy No. NMP210. Published March 2005. Updated March 2015. https://www.healthnet.com/static/general/unprotected/pdfs/national/policies The foremost priority in clinical oncology is pro- /ChemotherapySensitivityandResistanceAssays.pdf. Accessed April 17, 2015. viding personalized treatment approaches for more 11. Health Net. National medical policy. ChemoFX assay for prediction of response to chemotherapy. Policy No. NMP459. Published June 2009. Up- effective and better-tolerated therapies. One major dated May 2014. https://www.healthnet.com/static/general/unprotected/pdfs step toward offering personalized therapy in cancer /national/policies/ChemoFXAssayforPredictionofResponsetoChemotherapy. pdf. Accessed April 17, 2015. is the identification of actionable mutations so that 12. Reed D, Altiok S. Metastatic soft tissue sarcoma chemotherapy: an health care professionals can select the most appro- opportunity for personalized medicine. Cancer Control. 2011;18(3):188-195. 13. The Regence Group; Blue Cross/Blue Shield of Oregon, Utah, Ida- priate treatment for their patients. New technologies, ho, and Select Counties of Washington. Medical policy manual: in vitro che- such as next-generation sequencing (NGS), offer a moresistance and chemosensitivity assays. Policy No. 6. Published January 1996. Updated March 2014. http://blue.regence.com/trgmedpol/lab/lab06.pdf. comprehensive analysis of specific mutations present Accessed April 17, 2015. in the cancer of each individual patient and are like- 14. Mason JM, Drummond MF, Bosanquet AG, et al. The DiSC assay. A cost-effective guide to treatment for chronic lymphocytic leukemia? Int J ly to yield information regarding tumor biology and Technol Assess Health Care. 1999;15(1):173-184. response to various treatment modalities. However, 15. Solé RV, Valverde S, Rodriguez-Caso C, et al. Can a minimal rep- licating construct be identified as the embodiment of cancer? Bioessays. mutations identified through NGS analysis may not 2014;36(5):503-512. be actionable driver mutations, in that they do not 16. Weisenthal LM. Differential staining cytotoxicity assay: a review. Meth- ods Mol Biol. 2011;731:259-283. lend themselves to specific and effective therapies. 17. Nagourney RA, Blitzer JB, Shuman RL, et al. Functional profiling to Rather, we propose that combining ex vivo drug select chemotherapy in untreated, advanced or metastatic non-small cell lung cancer. Anticancer Res. 2012;32(10):4453-4460. testing with NGS analysis may help distinguish ran- 18. Aetna. Clinical policy bulletin: tumor chemosensitivity assays. No. 0245. dom passenger mutations from true drivers. Fresh Published May 26, 1998. Updated May 12, 2014. http://www.aetna.com/cpb /medical/data/200_299/0245.html Accessed April 1, 2015. tumor samples can be directly treated ex vivo with 19. Yung WK. In vitro chemosensitivity testing and its clinical application a particular drug or drug combination for the tar- in human gliomas. Neurosurg Rev. 1989;12(3):197-203. 20. Kern DH, Weisenthal LM. Highly specific prediction of antineoplastic geted blockade of signaling pathways aberrantly ac- drug resistance with an in vitro assay using suprapharmacologic drug expo- tivated through specific mutations in each cancer. sures. J Natl Cancer Inst. 1990;82(7):582-588. 21. BlueCross BlueShield of North Carolina. Corporate medical policy: in The drug-mediated inhibition of proximal and distal vitro chemoresistance and chemosensitivity assays. Published March 1997. signaling pathways accompanied by decreased cell Updated April 2014. https://wwwps.bcbsnc.com/assets/services/public/pdfs /medicalpolicy/in_vitro_chemoresistance_and_chemosensitivity_assays.pdf. viability would indicate a therapeutic opportuni- Accessed April 9, 2015. ty with a specific drug for each individual patient. 22. Gazdar AF, Steinberg SM, Russell EK, et al. Correlation of in vitro drug sensitivity testing results with response to chemotherapy and survival A similar approach could also be used as an enrich- in extensive stage small cell lung cancer: a prospective clinical trial. J Natl ment strategy in designing clinical trials in which Cancer Inst. 1990;82(2):117-124. 23. Agiostratidou G, Sgouros I, Galani E, et al. Correlation of in vitro cy- the tumor samples of study volunteers are tested ex totoxicity nd clinical response to chemotherapy in ovarian and breast cancer vivo for their responses to drugs used in the clinical patients. Anticancer Res. 2001;21(1A):455-459. 24. Konecny G, Crohns C, Pegram M, et al. Correlation of drug response trials and then volunteers are assigned to clinical trials with the ATP tumor chemosensitivity assay in primary FIGO stage III ovarian based on their ex vivo response to treatments. It is cancer. Gynecol Oncol. 2000;77(2):258-263. 25. Salom E, Penalver M, Homesley H, et al. Correlation of pretreatment plausible that ex vivo functional assays may serve as drug induced apoptosis in ovarian cancer cells with patient survival and clinical the basis for inclusion or exclusion for clinical trials response. J Transl Med. 2012;10:162. 26. Strickland SA, Raptis A, Hallquist A, et al. Correlation of the microculture-ki- that can incorporate targeted agents in addition to netic drug-induced apoptosis assay with patient outcomes in initial treatment of cytotoxic therapy.12 adult acute myelocytic leukemia. Leuk Lymphoma. 2013;54(3):528-534. 27. Bhagwandin S, Naffouje S, Salti G. Utility of chemoresponse assay in patients undergoing cytoreductive surgery plus hyperthermic intraperitoneal References chemotherapy. Ann Surg Oncol. 2015. [Epub ahead of print]. 28. Burstein HJ, Mangu PB, Somerfield MR, et al. American Society of 1. American Cancer Society. Cancer Facts & Figures 2015. Atlanta, GA: Clinical Oncology clinical practice guideline update on the use of chemotherapy American Cancer Society; 2015. http://www.cancer.org/acs/groups/content/@ sensitivity and resistance assays. J Clin Oncol. 2011;29(24):3328-3330. editorial/documents/document/acspc-044552.pdf. Accessed April 9, 2015. 29. Schrag D, Garewal HS, Burstein HJ, et al. American Society of Clinical 2. Wagner AJ. Patient-derived sarcoma xenografts for individual-patient se- Oncology technology assessment: chemotherapy sensitivity and resistance lection of chemotherapy-ready for prime time? Cancer. 2014;120(13):1917-1919. assays. J Clin Oncol. 2004;22(17):3631-3638. 3. Gerlinger M, Rowan AJ, Horswell S, et al. Intratumor heterogeneity 30. Laboratory and Diagnostic Services Panel. Human tumor assay and branched evolution revealed by multiregion sequencing. N Engl J Med. systems. Presented at the Medicare Evidence Development & Coverage 2012;366(10):883-892. Advisory Committee Meeting, November 15–16, 1999; Baltimore, MD. 4. Barber LJ, Davies MN, Gerlinger M. Dissecting cancer evolution at the http://www.cms.gov/medicare-coverage-database/details/medcac-meeting macro-heterogeneity and micro-heterogeneity scale. Curr Opin Genet Dev. -details.aspx?MEDCACId=14&year=1999&bc=AAAIAAAAAAAAAA%3d%3d&.

230 Cancer Control April 2015, Vol. 22, No. 2 Accessed April 9, 2015. 31. Stebbing J, Paz K, Schwartz GK, et al. Patient-derived xenografts for individualized care in advanced sarcoma. Cancer. 2014;120(13):2006-2015. 32. Ryan CW, Dolan ME, Brockstein BB, et al. A phase II trial of O6-ben- zylguanine and carmustine in patients with advanced soft tissue sarcoma. Cancer Chemother Pharmacol. 2006;58(5):634-639. 33. Dong X, Guan J, English JC, et al. Patient-derived first generation xenografts of non-small cell lung cancers: promising tools for predicting drug responses for personalized chemotherapy. Clin Cancer Res. 2010;16(5):1442-1451.

April 2015, Vol. 22, No. 2 Cancer Control 231 Editorial

Should We Preclude Phase 1 Clinical Trials From Enrolling Elderly Individuals?

A review by Mahipal and colleagues of phase 1 risks of potential drug interactions were present, even trial enrollment among patients 65 years and old- when such an interaction was not associated with er at the H. Lee Moffitt Cancer Center & Research chemotherapy agents. Drug interactions can typically Institute in Tampa, Florida, appears in this issue of be expected in any patient who takes at least 8 med- Cancer Control. The study provides a recent and ications every day.9 Another limitation is the absence extensive — albeit not exclusive — exploration of the of data related to phase 1 studies of targeted thera- subject, encompassing 39 clinical trials (31 no-transplan- py,10 which is posed to be the most promising form tation and 8 transplantation studies) conducted within of cancer treatment in the next decade. Yet another a 10-year period involving 1,162 study participants, limitation is that the authors stopped their review 380 of whom were between the ages of 65 and at the year 2007 because they wanted to limit their 91 years. analysis to studies involving cytotoxic chemotherapy. In their research, Mahipal and coauthors Some readers may question whether these old data confirmed previous findings, including that: are still relevant to the evolving landscape of cancer • The mean age of individuals involved in phase treatment. 1 trials is approximately 10 years younger than Despite these limitations, the Cancer Control the mean age of the general cancer population1 editors elected to publish this study for 3 reasons: • Age between 65 and 75 years does not appear (1) reaffirm the importance of physiological — rath- to affect the effectiveness or toxicity of drugs er than chronological — age for the enrollment of used in the no-transplantation setting2-5 older individuals in clinical trials of cancer chemo- In addition, they reported that the risk of toxic- therapy, (2) discuss which provisions would render ity increases with age in the transplantation setting. more meaningful results from clinical trials with el- This finding is new but not unexpected because derly participants, and (3) address the discussion as age is a recognized risk factor for toxicity and to whether a number of positions in phase 1 trials treatment-related mortality in patients receiving bone should be reserved for older individuals. marrow transplantation. Undoubtedly, targeted therapy has produced The results of the review by Mahipal and col- significant improvement in the prognoses of numer- leagues are robust because they are based on the ous malignancies,10 but predicting the demise of cy- experience of 1 institution, which keeps careful- totoxic chemotherapy would be premature because ly recorded data relating to outcomes and toxici- this treatment modality currently represents the only ties. The Cancer Control editors decided to publish curative treatment for germ cell tumors, some forms this contribution as a reminder that chronological of lymphoma and leukemia, and a number of breast, age — at least up to the age of 75 years — should colorectal, and pulmonary cancers. It is reasonable to not represent an impediment for patients to enroll in expect that cytotoxic agents will be the main bulwark phase 1 clinical trials. against the spread of cancers with complex genomic Like all retrospective analyses of data, the review alterations that may vary from cell to cell.11 Thus, of Mahipal and coauthors has several limitations, many phase 1 trials of cytotoxic chemotherapy agents are of which the authors have acknowledged, including still relevant to individuals of all ages. When these an inadequate assessment of age-related prognostic trials are conducted in patients of advancing age, a factors (eg, function, polymorbidity) and, more gen- wealth of desirable information may help contextu- erally, an under-representation of the older popula- alize the effects of a new drug. tion.6,7 In addition, the authors did not include study All clinical cancer trials in the elderly, including participants enrolled in cooperative phase 1/2 trials. phase 1 trials, should provide an estimate of physio- However, an additional limitation not mentioned logical age (ie, life expectancy and tolerance of stress). was a failure to account for polypharmacy that, by This estimate may be obtained from a comprehen- itself, could represent a risk factor for chemothera- sive geriatric assessment validated for this purpose.12 py-related toxicity. Popa et al8 reported that the risk of A reconciled list of medications may be important so chemotherapy-related toxicity was increased when the that toxicities and unsuspected drug interactions can

232 Cancer Control April 2015, Vol. 22, No. 2 be accounted for and documented. Although the com- not be excluded from phase 1 trials; however, at the prehensive geriatric assessment currently provides same time, they should not receive special treatment. the most reliable assessment of physiological age, The goal of a phase 1 trial is to establish the maximum some laboratory assays may help such determination tolerated dose of a drug, and that is best accomplished in the near future. Among the most promising are without having to accommodate a special population the so-called inflammatory index and the leukocyte such as the elderly. Thus, I would recommend that telomeres length.13-15 researchers reserve an adequate number of enrollment One may ask whether second or third malignan- positions to individuals 70 years of age or older for cies should exclude patients from enrolling in a phase phase 2 trials to establish how age may affect the 1 clinical trial. In approximately 20% of patients with pharmacology of a drug and whether a dose reduction cancer 70 years of age or older, a second primary may obtain similar results with less toxicity. One may malignancy will be present.16 In the majority of cases, argue that reserving a number of positions in phase the second neoplasm may be indolent (eg, localized 1 trials to older individuals, to those with a poor per- prostate cancer, low-grade lymphoma) and may not formance status, or those with significant comorbidity compromise patient survival for several years. There- may produce a plethora of adverse events among the fore, in my opinion, it is difficult to justify excluding different populations and, thus, prevent the drug from these patients from enrolling in either phase 1 or misuse in clinical practice. I believe that the terms of 2 clinical trials. As the US population continues to the controversy should be exposed, but I doubt they age, one may consider a change in the recruitment can be solved in this context. criteria for clinical trials so that they more realistically From time to time, the editors of Cancer Con- represent the demographics of patients with cancer. trol publish studies with methodological flaws when It is also desirable to examine the circumstances the results of such studies are robust and important that may prevent older individuals from enrolling in enough not to be ignored. I hope that the readership phase 1 trials and to study how these barriers can be appreciates access to this information as well as the reversed.17 In addition to the increased prevalence of opportunity to discuss the important topics emerging medical conditions that could potentially disqualify from these studies. patients from trial participation, other barriers may include limited availability of transportation and health Lodovico Balducci, MD care professional and family prejudices. Senior Member Data are also needed related to the effects of Program Leader, Senior Adult Oncology Program Editor, Cancer Control phase 1 clinical trials in individuals 75 years of H. Lee Moffitt Cancer Center & Research Institute age or older. A single report of phase 1 studies Tampa, Florida collected data from 28 individuals aged 80 years [email protected] and older, and found that, in this small patient population — the majority of whom was chemother- References apy naïve — dose-limiting toxicities were higher than 1. Talarico L, Chen G, Pazdur R. Enrollment of elderly patients in clinical 18 19 trials of cancer drug registration: a 7-year experience by the US Food and in younger individuals. Likewise, Schwandt et al re- Drug Administration. J Clin Oncol. 2004;22(22):4626-4631. ported that dose-limiting toxicities increased with age 2. Lutfullin A, Kuhlmann J, Wensing G. Adverse events in volunteer participating in phase I clinical studies: a single-center 5-year survey in 1,559 in patients older than 70 years but remained within an subjects. Intern J Clin Pharmacol Ther. 2005;43(5):227-236. acceptable threshold. These limited findings should 3. Borkowski JM, Duerr M, Donehower RC, et al. Relation between age and clearance rate of nine investigational anticancer drugs from phase I phar- make us question whether a chronological age thresh- macokinetic data. Cancer Chemother Pharmacol. 1994;33(6):493-496. old exists beyond which we can expect a reduction 4. van Heeckeren WJ, Fu P, Barr PM, et al. Safety and tolerability of phase I/II clinical trials among older and younger patients with acute myelogenous in dose-limiting toxicities in the majority of patients. leukemia. J Geriatr Oncol. 2011;2(3):215-221. For more than 20 years, health care professionals have 5. Zafar SF, Heilbrun LK, Vishnu P, et al. Participation and survival of geriatric patients in phase I clinical trials: the Karmanos Cancer Institute (KCI) advocated that physiological age should supersede experience. J Geriatr Oncol. 2011;2(1):18-24. chronological age in the care of older patients and in 6. Extermann M, Boler I, Reich RR, et al. Predicting the risk of chemo- 12 therapy toxicity in older patients: the Chemotherapy Risk Assessment Scale clinical trials of cancer treatment. While I still stand in High age patients (CRASH) score. Cancer. 2012;118(13):3377-3386. by this principle, I recognize that the prevalence of 7. Hurria A, Togawa K, Mohile ST, et al. Predicting chemotherapy toxicity 20 in older adults with cancer: a prospective multicenter study. J Clin Oncol. the so-called oldest old is ever increasing, so it is 2011;29(25):3457-3465. legitimate to ask whether chronological age accurately 8. Popa MA, Wallace KJ, Brunello A, et al. Potential drug interactions and chemotoxicity in older patients with cancer receiving chemotherapy. J Geriatr reflects a critical reduction of physiological reserve in Oncol. 2014;5(3):307-314. the majority of individuals older than 85 years of age. 9. Balducci L, Goetz-Parten D, Steinman MA. Polypharmacy and the management of the older cancer patient. Ann Oncol. 2013;(24 suppl 7):vii36-vii40. Should we have phase 1 trials reserved for older 10. Kelly CM, Power DG, Lichtman SM. Targeted therapy in older patients participants? This question may sound more philo- with solid tumors. J Clin Oncol. 2014;32(24):2635-2646. 11. Korkola J, Gray JW. Breast cancer genome--form and function. Curr sophical than practical. Personally, my position has Opin Genet Dev. 2010;20(1):4-14. been and continues to be that older individuals should 12. Balducci L. Studying cancer treatment in the elderly patient population.

April 2015, Vol. 22, No. 2 Cancer Control 233 Cancer Control. 2014;21(3):215-220. 13. Varadhan R, Yao W, Matteini A, et al. Simple biologically informed inflammatory index of two serum cytokines predicts 10 year all-cause mortality in older adults. J Gerontol A Biol Sci Med Sci. 2014;69(2):165-173. 14. Qu S, Wen W, Shu XO, et al. Association of leukocyte telomere length and breast cancer risk: nested case-control findings from the Shanghai Wom- en’s Health study. Am J Epidemiol. 2013;177(7):617-624. 15. Falandry C, Bonnefoy M, Freyer G, et al. Biology of cancer and aging: a complex association with cellular senescence. J Clin Oncol. 2014;32(24):2604-2610. 16. Luciani A, Balducci L. Multiple primary malignancies. Semin Oncol. 2004;31(2):264-273. 17. Denson AC, Mahipal A. Participation of the elderly population in clinical trials: barriers and solutions. Cancer Control. 2014;21(3):209-214. 18. Gaddipati H, Fu P, Dowlati A. Phase I clinical trials in patients ≥80. J Geriatr Oncol. 2011;2(2):142-146. 19. Schwandt A, Harris PJ, Hunsberger S, et al. The role of age on dose-limiting toxicities in phase I dose-escalation trials. Clin Cancer Res. 2014;20(18):4768-4775. 20. Nosraty L, Jylhä M, Raittila T, et al. Perceptions of the oldest old on successful aging, Vitality 90+ Study. J Aging Study. 2015;32:50-58.

234 Cancer Control April 2015, Vol. 22, No. 2 Original Research

Effect of Age on Clinical Outcomes in Phase 1 Trial Participants Amit Mahipal, MD, Aaron C. Denson, MD, Benjamin Djulbegovic, MD, PhD, Richard Lush, PhD, Ambuj Kumar, MD, Tzu-Hua Juan, MS, Michael J. Schell, PhD, and Daniel M.Sullivan, MD

Background: Most persons with cancer living in the United States are older than 65 years of age; however, in general, elderly persons are under-represented in clinical trials and outcomes data are lacking. Methods: Outcomes data were analyzed of elderly participants (≥ 65 years of age) enrolled in phase 1 clinical trials and the results compared with those of younger patients. All consecutive, single-center, phase 1 oncology trials initiated and completed at the H. Lee Moffitt Cancer Center & Research Institute between 1997 and 2007 were included. Patient data (including survival, response, and toxicity rates) were extracted from a cancer registry database and electronic medical records at Moffitt Cancer Center. Results: After excluding multi-institution trials, we analyzed 39 trials for a total of 1,162 enrolled study participants, 32.7% of whom were elderly. Among patients who underwent transplantation, median survival rates were worse in those who were elderly compared with those who were younger (44.9 vs 32.9 months; P = .0037). However, in the no-transplantation setting, participants who were elderly had a median survival rate of 10.9 months (95% confidence interval [CI]: 8.9–13.1) compared with 8.8 months (95% CI: 7.9–10.3) in those who were younger (P = .15). Both groups had similar overall response rates (15.2% vs 13.1%) and similar treatment-related mortality rates (1% vs 0.9%, respectively). Adverse events occurring among the elderly and younger participants were not statistically significant. Conclusions: Survival, response, toxicity, and treatment-related mortality rates were not significantly different between the elderly and younger phase 1 trial participants in the no-transplantation setting. Regardless of the complex pharmacological profiles and logistical issues involved in treating the elderly population, our data imply that elderly study participants do at least as well as their younger counterparts, contributing to the justification of increasing the phase 1 trial enrollment of elderly patients.

Introduction effectiveness of cancer treatments when compared In the United States, the majority of all cases of with younger populations.5,6 In 1 trial of 16,396 cancer are diagnosed in people 65 years of age and patients, Hutchins et al1 reported that 25% of study older.1,2 Despite a growing elderly population, and participants were 65 years of age or older. Similarly, a subsequently increasing cancer burden among in another study of 28,766 persons with cancer who the elderly, patients older than 65 years of age are were enrolled in registration trials of novel cancer frequently under-represented in oncology clinical drugs or for new indications of cancer drugs already trials.3,4 Limited data are available for this age group approved by the US Food and Drug Administration in terms of pharmacokinetics, toxicities, and the between 1995 and 2002, a total of 36% of patients were 65 years or older.7 Yet another study of 59,300 participants recruited in National Cancer Institute– From the Clinical Research Unit (AM, RL, DMS) and Department of sponsored, cooperative group trials between 1997 Biostatistics and Bioinformatics (T-H J, MJS), H. Lee Moffitt Cancer Center & Research Institute, and the Department of Internal Med- and 2000 reported that 32% of participants were icine (ACD, BD, AK), University of South Florida Morsani College 65 years of age or older.8 of Medicine, Tampa, Florida. Barriers to clinical trial participation among elder- Submitted February 10, 2015; accepted February 16, 2015. ly persons include stringent exclusion criteria, logisti- Address correspondence to Amit Mahipal, MD, Clinical Research cal challenges, and misconceptions among health care Unit, Moffitt Cancer Center, 12902 Magnolia Drive, Tampa, FL 33612. E-mail: [email protected] professionals and patients about the risks of partici- 9 No significant relationships exist between the authors and the com- pation. Cancer in geriatric patients may be difficult to panies/organizations whose products or services may be referenced treat because of multiple factors, including age-related in this article. physiological changes, significant comorbidities, and We thank Rasa Hamilton for editorial assistance. Our study received possible drug–drug interactions. These complexities valuable assistance from the Biostatistics and Bioinformatics Cores at Moffitt Cancer Center supported by a grant from the National may sway a health care professional’s decision to Institutes of Health (P30-CA76292). approach elderly patients with cancer for participation

April 2015, Vol. 22, No. 2 Cancer Control 235 in clinical trials.10,11 In addition, many physicians cite Data Extraction lack of information on elderly patients, calling on the Data were extracted using a standardized data need to increase data in patients aged 65 years and extraction form. Variables of interest included demo- older.12-14 Although the under-representation of elderly graphical information, relevant dates for trial par- persons enrolled in clinical trials is gaining atten- ticipation (enrollment, on treatment, off treatment, tion, the actual involvement of elderly persons with off study, and last contact), vital status, response, cancer in clinical trials has yet to reach appropriate off-study reason, and toxicity grades and types. Ran- proportions.15 Recently, however, some randomized dom reviews of the dataset were performed to ensure trials have specifically evaluated the role of therapy accuracy of the data collection. More than 10% of the in elderly persons with cancer.16,17 data was reviewed for accuracy. With regard to phase 1 trials, information on the Information on survival was primarily extracted enrollment of elderly participants and their clinical from the cancer registry data. We supplemented and outcomes is limited.6,18 Phase 1 trials are crucial for cross-referenced these data with the CRFs and EMRs. evaluating the safety profile of a drug, determining a Data on response rate were extracted directly from the safe dosage range, and identifying any adverse events. CRFs or EMRs for trials using the assessment noted For the majority of study participants, phase 1 clinical by investigators. Investigator assessment was utilized trials are safe and beneficial.18 In addition, prognostic for treatment decision-making on the trial and was indices are being developed to better predict survival considered accurate. For trials enrolling patients with rates and treatment tolerances in phase 1 studies.19,20 solid tumors, investigators used the Response Eval- However, the impact of age in predicting response, uation Criteria in Solid Tumors for response assess- treatment tolerance, survival rate, and possible com- ment.23 Investigators used disease-specific response plications remains largely unknown.21 criteria for assessment in hematological malignancy In this study, we sought to analyze the clinical trials. The CRFs for 1 myelodysplastic syndrome tri- outcomes of elderly participants (defined as al reported responses in terms of major and minor age ≥ 65 years) enrolled in phase 1 clinical trials at erythroid responses rather than complete and partial H. Lee Moffitt Cancer Center & Research Institute responses, among others. For consistency in reporting, (Tampa, Florida) and compared their results with we used the guidelines suggested by Cheson et al24 those of younger study participants (defined as age to convert these responses. < 65 years). We extracted data on toxicities assessed by investigators according to the Common Toxicity Crite- Methods ria (although the exact version varied depending on All consecutive, single-center, phase 1 oncology trials when the protocol was originally initiated) directly initiated and completed at Moffitt Cancer Center be- from CRFs or EMRs.25 Because toxicities were reported tween 1997 and 2007 were included in this analysis. as adverse events (AEs) in CRFs, we extracted data The dataset was included until 2007 to primarily focus for all grades 3, 4, and 5 AEs and noted whether on chemotherapeutic trials. All trials were registered they were attributed to treatment. Grade 3/4 toxicities and received Institutional Review Board approval. were defined as being severe to life threatening and All patients who gave informed consent to partici- either definitely, probably, or possibly attributed to pate on a phase 1 study by December 31, 2007, were the study drug.25 included on an intent-to-treat basis (ie, regardless of We used 2 methods to calculate mortality rate. whether they received the study regimen).22 Patients Any death within 30 days following the admin- who withdrew consent prior to receiving the study istration of the last dose of the study drug was drug were not included in the cohort. considered to be the 30-day mortality rate. The second method employed a direct extraction of the Data Sources data from the CRFs or EMRs on death attributed to An OnCore system (Forte Research Systems, Madison, treatment. We reported both methods to account Wisconsin) database was used to generate the list of for patients who succumbed to disease within phase 1 trials conducted at Moffitt Cancer Center. 30 days of being withdrawn from the study due to Patients who consented to more than 1 clinical trial disease progression without having experienced during the course of therapy were considered for the any treatment toxicity. first Moffitt Cancer Center–only trial for which they received treatment. The Moffitt Cancer Center cancer Statistical Analysis registry database, case report forms (CRFs) contain- The outcomes of this study were overall survival ing original trial data, and electronic medical records (OS) rate, overall response rate (ORR), any grade (EMRs) were data sources used to extract individual 3/4 toxicity, and treatment-related mortality and patient data. 30-day mortality rates. For the purpose of this anal-

236 Cancer Control April 2015, Vol. 22, No. 2 ysis, patients 65 years of age and older were con- Baseline Characteristics sidered to be elderly. Although analysis for benefits Table 1 lists the baseline characteristics of partici- (ie, survival and response rates) was performed pants included in this study. Of the entire cohort on an intent-to-treat basis, analysis for harms (ie, (N = 1,162), 32.7% (n = 380) were elderly. The me- toxicities and treatment-related mortality rates) was dian ages of the elderly and younger participant co- performed per protocol. We planned a priori subgroup analyses on outcomes according to disease Table 1. — Characteristics of the Volunteer Cohorts and treatment categories (eg, transplantation vs Category No. of Participants no-transplantation setting). Age < 65 y (%) Age ≥ 65 y (%) P Value Survival was calculated using the Kaplan–Meier Overall 782 (67.3) 380 (32.7) method from date of enrollment to last date of contact or death. A log-rank test was used to compare differ- Sex ences in survival between the subgroups. Differences Male 385 (49.2) 133 (35) < .0001 in response rates and toxicity grades were calculated Female 397 (50.8) 247 (65) using the chi-square and Jonckheere–Terpstra tests, Median age, y 53 (18–64) 69 (65–91) respectively. All reported P values were 2 sided, and a (range) significance level of .05 was used. Statistical analyses Hematological 303 (38.7) 107 (28.2) .0005 were performed using SAS 9.3 (SAS Institute, Cary, malignancy North Carolina). Solid tumor 479 (61.3) 273 (71.8) Underwent 295 (37.7) 65 (17.1) < .0001 Results transplantation Study Selection No transplan- 487 (62.3) 315 (82.9) Between 1997 and 2007, a total of 147 phase 1 tri- tation als were opened at Moffitt Cancer Center. Of these, Type of treatment 108 were excluded due to multi-institution involve- Chemotherapy 393 (50.3) 232 (61.0) < .0001 ment, resulting in 39 trials (31 no-transplantation and alone 8 transplantation trials) for a total of 1,162 study par- Chemotherapy 295 (37.73) 65 (17.1) + transplan- ticipants at Moffitt Cancer Center alone (Fig 1). These tation study participants formed the intent-to-treat cohort. Immunotherapy 14 (1.8) 17 (4.5) The median age in this cohort was 59 years (range: 18–91 years). Of the 1,162 study participants, 91% Gene therapy 59 (7.5) 47 (12.4) (n = 1,057) received treatment on study, 4.5% (n Chemotherapy 21 (2.7) 19 (5) + radiation = 52) never received treatment, and 4.6% (n = 53) received an alternative treatment off study. Rea- Disease category sons for study participants not receiving treat- Hematological 303 (38.7) 107 (28.2) < .0001 ment during the clinical trial included death, malignancies insurance issues, and the initiation of other therapies. Lung cancer 99 (12.7) 70 (18.4) Melanoma 83 (10.6) 43 (11.3) and other skin cancer

Potentially relevant phase 1 oncology trials Gastrointestinal 74 (9.5) 73 (19.2) identified in Moffitt Cancer Center clinical trials cancer database and conducted between 1997 and 2007 Breast cancer 63 (8.1) 7 (1.8) (n = 147) Renal cancer 31 (4.0) 21 (5.5) Hepatocellular 21 (2.7) 11 (2.9) carcinoma Excluded because trials were conducted at multiple institutions Soft-tissue 17 (2.2) 5 (1.3) (n = 108) sarcoma Sarcoma 16 (2.0) 3 (0.8) Gynecological 16 (2.0) 2 (0.5) cancer Phase 1 oncology trials that met inclusion criteria Mesothelioma 10 (1.3) 19 (5) (n = 39); enrolled 1,162 volunteers Cancer of the 9 (1.1) 5 (1.3) endocrine glands Other 40 (5.1) 14 (3.7) Fig 1. — Selection of studies.

April 2015, Vol. 22, No. 2 Cancer Control 237 horts were 69 years (range: 65–91 years) and 53 years younger study participants in the no-transplantation (range: 18–64 years), respectively. A single study cohort (n = 487), the clinical benefit rate was 41.5%; evaluating chemotherapy in 9 participants with CR was seen in 4.7%, PR in 8.4%, and SD in 28.3%. non–small-cell lung cancer exclusively enrolled elder- The differences in response rates were not statistically ly volunteers (defined as age ≥ 70 years). In the elderly significant between the 2 age groups (P = .40). population (n = 380), hematological malignancies rep- Among the study participants who received resented 28.2% (n = 107) of cancers, gastrointestinal stem cell transplantation, CR, PR, and SD were ob- cancers accounted for 19.2% (n = 73), and lung cancer served in 23.1%, 27.7%, and 23.1% of the elderly accounted for 18.4% (n = 70). In the younger popula- study participants, respectively. Progressive disease tion (n = 782), hematological malignancies, gastroin- (PD) was observed in 1.5% of study participants, and testinal cancers, and lung cancer were present in 38.7% 24.6% were not evaluable. Among the younger study (n = 303), 9.5% (n = 74), and 12.7% (n = 99) of the participants, the ORR was 57.3% (CR: 31.5%; study participants, respectively. These differences are PR: 25.8%). SD was reported in 18.6% of these study likely secondary to the age distribution of the re- participants and 7.5% had PD. spective diseases. Trials evaluating chemotherapy-only Overall Survival: In the no-transplantation regimens enrolled 61% (232/380) of elderly patients as group, the median OS rate was 10.9 months (95% compared with 50.3% (393/782) of younger patients. confidence interval [CI]: 8.9–13.1) in the elderly cohort Trials of stem cell transplantation enrolled 17.1% and and 8.8 months (95% CI: 7.9–10.3) in the younger 37.7% of elderly and younger patients, respectively. cohort (Fig 2; see Table 2). The difference in survival rates between the 2 age groups was not statistically Efficacy Outcomes significant (hazard ratio [HR] 0.89; 95% CI: 0.76–1.04; Response Rates: Among the study participants in the P = .145). No differences in the 2 age groups were no-transplantation cohort, the ORR was 14%; 5.7% of seen on multivariable analysis. Tumor category (solid study participants achieved complete response (CR) tumor vs hematological malignancy), sex, and type of and 8.2% of achieved partial response (PR; Table 2). treatment were significantly associated with outcomes In addition, 30.4% of study participants achieved stain the multivariable analysis (Table 3). ble disease (SD). Response could not be evaluated in In the transplantation group, elderly study par- 15.4% of study participants. The primary reason for ticipants had worse OS rates than younger study par- being unable to evaluate response included death or ticipants (HR 1.64; 95% CI: 1.17–2.29; P = .004). The withdrawal from the study prior to restaging scans, median OS rate was 32.9 months in the elderly group screen failure, or the initiation of alternative therapies. and 44.9 months in the younger group. In the multi- Among the elderly study participants in the variable analysis, age and type of tumor were signifi- no-transplantation cohort (n = 315), the ORR was cantly associated with survival (see Table 3). When 15.2% (CR: 7.3%; PR: 7.9%) and 33.7% had SD. The including both the transplantation and no-transplanta- clinical benefit rate (CR + PR + SD) was 48.9%. Among tion study cohorts, the OS rate was higher in younger study participants than in the Table 2. — Median Survival and Response Rates elderly (median OS: 15.8 vs 13.4 months; P = .002). Category No. of Outcomes Participants Median Survival Response Rates (%) Toxicity Outcomes Rate, mo (95% CI) CR PR SD PD NEa Among the elderly partici- Age, y pants, 31% and 30% of them < 65 782 15.8 (14.0–8.0) 14.8 15.0 24.7 30.3 15.2 experienced grade 3 and 4 toxicities, respectively. Grade ≥ 65 380 13.4 (11.3–16.2) 10.0 11.3 31.8 29.0 17.9 3 and 4 AEs were observed in Transplantation vs no transplantation 31.7% and 34% of the younger Transplantation 360 40.5 (37.6–48.5) 30.0 26.1 19.4 6.4 18.1 study participants, respective- < 65 y 295 44.9 (39.0–53.5) 31.5 25.8 18.6 7.5 16.6 ly. No significant differences ≥ 65 y 65 32.9 (22.4–39.9) 23.1 27.7 23.1 1.5 24.6 were seen in grade 3 and 4 No transplantation 802 9.5 (8.6–10.9) 5.7 8.2 30.4 40.4 15.2 AEs between the elderly and < 65 y 487 8.8 (7.9–10.3) 4.7 8.4 28.3 44.2 14.4 younger cohorts (P = .093). In the no-transplantation group, ≥ 65 y 315 10.9 (8.9–13.1) 7.3 7.9 33.7 34.6 16.5 the incidences of grade 3 and aA portion of patients were NE for response due to several reasons, including death prior to restaging scans, withdrawal from study prior to scans, alternate therapy, and screening failure. 4 AEs among the elderly were CI = confidence interval, CR = complete response, NE = not evaluable, PD = progressive disease, PR = partial response, 29.5% and 28.2%, respectively, SD = stable disease. and the incidences of grade 3

238 Cancer Control April 2015, Vol. 22, No. 2 Product-Limit Survival Estimates Table 3. — Multivariable Analysis for Transplantation 100 + Censored and No-Transplantation Cohorts Factor HR 95% CI P Value 80 No transplantation

60 Age, ≥ 65 y vs < 65 y 1.00 0.96–1.05 .8927 Male vs female 0.80 0.77–0.83 < .0001 40 Hematological malignancy 0.39 0.37–0.43 < .0001 vs solid tumor Survival Probability (%) 20 Type of treatment Chemotherapy + radiation 0.53 0.47–0.58 < .0001 0 vs chemotherapy 0 25 50 75 100 125 Gene therapy vs chemotherapy 0.56 0.52–0.59 Survival Time (mo) Immunotherapy vs chemotherapy 0.16 0.14–0.19 A Levels: ——— < 65 y — — — ≥ 65 y Transplantation Product-Limit Survival Estimates Age, ≥ 65 y vs < 65 y 1.90 1.73–2.09 < .0001 100 + Censored Male vs female 1.01 0.94–1.10 .7510 Hematological malignancy 0.41 0.37–0.45 < .0001 80 vs solid tumor CI = confidence interval, HR = hazard ratio. 60 similar in the study participants undergoing trans- 40 plantation; 78.6% of the elderly and 76.9% of the

Survival Probability (%) younger study participants experienced AEs of grade 20 3 or higher (P = .418).

0 Treatment-Related Mortality 0 25 50 75 100 125 For those in the no-transplantation group, the 30-day Survival Time (mo) B Levels: ——— < 65 y — — — ≥ 65 y mortality rate was 10% (30/300) among the elderly study participants and 9.2% (43/468) in the younger Product-Limit Survival Estimates study participants, whereas treatment-related mor- 100 + Censored tality rates were 1% (3/300) and 0.9% (4/468) for the elderly and younger cohorts, respectively. The 80 differences between these rates were not statistically significant. 60 Discussion 40 In our study cohorts (39 trials totaling 1,162 study

Survival Probability (%) participants), 32.7% of study participants were at least 20 65 years of age. Only 1 of the 39 studies evaluated was exclusively aimed at enrolling elderly participants. 0 To align the discrepancy between the limited num- 0 20 40 60 80 100 ber of trials focused on elderly participation and the Survival Time (mo) C Levels: ——— < 65 y — — — ≥ 65 y increasing number of elderly patients, more trials specifically designed for an elderly population are Fig 2A–C. — Overall survival. (A) Age < 65 years vs ≥ 65 years (N = 1,162). 21 (B) Age < 65 years vs ≥ 65 years among participants undergoing transplan- necessary. Due to the large percentage of patients tation (n = 360). (C) Age < 65 years vs ≥ 65 years among participants not with cancer aged 65 years and older,1 it is imperative undergoing transplantation (n = 802). to evaluate the safety and tolerability of possible treat- and 4 AEs among the younger population were 30.6% ments in elderly patients. This must be underscored by and 27.3%, respectively. Table 4 lists the AEs of at least the fact that pharmacokinetics and pharmacodynam- grade 3 experienced by all of the phase 1 clinical trial ics significantly differ from those in young patients cohorts in the no-transplantation setting. due to age-related biological changes.26 Furthermore, Gastrointestinal and metabolic nutritional toxic- elderly patients frequently have more comorbidities ities were observed at higher rates in younger study and higher rates of medication use than their young- participants, whereas general toxicities were higher er counterparts, thus making treatment options for in the elderly study participants. The toxicities were cancer more complicated.

April 2015, Vol. 22, No. 2 Cancer Control 239 Table 4. — Grade 3/4 Adverse Eventsa by Age we found no statistically significant dif- in the No-Transplantation Cohort ferences in survival, response, toxicity, Toxicity Age < 65 y Age ≥ 65 y P and treatment-related mortality rates Value Grade Grade Grade Grade between the age groups. To our knowl- 3, n (%) ≥ 4, n (%) 3, n (%) ≥ 4, n (%) edge, this is the first study to report tox- Hematological toxicities 81 (16.6) 115 (23.6) 45 (14.3) 72 (22.9) .4728 icity data comparisons between geriat- Cardiac disorders 14 (2.9) 3 (0.6) 2 (0.6) 5 (1.6) .3942 ric and younger populations enrolled in phase 1 clinical trials. Moreover, we did Ear and labyrinth disorders 1 (0.2) 0 (0) 2 (0.6) 0 (0) — not find any difference in the outcomes Gastrointestinal disorders 70 (14.4) 8 (1.6) 25 (8) 4 (0.5) .0056 with regard to whether the age cutoff of General disordersb 39 (8) 5 (1) 39 (12.4) 5 (1.6) .0307 70 years was used instead of 65 years. Infections 16 (3.3) 0 (0) 7 (2.2) 1 (0.3) .6725 Among the patients undergoing Metabolism and nutrition 45 (9.24) 12 (2.5) 17 (5.4) 2 (0.6) .0053 stem cell transplantation, age has been disorders demonstrated to be a major prognos- Musculoskeletal and 15 (3.1) 2 (0.4) 9 (2.9) 1 (0.3) .8003 tic factor in several retrospective stud- connective tissue disorders ies.29-32 Our results are consistent with Nervous system disorders 19 (3.9) 9 (1.8) 12 (3.8) 2 (0.6) .3557 prior studies, suggesting that elderly Psychiatric disorders 15 (3.1) 0 (0) 6 (1.9) 1 (0.3) .5176 patients have significant worse survival Renal and urinary disorders 6 (1.2) 0 (0) 6 (1.9) 2 (0.6) .1194 rates than younger patients. Respiratory disorders 33 (6.8) 4 (0.8) 22 (7) 7 (2.2) .3614 Our findings support the notion that age should not be used as a prognostic Skin and subcutaneous 13 (2.7) 0 (0) 16 (5.1) 0 (0) — disorders factor, particularly when determining Vascular disorders 17 (3.5) 7 (1.4) 4 (1.3) 12 (3.8) .8767 eligibility for phase 1 clinical trials. Furthermore, our finding of a median aGrouped by Common Terminology Criteria for Adverse Events. bThese include edema, pain, fatigue, and fever. survival rate of 10.9 months in elderly study participants not undergoing Several prognostic scores have been developed transplantation is comparable with results shown to predict clinical outcomes in patients participating in previous studies reporting on the outcomes of in phase 1 clinical trials. The Royal Marsden Hospi- participants enrolled in phase 1 clinical trials.18,28,33 tal prognostic score includes elevated lactate dehy- The current study also demonstrated that the drogenase levels, hypoalbuminemia, and more than incidence of high-grade toxicities is similar among 2 sites of metastasis as variables associated with elderly and younger study participants. By sup- poor prognosis.27 Wheler et al28 further validated the porting the finding that elderly study participants score at the phase 1 clinic at the MD Anderson Can- do not have worse outcomes than their younger cer Center (Houston, Texas). Gastrointestinal tumor counterparts in phase 1 clinical trials, our study may type and an Eastern Cooperative Oncology Group provide the confidence health care professionals performance status of 1 or more were both includ- need to help their elderly patients decide whether ed as additional factors associated with prognosis. to enroll in phase 1 clinical trials. Study participants with none of these risk factors had a median OS rate of 24 months compared with Limitations 4.1 months in study participants with the highest There are several limitations to our study, including risk scores. However, none of these scores included its retrospective nature. All trials included were from age as a prognostic factor. Moffitt Cancer Center, thus providing a single-center To our knowledge, only 1 previous study has perspective. It is unknown whether the data would reported on survival rates in geriatric patients partic- differ if data from other centers were included. Other ipating in phase 1 clinical trials.6,18 Zafar et al6 found limitations include the high probability for selection that elderly patients who enrolled in clinical trials bias among elderly patients. The low enrollment num- (n = 95) had a median OS rate of 8.4 months com- bers of elderly patients, relative to their proportion of pared with elderly patients who consented but were disease, may lead to enrollment of only the healthiest ineligible for study inclusion (n = 114) whose median of elderly patients. This possible bias is impossible to OS rate was 3.9 months. accommodate for and difficult to assess. We performed a systematic analysis of all con- Our study primarily included treatment with secutive oncology phase 1 trials conducted at Mof- cytotoxic or immunotherapeutic agents rather than fitt Cancer Center between 1997 and 2007 and com- targeted agents, which are much more common today. pared the benefits and harms in elderly and younger However, the prognostic scores mentioned above that study participants. In the no-transplantation setting, do not include age have been validated in patients

240 Cancer Control April 2015, Vol. 22, No. 2 receiving cytotoxic as well as targeted agents. 18. Italiano A, Massard C, Bahleda R, et al. Treatment outcome and survival in participants of phase I oncology trials carried out from 2003 to 2006 Although more studies are needed, our data at Institut Gustave Roussy. Ann Oncol. 2008;19(4):787-792. demonstrate that clinical outcomes, effectiveness, 19. Arkenau HT, Barriuso J, Olmos D, et al. Prospective validation of a prognostic score to improve patient selection for oncology phase I trials. J and rates of toxicities are not substantially different Clin Oncol. 2009;27(16):2692-2696. between elderly and younger study participants in 20. Garrido-Laguna I, Janku F, Vaklavas C, et al. Validation of the Royal Mars- den Hospital prognostic score in patients treated in the phase I clinical trials phase 1 clinical trials. program at the MD Anderson Cancer Center. Cancer. 2012;118(5):1422-1428. 21. Aapro MS, Köhne CH, Cohen HJ, et al. Never too old? Age should not be a barrier to enrollment in cancer clinical trials. Oncologist. 2005;10(3):198-204. Conclusions 22. Gravel J, Opatrny L, Shapiro S. The intention-to-treat approach in ran- Participation in phase 1 clinical trials should not domized controlled trials: are authors saying what they do and doing what they say? Clin Trials (London). 2007;4(4):350-356. be hindered by age. Rather, efforts should be made 23. Therasse P, Arbuck SG, Eisenhauer EA, et al; European Organization to increase clinical study participation among the for Research and Treatment of Cancer; National Cancer Institute of the United States; National Cancer Institute of Canada. New guidelines to evaluate the elderly patient population so as to decrease the gap response to treatment in solid tumors. J Natl Cancer Inst. 2000;92(3):205-216. between those who bear the disease burden and 24. Cheson BD, Greenberg PL, Bennett JM, et al. Clinical application and proposal for modification of the International Working Group (IWG) response those enrolled in clinical trials. By accumulating criteria in myelodysplasia. Blood. 2006;108(2):419-425. evidence in the elderly population, health care pro- 25. Cancer Therapy Evaulation Program. Common Terminology Criteria for Adverse Events). Version 3.0. Bethesda, MD: National Cancer Institute; 2003. fessionals may be encouraged to treat their elderly 26. Hubbard RE, O’Mahony MS, Woodhouse KW. Medication prescribing patients in an appropriately aggressive manner. As in frail older people. Eur J Clin Pharmacol. 2013;69(3):319-326. 27. Bachelot T, Ray-Coquard I, Catimel G, et al. Multivariable analysis the US population continues to age, information on of prognostic factors for toxicity and survival for patients enrolled in phase I treatment options for elderly persons has become clinical trials. Ann Oncol. 2000;11(2):151-156. 28. Wheler J, Tsimberidou DH, Hong D, et al. Survival of 1,181 patients in a — and will continue to be — increasingly crucial phase I clinic: the MD Anderson Clinical Center for targeted therapy experience. to patient care. Clin Cancer Res. 2012;18(10):2922-2929. 29. Sierra J, Pérez WS, Rozman C, et al. Bone marrow transplantation from HLA-identical siblings as treatment for myelodysplasia. Blood. 2002;100(6):1997-2004. References 30. de Witte T, Hermans J, Vossen J, et al. Haematopoietic stem cell trans- 1. National Cancer Institute. Surveillance, Epidemiology, and End Results plantation for patients with myelo-dysplastic syndromes and secondary acute Program stat fact sheets: all cancer sites. http://seer.cancer.gov/statfacts/html/ myeloid leukaemias: a report on behalf of the Chronic Leukaemia Working all.html. Accessed February 20, 2015. Party of the European Group for Blood and Marrow Transplantation (EBMT). 2. American Cancer Society. Cancer Facts & Figures 2015. Atlanta, GA: Br J Haematol. 2000;110(3):620-630. American Cancer Society; 2015. http://www.cancer.org/acs/groups/content/@ 31. Deeg HJ, Storer B, Slattery JT, et al. Conditioning with targeted busulfan editorial/documents/document/acspc-044552.pdf. Accessed February 18, 2015. and cyclophosphamide for hemopoietic stem cell transplantation from relat- 3. Hutchins LF, Unger JM, Crowley JJ, et al. Underrepresentation of ed and unrelated donors in patients with myelodysplastic syndrome. Blood. patients 65 years of age or older in cancer-treatment trials. N Engl J Med. 2002;100(4):1201-1207. 1999;341(27):2061-2067. 32. Lim Z, Brand R, Martino R, et al. Allogeneic hematopoietic stem-cell 4. Townsley CA, Selby R, Siu LL. Systematic review of barriers to the transplantation for patients 50 years or older with myelodysplastic syndromes recruitment of older patients with cancer onto clinical trials. J Clin Oncol. or secondary acute myeloid leukemia. J Clin Oncol. 2010;28(3):405-411. 2005;23(13):3112-3124. 33. Wheler J, Tsimberidou AM, Hong D, et al. Survival of patients in 5. van Heeckeren WJ, Fu P, Barr PM, et al. Safety and tolerability of phase a phase 1 clinic: the M. D. Anderson Cancer Center experience. Cancer. I/II clinical trials among older and younger patients with acute myelogenous 2009;115(5):1091-1099. leukemia. J Geriatr Oncol. 2011;2(3):215-221. 6. Zafar SF, Heilbrun LK, Vishnu P, et al. Participation and survival of geriatric patients in phase I clinical trials: the Karmanos Cancer Institute (KCI) experience. J Geriatr Oncol. 2011;2(1):18-24. 7. Talarico L, Chen G, Pazdur R. Enrollment of elderly patients in clinical trials for cancer drug registration: a 7-year experience by the US Food and Drug Administration. J Clin Oncol. 2004;22(22):4626-4631. 8. Lewis JH, Kilgore ML, Goldman DP, et al. Participation of patients 65 years of age or older in cancer clinical trials. J Clin Oncol. 2003;21(7):1383-1389. 9. Lara PN Jr, Higdon R, Lim N, et al. Prospective evaluation of cancer clinical trial accrual patterns: identifying potential barriers to enrollment. J Clin Oncol. 2001;19(6):1728-1733. 10. Kornblith AB, Kemeny M, Peterson BL, et al; Cancer and Leukemia Group B. Survey of oncologists’ perceptions of barriers to accrual of older patients with breast carcinoma to clinical trials. Cancer. 2002;95(5):989-996. 11. Foster JA, Salinas GD, Mansell D, et al. How does older age influence oncologists’ cancer management? Oncologist. 2010;15(6):584-592. 12. Knickman JR, Snell EK. The 2030 problem: caring for aging baby boomers. Health Serv Res. 2002;37(4):849-884. 13. Hempenius L, Slaets JPJ, Boelens MA, et al. Inclusion of frail elderly patients in clinical trials: solutions to the problems. J Geriatr Oncol. 2013;4(1):26-31. 14. Unger JM, Coltman CA Jr, Crowley JJ, et al. Impact of the year 2000 Medicare policy change on older patient enrollment to cancer clinical trials. J Clin Oncol. 2006;24(1):141-144. 15. Townsley CA, Selby R, Siu LL. Systematic review of barriers to the recruitment of older patients with cancer onto clinical trials. J Clin Oncol. 2005;23(13):3112-3124. 16. Quoix E, Zalcman G, Oster JP, et al; Intergroupe Francophone de Cancérologie Thoracique. Carboplatin and weekly paclitaxel doublet chemotherapy compared with monotherapy in elderly patients with advanced non-small-cell lung cancer: IFCT-0501 randomised, phase 3 trial. Lancet. 2011;378(9796):1079-1088. 17. Muss HB, Berry DA, Cirrincione CT, et al; CALGB Investigators. Adju- vant chemotherapy in older women with early-stage breast cancer. N Engl J Med. 2009;360(20):2055-2065.

April 2015, Vol. 22, No. 2 Cancer Control 241 Case Report

Primary Enteropathy-Associated T-Cell Lymphoma Type 2: An Emerging Entity? Nicole M. Grigg-Gutierrez, MD, Rodolfo Estremera-Marcial, MD, William W. Cáceres, MD, and Doris H. Toro, MD

Summary: Intestinal T-cell lymphoma is a rare hemato- of all colorectal neoplasms.2,4 Establishing a diagnosis logical malignancy that can present as primary intesti- of primary colorectal lymphoma requires histological nal lymphoma or as a manifestation of systemic disease. confirmation that the lymphoproliferative neoplasm is Primary involvement accounts for approximately 0.1% isolated to the colon and the regional lymph nodes.1,4 to 0.5% of all colorectal neoplasms. It is an aggressive This hematological malignancy has a male pre- disease with a poor prognosis and low survival rate. dominance and a higher incidence in those 50 to Inflammatory bowel disease, celiac disease, im- 70 years of age.5 Inflammatory bowel disease, celiac munosuppression, and infectious etiologies, disease, immunosuppressive states, and infectious eti- such as Epstein–Barr and human T-lymphotropic ologies, such as Epstein–Barr virus (EBV) and human viruses, have been reported as risk factors, but no T-lymphotropic virus (HTLV), have been reported to direct causal link has been established. Herein, we be risk factors; however, no direct causal link has examine the case of a Hispanic man 69 years of age been established.5 diagnosed with positive CD3, CD7, CD8, CD43, and Upon presentation, symptoms are typically non- Bcl-2 diffuse primary colorectal T-cell lymphoma. specific and may include changes in bowel habits, The patient did not exhibit a concomitant autoim- weight loss, abdominal pain, diarrhea, and rectal mune or genetic disease. Because of the patient’s bleeding.1 Acute abdomen secondary to intestinal history of polyps, surveillance colonoscopy was obstruction and tumor perforation requiring emer- performed and the diagnosis was confirmed. gent surgical intervention have also been reported on initial presentation.6 Background Early diagnosis is crucial in obtaining local control Gastrointestinal (GI) lymphoid tissues exist in the of the disease as well as to increase the likelihood of intestinal epithelium, lamina propria, submucosa, achieving remission with surgical intervention and and lymph nodes.1 An accumulation of lymphocytic adjuvant chemotherapy.3,7 Although colonoscopy is a tumor cells can be seen in any of these tissue layers crucial diagnostic tool in the diagnosis of colorectal in GI lymphomas. Colorectal lymphomas manifest disease, little is known of its value in the diagnosis as either primary tumors or generalized lymphoma of primary colorectal lymphoma.7 with associated colorectal involvement.1 Primary ex- Herein we present the case of a man diagnosed tranodal GI lymphomas are rare neoplastic process- with diffuse primary colorectal T-cell lymphoma while es accounting for 1% to 4% of all GI malignancies.2 being evaluated for unexplained weight loss. Primary extranodal lymphomas of the GI tract are most common on the stomach (50%–60%) and small Case Report intestine (20%–30%).3 Cases of primary lymphomas A Hispanic man 69 years of age presented with a past of the colon are rare and most are of B-cell origin. medical history of hypertension, type 2 diabetes mel- Primary T-cell lymphomas of the lower GI tract are litus, gout, arthritis, and hyperlipidemia. He admitted infrequent, representing approximately 0.1% to 0.5% to tobacco use and alcohol abuse. He was evaluated for worsening dysphagia to solids and liquids that From the Veterans Affairs Caribbean Healthcare System, San Juan, had lasted for 5 months and was associated with a Puerto Rico. 30-pound weight loss. He denied fever, chills, nau- Submitted September 19, 2014; accepted February 3, 2015. sea, vomiting, abdominal pain or distention, changes Address correspondence to Doris H. Toro, MD, VA Caribbean in bowel habits, pencil-like stool, early satiety, food Healthcare System, 10 Casia Street, San Juan, PR 00921. E-mail: intolerance, odynophagia, tenesmus, or GI bleeding. [email protected] Findings on the physical examination were gross- No significant relationships exist between the authors and the companies/organizations whose products or services may be ly unremarkable. In particular, the abdomen was referenced in this article. soft, nontender to palpation, and had normal bowel

242 Cancer Control April 2015, Vol. 22, No. 2 sounds. No palpable organomegaly was present, nor findings, immunostains were performed to rule out were masses or lymphadenopathy. Findings on the lymphocytic colitis or neoplastic T-cell proliferation. rectal examination were also unremarkable. The pa- The specimens were sent for further evaluation to the tient had adequate sphincter tone with no evidence of National Cancer Institute. fissures, nodules, or ulcers. Routine blood work was Histologic examination of the biopsies also identified similarly unremarkable, with no evidence of anemia infiltration by monomorphic, medium-sized lympho- or leukocytosis, and the peripheral smear confirmed cytes with a rim of pale cytoplasm and round dark nuclei. a normal leukocyte distribution. Florid intraepithelial infiltration of the crypts was In the setting of chronic dysphagia and weight present as well as prominent intraepithelial lympho- loss accompanied by alcohol and tobacco use, up- cytosis. Minimal inflammatory background without per endoscopy was obtained. Findings were re- necrosis was also documented. Immunostains of the markable for diffuse, severe nonerosive gastritis crypt infiltrate were highlighted by CD3, CD8, CD56, and mild nonerosive duodenitis without evidence TIA-1, and T-cell receptor (TCR) γ; they were focally of mucosal scalloping or other duodenal pathology. positive for perforin; and they had weak staining for No evidence suggested the presence of esophageal CD2. CD20, CD4, CD5, TCR-β, granzyme B, and Ep- lesions, rings, webs, or mucosal defects. Results stein–Barr encoding region in situ hybridization were from the rapid urease test (also known as the Cam- negative. Polymerase chain reaction studies showed pylobacter-like organism test) and gastric biopsies evidence of clonal rearrangements of the TCR gene. were negative for Helicobacter pylori. The histology and immunophenotype supported the Colonoscopy was also scheduled due to the pa- diagnosis of enteropathy-associated T-cell lymphoma tient’s remote history of colon polyps of unknown (EATL) type 2 with evidence of the δ phenotype. histology diagnosed more than 5 years prior to his Staging computed tomography of the chest, abdomen, current visit. Colonoscopy revealed the presence of a and pelvis showed no evidence of lymphadenopathy or sessile polyp at the cecum (0.5 × 0.3 × 0.2 cm), 2 ses- extranodal organ involvement. Positron emission to- sile polyps at the ascending colon (0.5 × 0.3 × 0.3 cm), mography was obtained and revealed no avid lesions. 1 sessile polyp at the rectum (0.5 × 0.2 × 0.2 cm), and Results from HIV by enzyme-linked immunosorbent 1 polyp 15 cm from the anal verge (2 × 1 × 0.7 cm). assay (ELISA) and antibodies against HTLV types 1 The findings also demonstrated evidence of an erythem- and 2 were negative. EBV serology was remarkable for atous and edematous fold 60 cm from the entry site and negative antibodies against EBV viral-capsid antigen a flat lesion 27 cm from the anal verge (Fig 1). immunoglobulin (Ig) M, Epstein–Barr nuclear antigen, All polyps were excised, and the histology was and early antigen with positive IgG antibodies against consistent with tubular adenomas. Histological ex- viral capsid, which has been associated with primary amination of the colonic flat lesions biopsy sam- infection in the acute phase. ples revealed abundant intraepithelial lymphocytes Positive serology against EBV brought conten- (Figs 2 and 3), and immunophenotyping revealed tion regarding the diagnosis of EATL. However, an intraepithelial T-cell population positive for CD3 no evidence suggested the presence of pleomor- and negative for CD10 and CD5. Because of these phic lymphoma cells, polymorphic inflammatory

Fig 1. — A large, flat ulcerated lesion with surrounding erythema and edema Fig 2. — Section of colon showing neoplastic intraepithelial lymphocytes 27 cm from the anal verge. infiltrating the colonic crypts (hematoxylin and eosin, × 400).

April 2015, Vol. 22, No. 2 Cancer Control 243 infiltrate, angioinvasion, angiocentricity, or zon- Discussion al necrosis — all of which are features of the na- Lymphomas are hematological malignancies with a sal type of natural killer (NK)/T-cell lymphomas. wide variety of histological subtypes and a broad spec- Furthermore, the patient had TCR gene rearrange- trum of clinical behaviors, aggressiveness, and progno- ments positive for CD8, which are not seen in the sis.8 It is the eight and ninth most common cause of can- nasal type of NK/T-cell lymphoma. cer-related death in US women and men, respectively.9 Celiac disease was excluded based on undetect- The incidence rate for non-Hodgkin lymphoma (NHL) able tissue transglutaminase antibody (< 1 U/mL) in slightly increased among men between 2007 and 2011.9 the setting of an adequate IgA level (201 mg/dL; ref- In the United States, an estimated 80,900 new cases erence range, 131–407). Viral serology for hepatitis B of lymphoma will be diagnosed in 2015; of those, and C (hepatitis B core antibody, hepatitis B surface 71,850 will be NHL.9 The GI tract is the main site for antigen, hepatitis B surface antibody, and hepatitis C the extranodal dissemination of lymphoma and con- antibody by ELISA) was negative. Lactate dehydroge- stitutes 40% of all cases.4 However, primary colorectal nase (LDH) levels were within normal limits. Bone lymphoma is relatively rare, making up approximately marrow biopsy and aspirate were morphologically 0.1% to 0.5% of all colorectal neoplasms and represent- normal with no B or T cells. Monoclonal cells were ing the third most common large bowel malignancy apparent by flow cytometry. after adenocarcinoma and neuroendocrine tumor.4 Given the localized disease, the patient subse- The most common location for colorectal lymphoma quently underwent total abdominal colectomy with is the cecum and is suspected to be secondary to ileorectal anastomosis (Fig 4). The pathology report the abundance of lymphoid tissue in this anatomical identified 36 regional lymph nodes with focal areas region.5 Tumors of the descending and rectosigmoid of diffuse paracortical monotonous tumor consistent colon account for approximately 25% of all colorectal with a malignant lymphomatous tumor. Microscopic lymphomas.8 examination of the colon further demonstrated ex- EATL is the only well-defined clinicopathological tensive lymphomatous involvement extending to the entity of primary T-cell GI lymphomas. It represents distal and radial margins. a rare NHL of T-cell origin,10 and it accounts for less The immediate postoperative period was compli- than 1% of all NHLs.11 The World Health Organiza- cated by multiple episodes of diarrhea that required tion (WHO) classification of tumors of hematopoietic aggressive hydration and antidiarrheal medications, and lymphoid tissue distinguishes between 2 types including loperamide and diphenoxylate/atropine. of EATL, with type 1 accounting for 80% to 90% of Because of the extent of the disease, adjuvant chemo- cases.10 The 2 subtypes, types 1 and 2, are classified therapy was started with cyclophosphamide, hydroxy- based on histology, immunophenotype, and relation- daunorubicin, vincristine, and prednisone (CHOP). ship to celiac disease.12 EATL is an infrequent type of Response to the induction cycle was poor. Due to aggressive lymphoma in most parts of the world, with septic shock secondary to community-acquired pneu- an estimated annual incidence of 0.5 to 1 per 1 million monia, the patient died 3 months after the diagnosis people in Western countries.13 A higher incidence has was established. been reported in northern Europe, where it overlaps

Fig 4. — Gross surgical excision specimen showing an ulcerated lesion with Fig 3. — Section of colon showing transmural neoplastic lymphoid infiltrate irregular borders with a full thickness (transmural) infiltrate of neoplastic (hematoxylin and eosin, × 20). cells consistent with enteropathy-associated T-cell lymphoma.

244 Cancer Control April 2015, Vol. 22, No. 2 with celiac disease. Type 1 is more common in Europe and surface epithelium.10 Immunohistochemistry of and type 2 is more common in Asia, whereas both the tumor cells and intraepithelial lymphocytes on EATL types are equally common in North America.12 the adjacent mucosa will be negative for CD4 and EATL has a male predominance with a median onset positive for CD3, CD8, and CD56 in 80% to 90% of during the sixth decade of life.14 cases and TCR-β.17 This subtype is also characterized Type 1 accounts for two-thirds of all cases of EATL by chromosome 8q24 (MYC) amplifications and, less and is associated with celiac disease, which is a food commonly, by gains of 1q and 5q.10,16 intolerance disorder in Western populations that has a Both types of EATL can be distinguished from prevalence of 0.5% to 1%.13 The relationship between reactive expansion by testing the clonality of the EATL type 1 and celiac disease is well established TCR genes. and is characterized by positive results on serology tests, HLA DQ2/8 expression in up to 90% of cases, Clinical Presentation and associated clinical findings such as dermatitis The clinical presentation of primary colorectal lym- herpetiformis and hypospleism.10 phomas is diverse and the duration of the presenting symptoms can widely vary.11 Patients with EATL most Type 1 commonly present with abdominal pain, weight loss, Microscopic findings of EATL type 1 are characterized diarrhea, vomiting, fatigue, and anorexia.10,11 In addi- by medium- to large-sized lymphocytes with round tion, a palpable abdominal mass, bowel perforation, or angulated vesicular nuclei, prominent nucleoli, bowel obstruction, B symptoms, and hemophagocyt- and moderate to abundant pale-staining cytoplasm.12 ic syndrome have been reported in some cases.11,17 Neoplastic tissue tends to be infiltrated with abundant A history of adult-onset celiac disease with or without eosinophils and histiocytes, which may obscure the a disease-free period is typically present in patients small numbers of tumor cells.10 Coagulative necrosis with EATL type 1; childhood onset of celiac disease is common.12 The intestinal mucosa adjacent to the has also been associated with this subtype but not as tumor frequently has enteropathic features consisting frequently.10 In some patients, EATL is simultaneously of villous atrophy, crypt hyperplasia, increased inflam- diagnosed with celiac disease.10 matory cells in the lamina propria, and intraepithelial Our patient presented with unintentional weight lymphocytosis.10,12,14 loss alone, which has been reported in 27% to Because EATL type 1 may present with 80% cases in the literature.7 Serology tests for celi- monomorphic, small- to medium-sized cells, ac disease were negative. Patients typically ignore immunohistochemistry is generally help- this relatively nonspecific symptom for an extend- ful in distinguishing between the 2 subtypes. ed period of time, thus further contributing to an In general, type 1 tumors cells are positive for CD3, advanced stage of disease at the time of diagnosis. CD7, and CD103, but negative for CD5, CD4, and In addition, performance status is variable at the time CD56; however, 10% of cases are positive for CD56.15 of presentation, with most patients having a WHO Tumor cells can also be positive or negative for CD8 performance status of 1 to 3.11 and TCR-β, and a varying proportion of cells express More than 90% of EATLs arise in the CD30.10 Furthermore, EATL can conceal complex seg- small intestine; the jejunum and proximal il- mental chromosomal amplifications or deletions, and eum are the most frequent sites.10,12 EATL of type 1 frequently displays deletions of 16q12.1 and the large intestine, as in our patient, has been gains of 1q, 5q, and 9q31.3.10,16 reported in approximately 16% to 18% of cases.12,14 Growth patterns tend to be diverse and lesions Type 2 may range from multiple, ulcerating raised mucosal Type 2 accounts for 10% to 20% of all cases and masses to 1 or more ulcers, large exophytic masses, has a broader geographical distribution.10,15,17 Typ- strictures, or plaques.1,12 Classical lymphoma stag- ically, it occurs sporadically, but up to one-quarter ing systems have failed to provide adequate prog- of patients have a history of celiac disease and 30% nostic treatment guidance for patients with EATL. to 40% express HLA DQ2/8.12 Histologically, type Although they have not been validated, several case 2 is characterized by multiple foci of small, round series have identified other novel adverse predictors uniform cells with dark nuclei and a rim of pale of survival in patients with this type of lymphoma, cytoplasm. Heavy infiltration of the intestinal crypt such as elevated levels of C-reactive protein and LDH, epithelium is present and a few admixed inflam- tumor size larger than 5 cm, and a nonambulatory matory cells may be seen; coagulative necrosis is performance status.16,18 absent. The adjacent intestinal mucosa shows villous atrophy and crypt hyperplasia with marked in- Diagnosis traepithelial lymphocytosis involving both the crypt An exact diagnostic algorithm for EATL remains

April 2015, Vol. 22, No. 2 Cancer Control 245 undefined because of the lack of studies aimed at first-line treatment in CD30+ EATL in addition to current comparing the accuracy of the different diagnostic strategies to improve survival and response rates in pa- modalities.18 Colonoscopy is valuable in the diagnosis tients whose tolerance to anthracycline-based therapy of primary intestinal lymphoma, but a high index of is expected to be poor. Additional case reports using suspicion is needed for tissue biopsy and diagnosis.8 alemtuzumab, a monoclonal antibody against CD52, In 1 series, colonoscopy accurately diagnosed prima- and romidepsin, a histone deacetylase inhibitor, have ry intestinal lymphoma in 23% of cases.17 Explana- shown promising clinical responses.18 tion for this may involve an insufficient specimen for pathological evaluation, because random biopsies in Conclusion the absence of apparent pathology are often need- This case satisfies the histology and immunophenotypi- ed to accurately identify and determine the extent cal criteria for enteropathy-associated T-cell lymphoma of lymphoma. Colorectal lymphoma is most often type 2 and highlights the difficulties in appropriately imaged using computed tomography, which can provide diagnosing T-cell malignancies. The incidence and lo- extraluminal and anatomical information regarding calization rates of primary gastrointestinal lymphomas tumor size, depth of invasion, and regional lymph are known to vary based on race and geographical node involvement. Positron emission tomography location, but their incidence and prevalence in the may be of use for diagnosis and for following-up Hispanic population have not yet been determined. patients, but its role has not been determined.8 A A significant increase can be seen in the incidence study published in 2012 showed that EATL type 2 is of EATL in the United States, most likely reflecting fluorodeoxyglucose (FDG) nonavid in 67% of cases, the increasing seroprevalence of celiac disease and whereas type 1 is uniformly FDG avid.14 In this case, better recognition of the rare types of T-cell lympho- computed tomography and positron emission tomog- mas.23 It will be important to monitor the incidence of raphy imaging failed to discover any colorectal dis- both EATL and premalignant conditions as evidence ease, thus highlighting the importance of high-quality accumulates among whites and ethnic minorities. colonoscopy in detecting rare GI pathologies that Although emerging data are encouraging, a significant would have not been diagnosed otherwise. need still exists for improved treatment modalities in this disease.19 Treatment EATL has limited validated treatment strategies References because of the lack of randomized clinical trials. 1. Lee HJ, Han JK, Kim TK, et al. Primary colorectal lymphoma: spectrum imaging findings with pathologic correlation. Eur Radiol. 2002;12(9):2242-2249. Historically, treatment options have included surgi- 2. Okada M, Maeda K, Suzumiya J, et al. Primary colorectal T-cell lym- cal resections with or without anthracycline-based phoma. J Gastroenterol. 2003;38(3):76-384. 10,13 3. Zinzani PL, Magagnoli M, Pagliani G, et al. Primary intestinal lym- chemotherapy ; however, results with these treat- phoma: clinical and therapeutic features of 32 patients. Haematologica. ment modalities are limited. The National Com- 1997;82(3):305-308. 4. Stanojevic GZ, Nestorovic MD, Brankovic BR, et al. Primary colorectal prehensive Cancer Network recommends an anth- lymphoma: an overview. World J Gastrointest Oncol. 2011;3(1):14-18. racycline-based combination therapy followed by 5. Wong MT, Eu KW. Primary colorectal lymphomas. Colorectal Dis. 2006;8(7):586-591. autologous stem cell transplantation in eligible pa- 6. Zettl A, deLeeuw R, Haralambieva E, et al. Enteropathy-type T-cell tients.19 A CHOP-like regimen, similar to what our lymphoma. Am J Clin Pathol. 2007;127(5):701-706. 7. Wang MH, Wong JM, Lien HC, et al. Colonoscopic manifestations of patient received, has an overall response rate of 30% primary colorectal lymphoma. Endoscopy. 2001;33(7):606-609. to 60%.20 With current treatment strategies, the median 8. Quayle FJ, Lowney JK. Colorectal lymphoma. Clin Colon Rect Surg. 2006;19(2):49-53. overall survival rate is approximately 10 months; the 9. American Cancer Society. Cancer Facts & Figures 2015. Atlanta, GA: 5-year estimated survival rate is 20%.16 As in our pa- American Cancer Society; 2015. http://www.cancer.org/acs/groups/content/@ editorial/documents/document/acspc-044552.pdf. Accessed February 19, 2015. tient, due to comorbidities, performance status, and his 10. Sieniawski MK, Lennard AL. Enteropathy-associated T-cell lymphoma: fragile state following total colectomy, only 50% of pa- epidemiology, clinical features, and current treatment strategies. Curr Hematol 18 Malig Rep. 2011;6(4):231-240. tients are eligible to undergo planned chemotherapy. 11. Gale J, Simmonds PD, Mead GM, et al. Enteropathy-type intestinal Given data regarding their use in other CD30+ T-cell T-cell lymphoma: clinical features and treatment of 31 patients in a single center. J Clin Oncol. 2000;18(4):795-803. lymphomas, agents such as brentuximab vedotin, an 12. Rand AJ, Cardona DM, Proia AD, et al. Clinically undiagnosed en- anti-CD30 conjugated antibody, can be considered in teropathy associated T-cell lymphoma type II presenting with prolonged lower gastrointestinal tract symptoms: report of an autopsy case and review of di- patients with evidence of disease progression af- agnostic challenges and clinicopathological correlation. J Gastrointest Oncol. ter induction therapy; a median duration of re- 2013;4(1):103-108. 21 13. Sieniawski M, Angamuthu N, Boyd K, et al. Evaluation of enteropa- mission of 13.6 months has been reported. thy-associated T-cell lymphoma comparing standard therapies with a novel reg- Khalaf et al22 reported on a patient with EATL imen including autologous stem cell transplantation. Blood. 2010;115(18):3664- 3670. successfully treated with brentuximab vedotin as sal- 14. Tse E, Gill H, Loong F, et al. Type II enteropathy-associated T-cell vage treatment who had a good response and disease lymphoma: a multicenter analysis from the Asia Lymphoma Study Group. Am J Hematol. 2012;87(7):663-668. remission at 9 months of follow-up. Due to its efficacy 15. Sun ZH, Zhou HM, Song GK, et al. Intestinal T-cell lymphomas: a ret- and tolerance, this medication could be considered as rospective analysis of 68 cases in China. World J Gastroenterol. 2014;20(1):

246 Cancer Control April 2015, Vol. 22, No. 2 296-302. 16. Delabie J, Holte H, Vose JM, et al. Enteropathy-associated T-cell lymphoma: clinical and histological findings from the International Peripheral T-Cell Lymphoma Project. Blood. 2011;118(1):148-155. 17. Sun J, Lu Z, Yang D, et al. Primary intestinal T-cell and NK-cell lymphomas: a clinicopathological and molecular study from China focused on type II enteropathy-associated T-cell lymphoma and primary intestinal NK-cell lymphoma. Mod Pathol. 2011;24(7):983-992. 18. Di Sabatino A, Biagi F, Gobbi PG, et al. How I treat enteropathy-associated T-cell lymphoma. Blood. 2012;119(11);2458-2468. 19. National Comprehensive Cancer Network. NCCN Clinical Practice Guidelines in Oncology for Non-Hodgkin’s Lymphomas. v 1.2015. http://www. nccn.org/professionals/physician_gls/pdf/nhl.pdf. Accessed February 19, 2015. 20. Daum S, Ulrich R, Heise W, et al. Intestinal Non-hodgkin’s lymphoma: a multicenter prospective clinical study for the German Study Group on Intestinal lymphoma. J Clin Oncol. 2003;21(14):2740-2746. 21. Pro B, Advani R, Brice P, et al. Brentuximab vedotin (SGN-35) in patients with relapsed or refractory systemic anaplastic large cell lymphoma: results of a phase II study. J Clin Oncol. 2012;30(18):2190-2196. 22. Khalaf WF, Caldwell ME, Reddy N. Brentuximab in the treatment of CD30-positive enteropathy-associated T-cell lymphoma. J Natl Compr Canc Netw. 2013;11(2):137-140. 23. Sharaiha RZ, Lebwohl B, Reimers L, et al. Increasing incidence of enteropathy-associated T-cell lymphoma in the United States, 1973-2008. Cancer. 2012;118(15):3786-3792.

April 2015, Vol. 22, No. 2 Cancer Control 247 Special Report

Cost Effectiveness of Colorectal Cancer Screening Strategies Shaan S. Patel and Meredith L. Kilgore, PhD

Background: Several screening tests are available to detect colorectal cancer (CRC) and reduce the incidence and mortality of CRC. The purpose of this review was to determine how current CRC screening strategies for CRC compare with no screening and whether agreement exists with regard to the cost effectiveness of different strategies. Methods: Databases were searched for cost-effectiveness analyses focused on CRC screening strategies in the United States and published between May 2007 and February 2014. We analyzed the uses of fecal occult blood, fecal immunochemistry, and stool DNA tests, as well as sigmoidoscopy, colonoscopy, and virtual colonoscopy. A paired comparison of each screening strategy with no screening across each of the studies reviewed was conducted. A series of paired comparisons of the results reported in each of the studies is also included. Results: When compared with no screening, all CRC screening strategies evaluated in this review were cost effective. There was disagreement as to which screening strategy was the most cost effective. However, sigmoidoscopy combined with fecal blood testing always dominated either strategy alone. Studies comparing colonoscopy with fecal blood testing, sigmoidoscopy, or both had mixed results. Conclusions: Compared with no screening, all CRC screening strategies are more cost effective. Study results disagree as to which screening strategy should be the preferred method.

Introduction Identifying the most cost-effective screening strategy Colorectal cancer (CRC) is the third leading cause of for CRC can inform health care professionals as to cancer-related deaths among both men and women which screening strategy to recommend for patients, in the United States, and 132,700 estimated new and, thus, reduce the cost and maximize the effec- cases of CRC and 49,700 CRC-related deaths will tiveness of CRC screening. occur in the United States in 2015.1 The use of CRC The purpose of this study was to conduct screening tests can reduce the incidence and mor- a systematic review, similar to the review of tality of the disease through early detection so that Lansdorp-Vogelaar et al,7 of the cost-effectiveness the cancer can be removed before it progresses.2-5 analyses of various CRC screening strategies. In con- Guidelines from the US Preventive Services Task trast to the review of Lansorp-Vogelaar et al,7 this Force (USPSTF), which were last updated in 2008 current review focuses on cost-effectiveness analyses and are in the process of being revised, recommend conducted in the United States alone since May 2007. screening for CRC using fecal occult blood testing, Our review examined the following questions: sigmoidoscopy, or colonoscopy starting at 50 years 1. How does the cost effectiveness of current of age and ending at 75 years of age.6 According CRC screening strategies compare with no to these guidelines, evidence is insufficient to de- screening? termine whether virtual colonoscopy, also called 2. Is there agreement as to the cost effective- computed tomography (CT) colonography, or stool ness of different screening strategies? DNA testing is a cost-effective screening modality.6 Materials and Methods Identification of Cost-Effectiveness Analyses From the University of Alabama at Birmingham School of Medicine The method for identifying studies and extracting (SSP) and the Department of Health Care Organization and Policy (MLK), University of Alabama at Birmingham School of Public data from the studies was described in previous stud- Health, Birmingham, Alabama. ies conducting economic analyses of CRC screening Address correspondence to Meredith L. Kilgore, PhD, 1530 tests.7,8 PubMed, the Cost-Effectiveness Analysis Reg- Third Avenue South, RPHB 330, Birmingham, AL 35294-0022. istry, and the York Database of Abstracts of Reviews E-mail: [email protected] of Effects were searched for cost-effectiveness stud- Submitted September 16, 2014; accepted January 6, 2015. ies on CRC screening strategies, focusing on anal- No significant relationships exist between the authors and the companies/organizations whose products or services may be yses published between May 2007 and February referenced in this article. 2014. The literature search was limited to the last

248 Cancer Control April 2015, Vol. 22, No. 2 7 years to assess the cost effectiveness of the most We evaluated the following strategies for this question: up-to-date screening strategies used in the clinical setting. • No screening (The review by Lansorp-Vogelaar et al7 assessed stud- • Annual fecal occult blood test ies up to January 2010.) (Hemoccult II and Hemoccult SENSA In PubMed, we searched for terms such as “col- [Beckman Coulter]) orectal screening” and “cost analysis,” and limited • Annual fecal immunochemical test our search to English-language studies published • Sigmoidoscopy every 5 or 10 years May 2007 or later. This search identified 529 studies. • Combination sigmoidoscopy every 5 years The Cost-Effectiveness Analysis Registry was searched and annual fecal occult blood test or annual for the term “colorectal cancer screening,” for which fecal immunochemical test 5 studies were identified with the same limitations • Colonoscopy every 10 years as mentioned above. The York Database of Abstracts • Colonoscopy with various time intervals of Reviews of Effects was searched using the search and age of first colonoscopy terms “colorectal” and “cost,” for which 3 additional • Virtual colonoscopy every 5 or 10 years studies were identified using the same limitations. • Virtual colonoscopy every 5 years with The reference list of the review by Lansdrop-Vo- 2- or 3-dimensional imaging gelaar et al,7 which analyzed the cost effectiveness • Virtual colonoscopy every 5 years with of CRC screening tests from 23 studies, was manually 100% or 50% adherence searched to identify studies not gathered by the da- • Stool DNA test every 2, 3, or 5 years tabase searches. By evaluating information found in the titles, These screening tests were chosen due to their abstracts, and full texts of the studies gathered from inclusion in the CRC screening guidelines from the the database searches, we excluded unoriginal studies, USPSTF.6 Some studies specified whether sigmoid- those that did not analyze at least 1 of the screening oscopy was followed with biopsy or not. Screening tests for CRC we were assessing, those that lacked strategies from studies were eliminated if they were an analysis of cost-effectiveness analysis, and those not distinct (eg, individualized colonoscopies). If a that analyzed special groups ineligible for routine study did not specify the type of annual fecal oc- screening. In contrast to previous reviews, we exclud- cult blood test, then we assumed that the test was ed studies that did not analyze a US population or the annual Hemoccult II. If a study evaluated virtual report the cost in US dollars. We used these criteria colonoscopy, then cost and effectiveness estimates for exclusion because the difference in price struc- pertaining to extracolonic findings were eliminated. tures among health care systems in different countries The commonly cited threshold of $50,000/LYG was makes a cross-comparison of the screening strategies set for assessing cost effectiveness.7 uninformative. Collective agreement by the authors determined whether a study should be included if Assessing the Consistency of Study Findings there was uncertainty. We conducted a series of paired comparisons of the results reported in each of the studies included in Comparing the Cost Effectiveness of Current this review. The differences in expected costs per Screening Strategies With No Screening person and the differences in LYG per person for A paired comparison of each screening strategy with each paired comparison were examined. The ICERs no screening (eg, annual Hemoccult II [Beckman were examined for nondominated strategies using the Coulter, Brea, California] vs no screening) across each following equation: of the studies was conducted to answer this question. The differences in expected costs per person and Total Cost Total Cost strategy A — strategy B the differences in life-years gained (LYGs) per person ––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– for each paired comparison were examined. If one Total Effectiveness — Total Effectiveness strategy A strategy B strategy was less costly and more effective than no screening, then it was characterized as being dominant. The incremental cost-effectiveness ratios (ICERs) If strategy A was less costly and more effective were examined for nondominated strategies using the than strategy B, then strategy A was characterized following equation: as dominant. A threshold of $50,000/LYG was set for assessing Total Cost Total Cost cost effectiveness. The same strategies were similarly screening strategy — no screening ––––––––––––––––––––––––––––––––––––––––––––––––––––––––––––– evaluated as in the first study question except for Total Effectiveness — Total Effectiveness virtual colonoscopy and the stool DNA test. These screening strategy no screening 2 strategies were not included because the compar-

April 2015, Vol. 22, No. 2 Cancer Control 249 isons were limited and their associated costs were higher than other screening strategies. 543 studies identified

Results Title Selection Identification of Cost-Effectiveness Analyses Excluded 276 Studies: The search for relevant articles was initiated in Febru- 72 Non–US studies ary 2014; we were able to identify 543 studies. After 95 Non–screening intervention analyzing the title of the studies, 267 of those studies 44 Special group not eligible for routine screening were included in our analysis. After analyzing the 13 No cases of colorectal cancer abstracts of those studies, we chose 125 to include 35 Not principal modality in our review, and their full texts were analyzed. Of 16 Unoriginal studies these, 17 studies met the criteria for inclusion (Fig). 1 Duplicate study Each data element abstracted from each of the included studies can be found in Tables 1 and 2.9-26 Table 1 lists the perspective taken, models used, strat- 267 studies kept after title inspection egies evaluated, and basic results for each study.9-26 If a study used more than 1 model, then an additional Abstract Selection entry was made in Table 1 for the given study. Table 2 provides further details, particularly on the sensitivity Excluded 142 Studies: and specificity rates, costs for tests, and the types of 16 Non–US studies sensitivity analyses performed.9-26 This review includes 26 Non–screening intervention 13 studies not included in the review published in 9 Special group not eligible for routine screening 2011 by Lansdorp-Vogelaar et al.7 4 No colorectal cancer 3 Not principal modality Cost Effectiveness of Current Screening 36 Unoriginal studies Strategies vs No Screening 31 No cost-effectiveness analyses For 34% of comparisons made, CRC screening — re- 17 Adherence studies gardless of the strategy evaluated — was less costly and more effective than no screening. The high- 125 studies kept after abstract inspection est ICER, which was for stool DNA testing every 5 years, was $34,258 per life-year saved; nearly all Full-Text Selection other screening strategies had ICERs below $20,000/ LYG compared with no screening. The complete re- Excluded 108 Studies: sults of this analysis are provided in the Supplemental 25 Non–US studies Table (online).9-11,13-26 8 Non–screening intervention Hemoccult II was less costly and more effective 11 No principal modality than no screening in 89% of the comparisons made. 10 Unoriginal studies The annual Hemoccult SENSA and annual fecal immu- 50 No cost-effectiveness analyses nochemical strategies either dominated no screening or 4 Duplicate studies had an ICER below $1,000/LYG. Overall, 60% and 78% of comparisons made showed that Hemoccult SENSA and an annual fecal immunochemical test, respectively, were 17 studies included in review less costly and more effective than no screening at all. Sigmoidoscopy every 5 years either dominated no Fig. — Method for identifying studies for inclusion. screening or had an ICER below $2,000/LYG. Overall, 64% of comparisons made showed that undergoing that sigmoidoscopy every 5 years plus an annual sigmoidoscopy every 5 years was less costly and more ecal immunochemical test was less costly and more effective than no screening. Sigmoidoscopy every effective than no screening. Sigmoidoscopy every 5 years in combination with any annual fecal occult 10 years had an ICER below $10,000/LYG. blood test (annual Hemoccult II or annual Hemoc- Colonoscopy every 10 years beginning at cult SENSA) either dominated no screening or had 50 years of age either dominated no screening or an ICER below $15,000/LYG. Approximately 50% of had an ICER below $28,000/LYG. Overall, 18% of comparisons made showed that sigmoidoscopy ev- comparisons made showed that undergoing colo- ery 5 years plus any annual fecal occult blood test noscopy every 10 years was less costly and more was less costly and more effective than no screening. effective than no screening. Virtual colonoscopy Approximately 44% of comparisons made also showed every 5 years did not dominate no screening in

250 Cancer Control April 2015, Vol. 22, No. 2 Table 1. — Descriptions of Included Studies Reference Modeling Perspective Strategy Evaluated Reference Modeling Perspective Strategy Evaluated Approach Approach Dinh10 Archimedes Societal NS Parekh19 Markov Payer NS aFIT aHemSENSA SIG5 aFIT COL10 COL10 DNA3 (v 1.0) Hassan9 Markov Payer NS DNA3 (v 1.1) CTC10 DNA3 (v 1.0) Hassan11 Markov Societal NS DNA3 (v 2.0) CTC10 Pickhardt20 Markov Payer NS COL10 SIG10 Hassan12 Markov Societal SIG5 COL10 COL10 CTC10 CTC5 Pickhardt21 Markov Societal COL10 Hassan13 Markov Payer NS CTC10 COL10 CTC5 Jonas14,a Markov Payer COL10 Ramsey22 MISCAN Societal NS Societal COL10

16 COL10/40 Knudsen MISCAN Payer NS COL5/50 SimCRC aHemII COL5/40 CRC-SPIN aHemSENSA aFIT Regge23 Markov Societal NS SIG5 SIG10 SIGB5 COL10 SIGB5 + aHemII CTC10/IR/without CAD SIG5 + aHemII CTC10/ER/without CAD SIGB5 + aHemSENSA CTC10/IR/with CAD SIG5 + aHemSENSA CTC10/ER/with CAD SIGB5 + aFIT Sharaf24 Markov Payer NS SIG5 + aFIT aHemII COL10 aFIT CTC5 SIG5 Ladabaum17 Decision Payer NS COL10 analytic aHemII Vanness25 MISCAN Payer NS model aFIT SimCRC SIG5 + aHemII SIG5 SIG5 + aFIT SIG5 + aFOBT COL10 SIG5 + aFIT CTC5 COL10 CTC10 DNA2 (SEPT9-2-well) DNA2 (SEPT9-3-well) Vijan26 Markov Payer NS aHemII Lansdorp- MISCAN Payer NS SIG5 Vogelaar18 SimCRC aHemII SIG5 + aHemII aHemSENSA COL10 aFIT CTC5/2D SIG5 CT10/2D SIGB5 CTC5/3D SIGB5 + aHemII CTC10/3D SIG5 + aHemII SIGB5 + aHemSENSA SIG5 + aHemSENSA SIGB5 + aFIT SIG5 + aFIT COL10 DNA3 (v 1.0) DNA3 (v 1.1)

aThis study also included data from Sonnenberg et al.15 aFIT = annual fecal immunochemical test, aHemII = annual Hemoccult II, aHemSENSA = annual Hemoccult SENSA, COL10 = colonoscopy every 10 y, COL 10/40 = colonoscopy every 10 y starting at age 40 y, CRC-SPIN = colorectal cancer–simulated population model for incidence and natural history, CTC5 = computed tomography colonography every 5 y, CTC10 = computed tomography colonography every 10 y, CTC10/2D = 2-dimensional computed tomography colonography every 10 y, CTC10/3D = 3-dimensional computed tomography colonography every 10 y, CTC10/ER/with CAD = computed tomography colonography with a computer-aided design every 10 y (read by an experienced radiologist), CTC10/IR/with CAD = computed tomography colonography with a computer-aided design every 10 y (read by an inexperienced radiologist), DNA2 (SEPT9-2-well) = stool DNA test using 2-well methylated Septin 9 DNA assays every 2 y, DNA3 = stool DNA test every 3 y, DNA5 = stool DNA test every 5 y, MISCAN = microsimulation screening analysis, NS = no screening, SIG5 = sigmoidoscopy every 5 y, SIG10 = sigmoidoscopy every 10 y, SIGB5 = sigmoidoscopy with biopsy every 5 y, SimCRC = simulation model of colorectal cancer.

April 2015, Vol. 22, No. 2 Cancer Control 251 Table 2. — Model Inputs for Each Study Included Table continued on next 3 pages Characteristic Dinh10 Hassan9 Hassan11 Hassan12 Hassan13 Jonas14,a Knudsen16 Ladabaum17 Lansdorp-Vogelaar18 Age for screening, y 50–75 50–80 50–80 50–80 50–80 50–80 65–80 50–80 65–80 Test characteristics Sensitivity rate of HemII, % ≤ 5-mm polyps NR NR NR NR NR NR 2 5 2 6–9-mm polyps NR NR NR NR NR NR 5 5 5 ≥ 10-mm polyps NR NR NR NR NR NR 12 11 12 CRC NR NR NR NR NR NR 40 40 40 Sensitivity rate of HemSENSA, % ≤ 5-mm polyps NR NR NR NR NR NR 7 NR 7.5 6–9-mm polyps NR NR NR NR NR NR 12 NR 12.4 ≥ 10-mm polyps NR NR NR NR NR NR 24 NR 23.9 CRC NR NR NR NR NR NR 70 NR 70 Sensitivity rate of FIT, % ≤ 5-mm polyps NR NR NR NR NR NR 5 10 5 6–9-mm polyps NR NR NR NR NR NR 10 10 10 ≥ 10-mm polyps 25 NR NR NR NR NR 22 24 22 CRC 66 NR NR NR NR NR 70 70 70 Sensitivity rate of SIG, % ≤ 5-mm polyps NR NR NR 45 NR NR 75 85 75 6–9-mm polyps NR NR NR 65 NR NR 85 85 85 ≥ 10-mm polyps NR NR NR NR NR NR 95 90 95 CRC NR NR NR 65 NR NR 95 95 95 Sensitivity rate of COL, % ≤ 5-mm polyps 75 NR 80 80 80 NR 75 85 75 6–9-mm polyps 85 NR 85 85 85 NR 85 85 85 ≥ 10-mm polyps 95 NR 90 90 90 NR 95 90 95 CRC NR NR 95 95 95 NR 95 95 95 Sensitivity rate of CTC, % ≤ 5-mm polyps NR 0 0 NR NR 80 NR NR NR 6–9-mm polyps NR 70 70 80 NR 80 57–84 NR NR ≥ 10-mm polyps NR 85 85 91 NR 80 84–92 NR NR CRC NR 95 95 95 NR NR 84–92 NR NR Sensitivity rate of DNA stool test, % ≤ 5-mm polyps NR NR NR NR NR NR NR 7 (2-well) 4-5 6–9-mm polyps NR NR NR NR NR NR NR 7 (2-well) 12 ≥ 10-mm polyps NR NR NR NR NR NR NR 11 (2-well) 15–43 14 (3-well) CRC NR NR NR NR NR NR NR Varied 52–70 Specificity rate, % HemII NR NR NR NR NR NR 98 97 98 HemSENSA NR NR NR NR NR NR 93 NR 92.5 FIT 97 NR NR NR NR NR 95 95 95 SIG NR NR NR 90 NR NR 92–100 92 92 COL 95 NR 99 99 99 NR 90 NR 90 CTC NR 86 86 84 NR 95 80–88 NR NR hDNA stool test NR NR NR NR NR NR NR 91 (2-well) 95–96 88 (3-well)

252 Cancer Control April 2015, Vol. 22, No. 2 Table 2. — Model Inputs for Each Study Included (cont) Characteristic Parekh19 Pickhardt20 Pickhardt21 Ramsey22 Regge23 Sharaf24 Vanness25 Vijan26 Age for screening, y 50–80 50–80 65–80 Varies 50–80 50–80 50–80 50–80 Test characteristics Sensitivity rate of HemII, % ≤ 5-mm polyps NR NR NR NR NR 5 5 5 6–9-mm polyps NR NR NR NR NR 5 9.7–11.1 5 ≥ 10-mm polyps 10 NR NR NR NR 11 20.4–24.1 5 CRC 40 NR NR NR NR 40 70 30–50 Sensitivity rate of HemSENSA, % ≤ 5-mm polyps NR NR NR NR NR NR NR NR 6–9-mm polyps NR NR NR NR NR NR NR NR ≥ 10-mm polyps NR NR NR NR NR NR NR NR CRC NR NR NR NR NR NR NR NR Sensitivity rate of FIT, % ≤ 5-mm polyps NR NR NR NR NR 10 7.5 NR 6–9-mm polyps NR NR NR NR NR 10 12–13.4 NR ≥ 10-mm polyps 40 NR NR NR NR 24 22.2–25.7 NR CRC 76 NR NR NR NR 70 70 NR Sensitivity rate of SIG, % ≤ 5-mm polyps NR 45 NR NR 45 85 75 85 6–9-mm polyps NR 45 NR NR 45 85 85 85 ≥ 10-mm polyps NR 60–65 NR NR 60–66 90 95 95 CRC NR 60–65 NR NR 60–66 95 75–95 95 Sensitivity rate of COL, % ≤ 5-mm polyps Small polyp 80 80 80 80 85 75 85 6–9-mm polyps < 10-mm: 85 85 85 85 85 85 85 85 ≥ 10-mm polyps 90 90 90 95 90 90 95 95 CRC 95 95 95 95 95 95 75–95 95 Sensitivity rate of CTC, % ≤ 5-mm polyps NR 48 0 NR 0 NR 17.2–54.5 33–46 6–9-mm polyps NR 70 70 NR 49.5–88.0 NR 61.5–66.7 50–83 ≥ 10-mm polyps NR 85 90 NR 64.8–91.0 NR 80.5–94.7 82–91 CRC NR 95 95 NR 95 NR 80.5–94.7 Sensitivity rate of DNA stool test, % ≤ 5-mm polyps NR NR NR NR NR NR NR NR 6–9-mm polyps NR NR NR NR NR NR NR NR ≥ 10-mm polyps 18 NR NR NR NR NR NR NR

CRC 52–88 NR NR NR NR NR NR NR Specificity rate, % HemII 92 NR NR NR NR 97 95 97.5 HemSENSA NR NR NR NR NR NR NR NR FIT 91 NR NR NR NR 95 92.5 NR SIG NR 90 NR NR 90 92 92 NR COL NR 90 99 100 99 NR 90 NR CTC NR 86 86 NR 81.0–88.7 NR 90.9 91 DNA stool test 82–94 NR NR NR NR NR NR NR

April 2015, Vol. 22, No. 2 Cancer Control 253 Table 2. — Model Inputs for Each Study Included (cont) Characteristic Dinh10 Hassan9 Hassan11 Hassan12 Hassan13 Jonas14,a Knudsen16 Ladabaum17 Lansdorp- Vogelaar18 Costs (updated to 2012 US dollars) Type of cost Direct Direct and Direct and Direct and Direct and Direct Direct Direct Direct included indirect indirect indirect indirect Source of Medicare Medicare Medicare Medicare Medicare US Health Care Medicare Medicare Medicare screening cost Financing information Administration HemII, $ NR NR NR NR NR NR 5.54 23 5.54 HemSENSA, $ NR NR NR NR NR NR 5.54 NR 5.54 FIT, $ 22.00 NR NR NR NR NR 24.36 23 24.36 SIG, $ 161.00 NR NR 223.94 NR NR NR 169 NR COL, $ 498.00 1,218.05 998.78 671.82 643.04 855.83 551.45 645 551.45 CTC, $ NR 901.36 757.34 709.14 NR 561.94 540.37 NR NR DNA stool NR NR NR NR NR NR NR 150 (SEPT9) test, $ Polypectomy, $ 150.00 1,647.69 1,440.66 986.40 1,046.21 1,339.00 718.65 1,013 Treatment of NR NR 18,654.51 17,467.24 16,718.98 NR 13,781.71 14,258 perforation, $ Indirect costs, $ HemII NR NR NR NR NR NR NR NR NR HemSENSA NR NR NR NR NR NR NR NR NR FIT NR NR NR NR NR NR NR NR NR SIG NR NR NR 79.98 NR NR NR NR NR COL NR NR 253.97 223.94 214.35 NR NR NR NR CTC NR NR 169.69 79.98 NR NR NR NR NR Source of Medicare Medicare Medicare Medicare Medicare US Health Care Medicare Medicare Medicare CRC care cost Financing information Administration Localized, $ — 57,359.19 58,992.90 53,318.81 53,076.11 Terminal care Variesb Variesb Variesb per patient with CRC: 40,087.84 Regional, $ Variesb 84,710.00 87,122.72 106,637.62 77,572.78 — — — — Distant, $ 88,585.62 91,108.73 213,275.24 81,655.56 — — — — Measured in 2012 2007 2006 2008 2011 2005 2007 2010 2007 US dollars from what year Measurement Quality- LGA LGA LGA LGA Life-y saved LGA Quality- LGA of effectiveness adjusted adjusted LGA LGA Sensitivity 1-way Monte Carlo Monte Carlo Probabilistic Systematic Tested Tested Tested Tested analysis sensitivity simulation; simulation (Monte Carlo sensitivity different wage multiple multiple multiple analysis systematic and simulation) analysis rates and time variables variables variables sensitivity systematic costs and simulation intervals any study, but it had an ICER below $22,000/ Cost Effectiveness of Different LYG. Virtual colonoscopy every 10 years domi- Screening Strategies nated no screening in 1 study,9 and the data from When annual fecal immunochemical testing was com- the rest of the studies reported an ICER below pared with annual Hemoccult II testing, annual fecal $23,000/LYG. Stool DNA testing every 3 years or immunochemical testing was either dominant or had 5 years did not dominate a no-screening strategy an ICER below $50,000/LYG in 100% of the simula- in any study, but these 2 strategies had an ICER tions from 5 different studies (Table 3). By contrast, below $35,000/LYG. annual Hemoccult SENSA testing dominated annual

254 Cancer Control April 2015, Vol. 22, No. 2 Table 2. — Model Inputs for Each Study Included (cont) Characteristic Parekh19 Pickhardt20 Pickhardt21 Ramsey22 Regge23 Sharaf24 Vanness25 Vijan26

Costs (updated to 2012 US dollars) Type of cost Direct Direct Direct and Direct and Direct and Direct Direct Direct included indirect indirect indirect Source of Medicare 1 article Medicare Medicare Medicare Medicare Medicare Medicare screening cost information HemII, $ 17.08 NR NR NR NR 24.22 5.03 22.46 HemSENSA, $ NR NR NR NR NR NR NR NR FIT, $ 25.05 NR NR NR NR 24.22 24.60 NR SIG, $ NR 456.68 NR NR 346.59 NR 237.83 485.39 COL, $ 1,047.75 792.65 847.77 NR 971.12 679.13 802.52 814.81 CTC, $ NR 544.37 718.74 NR 736.37 NR 569.62 697.52 DNA stool 341.66 NR NR NR NR NR NR NR test, $ Polypectomy, $ 1,537.46 1,297.16 1,043.98 NR 1,400.76 1,066.60 960.83 222.11 Treatment of NR 14,805.17 13,555.77 NR 18,137.91 15,012.43 16,272.81 27,787.11 perforation, $ Indirect costs, $ HemII NR NR NR NR NR NR NR NR HemSENSA NR NR NR NR NR NR NR NR FIT NR NR NR NR NR NR NR NR SIG NR NR NR NR 83.05 NR NR NR COL NR NR NR NR 246.93 NR NR NR CTC NR NR NR NR 83.05 NR NR NR Source of 9 articles 1 article Medicare 5 articles Medicare Medicare Medicare Medicare CRC care cost information Localized, $ 58,081.82 51,508.32 55,238.29 — 57,359.19 — — 42,799.32

Regional, $ 111,608.19 — 81,577.78 Variesb 84,710.00 Variesb Variesb 58,933.29 Distant, $ 227,771.83 — 85,310.10 — 88,585.62 — — 51,908.21 Measured in 2006 NR 2008 2005 2007 2010 2007 2003 US dollars from what year Measurement LGA LGA LGA LGA Life-y saved Quality- LGA LGA of effectiveness adjusted LGA Sensitivity 1- and 1-way Monte Carlo 1- and Monte Carlo Monte Carlo Adherence Monte Carlo analysis 2-way (10,000 multiway (10,000 (10,000 rates 50% (10,000 simulation) simulation) and simulation) vs 100% for simulation) and systematic analysis each test and 1-way systematic on all variables in the model aThis study also included data from Sonnenberg et al.15 bCRC care depends on stage of cancer and phase of care. COL = colonoscopy, CRC = colorectal cancer, CTC = computed tomography colonography, FIT = fecal immunochemical test, HemII = Hemoccult II, Hem- SENSA = Hemoccult SENSA, LGA = life-y gained, NR = not recorded, SEPT9 = DNA SEPT9 analysis, SIG = sigmoidoscopy. fecal immunochemical testing in 100% of the simula- py every 5 years dominated in 100% of the simulations from 2 different studies (see Table 3). The ma- tions. When compared with annual Hemoccult II, jority of comparisons made between sigmoidoscopy sigmoidoscopy every 5 years had an ICER below every 5 years and annual Hemoccult SENSA or fecal $50,000/LYG in 70% of the simulations from 5 different immunochemical testing showed that sigmoidosco- studies (see Table 3).

April 2015, Vol. 22, No. 2 Cancer Control 255 When sigmoidoscopy every 5 years plus any $50,000/LYG in 74% of the simulations when com- annual fecal occult blood test (Hemoccult II, Hemoc- pared with any type of annual fecal blood testing. cult SENSA, fecal immunochemical test) was com- When compared with sigmoidoscopy every 5 years, pared with annual fecal occult blood testing alone colonoscopy every 10 years either dominated or had or sigmoidoscopy alone every 5 years, sigmoidosco- an ICER below $50,000/LYG in 87% of the simula- py every 5 years plus any annual fecal occult blood tions. When compared with sigmoidoscopy every test had an ICER below $50,000/LYG or dominat- 5 years plus annual Hemoccult II or Hemoccult SENSA ed the other strategy in 99% of the simulations. The testing, colonoscopy every 10 years dominated or had majority of simulations showed that sigmoidoscopy an ICER below $50,000/LYG in 53% of the simulations. every 5 years plus annual Hemoccult SENSA testing Colonoscopy every 10 years was dominated or was either dominated or had an ICER below $50,000/LYG not cost effective in 35% of the simulations when com- when compared with sigmoidoscopy every 5 years pared with sigmoidoscopy every 5 years plus annual plus annual Hemoccult II testing. When comparing fecal immunochemical testing. sigmoidoscopy every 5 years plus annual fecal immunochemical testing with sigmoidoscopy every 5 years Discussion plus either annual Hemoccult II or Hemoccult SENSA Using US data published in 17 studies since May 2007, testing, the simulations showed that sigmoidoscopy this review reaffirms the results from previous re- plus annual fecal immunochemical testing had an views, such as the 2011 review by Lansdorp-Vogelaar ICER below $50,000/LYG in 27% of the simulations. et al,7 that any fecal occult blood test, sigmoidoscopy, Colonoscopy every 10 years had an ICER below colonoscopy, virtual colonoscopy, or stool DNA test

Table 3. — Summary of Comparison of Screening Strategies Strategy No. No. No. No. No. No. of Studiesa of Simulationsb Dominantc < $50,000/LYGd > $50,000/LYGd Dominatedc COL10 vs aHemII 6 9 0 6 3 0 COL10 vs aHemSENSA 2 5 0 5 0 0 COL10 vs aFIT 6 9 0 6 0 3 COL10 vs SIG5/SIGB5 7 15 4 9 2 0 COL10 vs SIG5/SIGB5 + aHemII/aHemSENSAe 5 28 4 11 3 7 SIG5/SIGB5 + aFIT vs COL10 4 17 3 3 4 7 aFIT vs aHemII 5 8 3 5 0 0 aFIT vs aHemSENSA 2 5 0 0 0 5 SIG5/SIGB5 vs aHemII 5 13 0 9 4 0 SIG5/SIGB5 vs aHemSENSA 2 10 0 0 0 10 SIG5/SIGB5 vs aFIT 5 13 0 0 0 13 SIG5/SIGB5 + aHemII/aHemSENSA 4 42 2 40 0 0 vs aHemII/aHemSENSA SIG5/SIGB5 + aFIT vs aHemII/aHemSENSA 3 21 0 21 0 0 SIG5/SIGB5 + aHemII/aHem SENSA vs aFIT 3 21 4 16 1 0 SIG5/SIGB5 + aFIT vs aFIT 3 11 0 10 1 0 SIG5/SIGB5 + aHemII/aHemSENSA 4 42 24 18 0 0 vs SIG5/SIGB5 SIG5/SIGB5 + aFIT vs SIG5/SIGB5 3 21 1 20 0 0 SIG5/SIGB5 + aHemSENSA vs SIG5/SIGB5 2 20 1 19 0 0 vs aHemII SIG5/SIGB5 + aFIT vs SIG5/SIGB5 4 47 0 13 16 18 + aHemII/aHemSENSA aNumber of studies evaluating a given comparison. bNumber of simulations in which model simulations were conducted for each comparison. cNumber of dominant = given a X vs Y comparison, the number of simulations resulting in X being dominant to Y. dNumber of studies with less than or more than $50,000/LYG = given a X vs Y comparison, the number of simulations resulting in X having an incremental cost effectiveness ratio less than or greater than $50,000/LYG when compared with Y. eThree simulations showed COL10 to be less costly and less effective than SIG + aFOBT. aFIT = annual fecal immunochemical test, aFOBT = annual fecal occult blood test, aHemII = annual Hemoccult II, aHemSENSA = annual Hemoccult SENSA, COL10 = colonoscopy every 10 y, LYG = life-years gained, SIG = sigmoidoscopy, SIG5 = sigmoidoscopy every 5 y, SIGB5 = sigmoidoscopy with biopsy every 5 y.

256 Cancer Control April 2015, Vol. 22, No. 2 is cost effective when compared with no screening. each screening strategy has its own advantages. Colo- Evidence suggests any of the current screening strat- noscopy is more sensitive and allows access to the egies will reduce mortality when compared with rectum and the entire colon, whereas sigmoidoscopy no screening. For 34% of the comparisons made, has fewer risks of complications and does not require CRC screening — regardless of the strategy evalu- sedation or extensive bowel preparation. ated — was less costly and more effective than no screening. The highest ICER calculated was approxi- Limitations mately $35,000/LYG, and the corresponding screening Data extracted from each study came from simula- strategy was stool DNA test (v 1.0) every 5 years. tions that assumed 100% adherence of using each This strategy was still below the typical cost-effec- screening strategy. Clinically, this does not reflect the tiveness threshold. actual adherence rate. An important gap in the litera- In addition, the current review confirms that un- ture is the need to incorporate compliance rates into certainty still exists as to which screening strategy cost-effectiveness estimations. It is likely that different is optimal in terms of cost effectiveness. Potential screening strategies should be recommended to dif- reasons for the disagreement among studies as to ferent patients, based on predicted adherence rates, which screening strategy is the most cost effective or that health care professionals present their patients have been explained by previous reviews on the cost with different options for screening strategies. effectiveness of CRC screening strategies.7,9 Each study Another limitation of this study is that we includ- uses a separate model to simulate the natural history ed studies of US patients alone and did not include of CRC. However, Knudsen et al16 used 3 different international study populations. Due to the relative models — microsimulation screening analysis, a simu- differences in prices between countries, the cost-ef- lation model of CRC, and a CRC-simulated population fectiveness data from this review cannot be applied model for incidence and natural history — in their to other CRC screening strategies in other countries. cost-effective analysis. The calculated ICERs between models were similar regardless of screening strate- Conclusions gy; microsimulation screening analysis had a slightly Solid agreement exists among all the studies reviewed higher ICER. In general, the decision analysis model that colorectal cancer screening is either dominant used by Ladabaum et al17 reported results similar to or cost effective compared with no screening at other studies that used the Markov and Archimedes all, regardless of the screening strategy employed. models. The use of different model inputs, such as We found considerable variation concerning which dwell time of precancerous and cancerous phases, screening strategies would be preferred, subject to a screening test costs, or screening adherence, may be prespecified level of cost per life-year gained deemed potential factors for disagreement among studies. acceptable. The sensitivity and specificity rates of screening tests also vary among studies, thus posing yet another References potential reason for different results. 1. American Cancer Society. Cancer Facts & Figures 2015. Atlanta, GA: American Cancer Society; 2015. http://www.cancer.org/acs/groups/ It is clear that sigmoidoscopy every 5 years plus content/@editorial/documents/document/acspc-044552.pdf. Accessed annual fecal occult blood testing is cost effective February 12, 2015. 2. Elmunzer BJ, Hayward RA, Schoenfeld PS, et al. Effect of flexible sig- when compared with either strategy alone. Colonos- moidoscopy-based screening on incidence and mortality of colorectal cancer: copy every 10 years is cost effective when compared a systematic review and meta-analysis of randomized controlled trials. PLoS Med. 2012;9(12):e1001352. with annual fecal occult blood testing or sigmoidos- 3. Libby G, Brewster DH, McClements PL, et al. The impact of popula- copy every 5 years. However, the results are mixed tion-based faecal occult blood test screening on colorectal cancer mortality: a matched cohort study. Br J Cancer. 2012;107(2):255-259. when colonoscopy every 10 years is compared with 4. Schoen RE, Pinsky PF, Weissfeld JL, et al; PLCO Project Team. Col- sigmoidoscopy every 5 years plus any annual fecal orectal-cancer incidence and mortality with screening flexible sigmoidoscopy. N Engl J Med. 2012;366(25):2345-2357. occult blood test. 5. Young PE, Womeldorph CM. Colonoscopy for colorectal cancer screen- Patients aged 50 years who have an average risk ing. J Cancer. 2013;4(3):217-226. 6. US Preventive Services Task Force. Colorectal cancer: screening. of CRC must be advised by their health care profes- Published October 2008. http://www.uspreventiveservicestaskforce.org/ sional as to which screening strategy is cost effective. Page/Topic/recommendation-summary/colorectal-cancer-screening. Accessed January 21, 2015. Based on this review, any screening strategy is more 7. Lansdorp-Vogelaar I, Knudsen AB, Brenner H. Cost-effectiveness of cost effective compared with no CRC screening, and colorectal cancer screening. Epidemiol Rev. 2011;33(1):88-100. 8. Pignone M, Saha S, Hoerger T, et al. Challenges in systematic reviews high-quality evidence supports this recommendation. of economic analyses. Ann Intern Med. 2005;142(12 pt 2):1073-1079. Furthermore, colonoscopy every 10 years or sigmoid- 9. Hassan C, Pickhardt PJ, Laghi A, et al. Impact of whole-body CT screening on the cost-effectiveness of CT colonography. Radiology. 2009;251(1):156-165. oscopy every 5 years plus any annual fecal occult 10. Dinh T, Ladabaum U, Alperin P, et al. Health benefits and cost-effective- blood test should be the first 2 screening strategies ness of a hybrid screening strategy for colorectal cancer. Clin Gastroenterol Hepatol. 2013;11(9):1158-1166. recommended to patients, but evidence is of moderate 11. Hassan C, Pickhardt PJ, Laghi A, et al. Computed tomographic colo- quality to support this recommendation. Obviously, nography to screen for colorectal cancer, extracolonic cancer, and aortic an-

April 2015, Vol. 22, No. 2 Cancer Control 257 eurysm: model simulation with cost-effectiveness analysis. Arch Intern Med. 2008;168(7):696-705. 12. Hassan C, Hunink MG, Laghi A, et al. Value-of-information analysis to guide future research in colorectal cancer screening. Radiology. 2009; 253(3):745-752. 13. Hassan C, Rex DK, Zullo A, Cooper GS. Loss of efficacy and cost-effectiveness when screening colonoscopy is performed by nongastroenterologists. Cancer. 2012;118(18):4404-4411. 14. Jonas DE, Russell LB, Sandler RS, et al. Value of patient time invested in the colonoscopy screening process: time requirements for colonoscopy study. Med Decis Making. 2008;2008:28(1):56-65. 15. Sonnenberg A, Delcò F, Bauerfeind P. Is virtual colonoscopy a cost-effective option to screen for colorectal cancer? Am J Gastroenterol. 1999;94(8):2268-2274. 16. Knudsen AB, Lansdorp-Vogelaar I, Rutter CM, et al. Cost-effectiveness of computed tomographic colonography screening for colorectal cancer in the Medicare population. J Natl Cancer Inst. 2010;102(16):1238-1252. 17. Ladabaum U, Allen J, Wandell M, et al. Colorectal cancer screening with blood-based biomarkers: cost-effectiveness of methylated septin 9 DNA versus current strategies. Cancer Epidemiol Biomarkers Prev. 2013;22(9):1567-1576. 18. Lansdorp-Vogelaar I, Kuntz KM, Knudsen AB, et al. Stool DNA testing to screen for colorectal cancer in the Medicare population: a cost-effectiveness analysis. Ann Intern Med. 2010;153(6):368-377. 19. Parekh M, Fendrick AM, Ladabaum U. As tests evolve and costs of cancer care rise: reappraising stool-based screening for colorectal neoplasia. Aliment Pharmacol Ther. 2008;27(8):697-712. 20. Pickhardt PJ, Hassan C, Laghi A, et al. Cost-effectiveness of colorectal cancer screening with computed tomography colonography: the impact of not reporting diminutive lesions. Cancer. 2007;109(11):2213-2221. 21. Pickhardt PJ, Hassan C, Laghi A, et al. CT colonography to screen for colorectal cancer and aortic aneurysm in the Medicare population: cost-effectiveness analysis. AJR Am J Roentgenol. 2009;192(5):1332-13400. 22. Ramsey SD, Wilschut J, Boer R, et al. A decision-analytic evaluation of the cost-effectiveness of family history-based colorectal cancer screening programs. Am J Gastroenterol. 2010;105(8):1861-1869. 23. Regge D, Hassan C, Pickhardt PJ, et al. Impact of computer-aided detection on the cost-effectiveness of CT colonography. Radiology. 2009;250(2):488-497. 24. Sharaf RN, Ladabaum U. Comparative effectiveness and cost-effectiveness of screening colonoscopy vs. sigmoidoscopy and alternative strategies. Am J Gastroenterol. 2013;108(1):120-132. 25. Vanness DJ, Knudsen AB, Lansdorp-Vogelaar I, et al. Comparative economic evaluation of data from the ACRIN National CT Colonography Trial with three cancer intervention and surveillance modeling network microsim- ulations. Radiology. 2011;261(2):487-498. 26. Vijan S, Hwang I, Inadomi J. The cost-effectiveness of CT colonography in screening for colorectal neoplasia. Am J Gastroenterol. 2007;102(2):380-390.

258 Cancer Control April 2015, Vol. 22, No. 2 Editorial

Colorectal Cancer: The Last Nail in the Coffin for Androgen Deprivation Therapy for Prostate Cancer?

In this issue of Cancer Control, the Special Report by underwent ADT as secondary treatment. Thus, the authors Lu and colleagues suggests that androgen deprivation divided patients from the registries into 2 cohorts, therapy (ADT) may be associated with an increased ie, those diagnosed between 1960 and 1980 and incidence of colorectal cancer (CRC). We elected to those diagnosed between 1981 and 2008. Finally, no publish this provocative paper for 2 reasons: (1) It consideration was given to the fact that male life ex- supports the results of an analysis of Surveillance, pectancy — and, hence, the possibility to diagnose new Epidemiology, and End Results Program data with the cases of CRC — has increased in Sweden since 1958.8 weight of a Swedish cancer registry, which includes Despite these limitations, the incidence of CRC in all cases of prostate cancer diagnosed since 1958 in Swedish men with prostate cancer has increased since Sweden,1 and (2) it is a reminder that some common 1980 and the results of this Special Report suggest and presumably innocuous interventions may have that ADT is a likely culprit of this increase. Thus, we long-term complications. The association of CRC and felt that these conclusions were important and robust ADT may, in fact, be the last nail in the coffin for the enough to justify the publication of this Special Report use of ADT in individuals with recurrences of pros- by Lu and colleagues in the hopes that we can open tate-specific antigen and without imaging evidence of a lively debate of the issue. metastatic disease. In these settings, the use of ADT Is this study practice changing? Definitely not. does not improve survival rates, but its use has been It presents one more reason to avoid the use of ADT associated with other serious complications, including when it is not indicated, but it does not offer reasons the loss of libido, hot flashes, fatigue, sarcopenia, oste- to avoid ADT in conditions in which there are evi- oporosis, diabetes, and an increased risk of infarction dence-based benefits to this type of treatment. Such in the setting of coronary artery disease.2,3 situations include using ADT as initial treatment for A hypothesis suggested by Lu and colleagues is patients with metastatic prostate cancer in combina- the possibility that estrogen therapy, which was wide- tion with radiotherapy in patients with locally ad- ly used in Sweden prior to the introduction of go- vanced disease or in those at high risk of recurrence, nadotropin-releasing hormone therapy, could reduce as well as adjuvant treatment following prostatectomy the risk of CRC. Undoubtedly, this article will elicit in patients with involved pelvic lymph nodes.3 controversy, and Lu and colleagues have thus detailed This Special Report is a reminder of the need to the limitations and weaknesses of their Special Report. monitor long-term treatments throughout a patient’s life- The authors use 3 sources of information: (1) a time, and this is particularly true of therapeutic options Swedish cancer registry initiated in 1958 that includes that may appear innocuous but may cause unwanted nearly all cases of cancers diagnosed in Sweden, long-term consequences. Currently, vitamin D deficiency (2) a Swedish patient registry that began in 1964 for is considered to be the cause of a variety of conditions, 2 counties and was then extended to the entire coun- including cancer, hematological disorders, and bone and try in 1987 and includes surgical (but not medical) cardiovascular diseases.8 As clinicians, we would be well treatments, and (3) the Swedish total population reg- advised to monitor the long-term effects of ongoing istry that includes demographic data for the entire vitamin D supplementation. In yet another example of Swedish population since 1961. Prior to 1980, Swed- the importance of monitoring long-term treatments, we ish patients with prostate cancer were treated with now know that vitamin E supplementation might have estrogen therapy but with orchiectomy, gonadotro- facilitated the development of prostate and lung can- pin-releasing hormone therapy, or other treatments cers — the exact types of cancer it was purported to after 1980.4-7 Furthermore, the patient registry did not prevent9! With the rising life expectancy of the global cover the entire Swedish population until 1987, so the community (the global life expectancy has increased by authors infer that it is unclear whether patients under- 6 years since 19908), clinical epidemiology has become went prostatectomy or orchiectomy prior to that date. an invaluable instrument to detect the long-term effects It is also unclear how many patients received radio- of medical intervention. therapy in lieu of prostatectomy or how many expe- This Special Report is also an opportunity to revisit rienced a recurrence following local treatment and a turning point in the management of prostate cancer.

April 2015, Vol. 22, No. 2 Cancer Control 259 Since the 1984 report of the Leuprolide Study Group References suggesting that ADT in combination with leuprolide 1. Gillessen S, Templeton A, Marra G, et al. Risk of colorectal cancer in men on long-term androgen deprivation therapy for prostate cancer. J Natl and a daily dose of 3 mg of diethylstilbestrol (DES), a Cancer Inst. 2010;102(23):1760-1770. synthetic estrogen, were comparable treatment options 2. Droz JP, Aapro M, Balducci L, et al. Management of prostate cancer 10 in older patients: updated recommendations of a working group of the Inter- for the management of metastatic prostate cancer, es- national Society of Geriatric Oncology. Lancet Oncol. 2014;15(9):e404-e414. trogen therapy was all but banned from US practice. 3. Saylor PJ, Smith MR. Metabolic complications of androgen deprivation therapy for prostate cancer. J Urol. 2013;189(1 suppl):S34-S44. In our opinion, this approach might have been prema- 4. Giertz G. Urology in Sweden: 1940-1990 [in Swedish]. Sydsven ture and ill advised. Estrogen therapy is associated with Medicinhist Sallsk Arssk. 1996;(suppl 21):1-284. 5. McLeod DG. Hormonal therapy: historical perspective to future directions. an increased risk of deep venous thrombosis (DVT), Urology. 2003;61(2 suppl 1):3-7. whereas leuprolide use increases the incidence of hot 6. Adolfsson J, Garmo H, Varenhorst E, et al. Clinical characteristics and 10 primary treatment of prostate cancer in Sweden between 1996 and 2005. flashes ; however, at the time, the cost of estrogen was Scand J Urol Nephrol. 2007;41(6):456-477. less than 1% of the cost of leuprolide. Furthermore, a 7. Cox RL, Crawford ED. Estrogens in the treatment of prostate cancer. J Urol. 1995;154(6):1991-1998. study comparing leuprolide with a daily dose of 1 mg 8. World Health Organization (WHO). World Health Statistics 2014. DES was never performed. Some evidence suggests that Geneva, Switzerland: WHO; 2014. http://apps.who.int/iris/bitstram/10665/ 112738/1/9789240692671_eng.pdf?ua=1. Accessed February 2, 2015. 1 mg and 3 mg DES may be equivalent but that the 9. Cardenas E, Ghosh R. Vitamin E: a dark horse at the crossroad of lower dose may be associated with a lower risk of DVT.11 cancer management. Biochem Pharmacol. 2013;86(7):845-852. 10. The Leuprolide Study Group. Leuprolide versus diethylstilbestrol for Finally, estrogens do not cause many of the complica- metastatic prostate cancer. N Engl J Med. 1984;311(20):1281-1286. tions of gonadotropin-releasing hormone therapy, such 11. Balducci L, Parker M, Hescock H, et al. Systemic management of as loss of libido, osteoporosis, and hot flashes.2,3 It may prostate cancer. Am J Med Sci. 1990;299(3):185-192. be too late to advocate a return to estrogen therapy, but the results from the Special Report by Lu and colleagues suggest that estrogen therapy might have had an additional advantage over ADT and that promoting the economic considerations of leuprolide might have trampled science and patient safety. We must be skeptical about some of the study conclusions. For example, no biological or epidemiological evidence supports the claim that prostatectomy may have caused the incidence of CRC to increase. It is impossible to accept this conclusion without knowing how many patients received radiotherapy, hormonal therapy, or both following prostatectomy. Some readers may question our decision to publish this Special Report despite its methodological flaws. As the editors of Cancer Control, we are priv- ileged to place the importance of information over academic purity. Despite its flaws, we felt that the conclusion that ADT was associated with an increased risk of CRC was justified and robust, and, as such, the results of this Special Report are relevant to all health care professionals involved in the management of prostate cancer.

Lodovico Balducci, MD Senior Member Program Leader, Senior Adult Oncology Program Editor, Cancer Control H. Lee Moffitt Cancer Center & Research Institute Tampa, Florida [email protected] Julio M. Pow-Sang, MD Chair, Department of Genitourinary Oncology Chief, Surgery Service Director of Moffitt Robotics Program Deputy Editor, Cancer Control H. Lee Moffitt Cancer Center & Research Institute Tampa, Florida [email protected]

260 Cancer Control April 2015, Vol. 22, No. 2 Editorial

Does Androgen Deprivation Therapy for Prostate Cancer Increase the Risk of Colorectal Cancer?

Colorectal cancer (CRC) is the third most common adiposity accumulating during the short-term use of cancer in the United States.1 The causality of CRC is ADT is correlated with resistance to insulin (within multifactorial and includes genetic factors, lifestyle 3 months for some cases) and increasing levels of factors (diet, physical activity, smoking, alcohol con- circulating insulin.17 sumption), obesity, metabolic syndrome, and sex hor- Although clinical recommendations for CRC mones. In the study by Lu and colleagues published screening among patients with prostate cancer re- in this issue of Cancer Control, Swedish patients with ceiving ADT may be unnecessary at this time, health prostate cancer were followed-up and an increased care professionals should be aware of the association risk of CRC was observed in patients diagnosed with between ADT and CRC in this patient population. It prostate cancer after 1980, which is when the use of may be advisable for physicians to order relatively androgen deprivation therapy (ADT; bilateral orchi- inexpensive laboratory studies (eg, fecal occult blood ectomy and gonadotropin-releasing hormone [GnRH] test) and suggest lifestyle modifications and early agonists) increased in Sweden.2-5 No increased risk treatment options for lipid disorders in their patients of CRC was found in patients treated with estrogen with prostate cancer receiving ADT. therapy. Thus, Lu and colleagues concluded that ADT might be a causal factor for the increased risk of Anna Martling, MD, PhD CRC seen in this cohort of patients. The similar but Professor stronger and dose-dependent effect of ADT associat- Department of Molecular Medicine and Surgery Karolinska Institutet ed with an increased risk of CRC was also found by Stockholm, Sweden 6 Gillessen et al. [email protected] Treatments for prostate cancer include active surveillance, radical prostatectomy, radiotherapy, or References hormone therapy. ADT is a first-line treatment option 1. American Cancer Society. Cancer Facts & Figures 2015. Atlanta: Amer- ican Cancer Society; 2015. for men with metastatic prostate cancer because it 2. Giertz G. Urology in Sweden: 1940-1990 [in Swedish]. Sydsven suppresses the binding of androgen to the androgen Medicinhist Sallsk Arssk. 1996;(suppl 21):1-284. 3. McLeod DG. Hormonal therapy: historical perspective to future direc- receptor. In Sweden, hormonal treatment for prostate tions. Urology. 2003;61(2 suppl 1):3-7. cancer has changed over time2-5; up until the 1980s, 4. Adolfsson J, Garmo H, Varenhorst E, et al. Clinical characteristics and primary treatment of prostate cancer in Sweden between 1996 and 2005. estrogen therapy was the predominant hormonal ther- Scand J Urol Nephrol. 2007;41(6):456-477. apy prior to the introduction of ADT.2 5. Cox RL, Crawford ED. Estrogens in the treatment of prostate cancer. J Urol. 1995;154(6):1991-1998. Animal studies have demonstrated that androgen 6. Gillessen S, Templeton A, Marra G, et al. Risk of colorectal cancer in may have a protective effect against colorectal car- men on long-term androgen deprivation therapy for prostate cancer. J Natl Cancer Inst. 2010;102(23):1760-1770. cinogenesis; by contrast, androgen deprivation may 7. Chen SY, Wulf G, Zhou XZ, et al. Activation of beta-catenin signaling in promote it. Evidence indicates that activating andro- prostate cancer by peptidyl-prolyl isomerase Pin1-mediated abrogation of the androgen receptor-beta-catenin interaction. Mol Cell Biol. 2006;26(3):929-939. gen receptors represses Wnt/β-catenin/T-cell factor 8. Teoh JY, Chan SY, Chiu PK, et al. Risk of acute myocardial infarction signaling in colon cancer cells and is associated with after androgen deprivation therapy for prostate cancer in the Chinese popu- 7 lation. BJU Int. 2014. Epub ahead of print. a decreased risk of CRC. This process may influence 9. Teoh JY, Chiu PK, Chan SY, et al. Risk of new-onset diabetes after competition among androgen receptors and T-cell fac- androgen deprivation therapy for prostate cancer in the Asian population. J Diabetes. 2014. Epub ahead of print. tors for β-catenin binding; by contrast, administering 10. Zhao J, Zhu S, Sun L, et al. Androgen deprivation therapy for prostate antiandrogen therapy may instead reverse it. In fact, cancer is associated with cardiovascular morbidity and mortality: a meta-analysis of population-based observational studies. PLoS One. 2014;9(9):e107516. the expression levels of androgen receptors in CRC 11. Cowey S, Hardy RW. The metabolic syndrome: a high-risk state for are lower than those found in samples of normal mu- cancer? Am J Pathol. 2006;169(5):1505-1522. 6 12. Stepien M, Rosniak-Bak K, Paradowski M, et al. Waist circumfer- cosa. Furthermore, testosterone deficiency has been ence, ghrelin and selected adipose tissue-derived adipokines as predictors linked to an increased risk of metabolic syndrome, of insulin resistance in obese patients: preliminary results. Med Sci Monit. 8-10 2011;17(11):PR13-PR18. diabetes, and cardiovascular disease in men ; in 13. Giovannucci E. Metabolic syndrome, hyperinsulinemia, and colon can- addition, strong risk factors for CRC include obesity, cer: a review. Am J Clin Nutr. 2007;86(3):s836-s842. 14. Esposito K, Chiodini P, Capuano A, et al. Colorectal cancer associa- hyperinsulinemia, type 2 diabetes, and metabolic syn- tion with metabolic syndrome and its components: a systematic review with drome.11-16 Research has also indicated that visceral meta-analysis. Endocrine. 2013;44(3):634-647.

April 2015, Vol. 22, No. 2 Cancer Control 261 15. Stocks T, Lukanova A, Bjørge T, et al; Metaboli Syndrome Cancer Project Me-Can Group. Metabolic factors and the risk of colorectal cancer in 580,000 men and women in the metabolic syndrome and cancer project (Me-Can). Cancer. 2011;117(11):2398-2407. 16. Aleksandrova K, Nimptsch K, Pischon T. Influence of obesity and related metabolic alterations on colorectal cancer risk. Curr Nutr Rep. 2013;2(1):1-9. 17. Harrington JM, Schwenke DC, Epstein DR, et al. Androgen-deprivation therapy and metabolic syndrome in men with prostate cancer. Oncol Nurs Forum. 2014;41(1):21-29.

262 Cancer Control April 2015, Vol. 22, No. 2 Special Report

Risk of Colorectal Cancer by Subsite in a Swedish Prostate Cancer Cohort Yunxia Lu, MD, PhD, Rickard Ljung, MD, PhD, Anna Martling, MD, PhD, and Mats Lindblad, MD, PhD

Background: The relationship between sex hormone–related treatment for prostate cancer and the risk of colorectal cancer is controversial. Methods: A prostate cancer cohort was initiated from the Swedish Cancer Registry of patients diagnosed between 1961 and 2008. Patients diagnosed with prostate cancer between 1961 and 1980 were generally treated with estrogen. The cohort diagnosed between 1981 and 2008 was further divided into 3 subcohorts of orchiectomy, prostatectomy, and other treatment. Standardized incidence ratios (SIRs) for developing colorectal adenocarcinoma were estimated and 95% confidence intervals (CIs) were used to compare relative risk among these patients and the general male population. Results: Of 601,542 person-years of follow-up, 1,698 cases of colorectal adenocarcinoma were identified. Com- pared with the general male population, no association was detected in the cohort diagnosed between 1961 and 1980, whereas an increased risk of colorectal adenocarcinoma was observed among patients diagnosed with prostate cancer who received treatments other than estrogen. Following bilateral orchiectomy, the SIR was 1.30 (95% CI: 1.14–1.47); after prostatectomy, the SIR was 1.22 (95% CI: 1.04–1.43); among those who received treatment other than estrogen, the SIR was 1.37 (95% CI: 1.29–1.45). The increased risks were more apparent in cases of adenocarcinoma of the distal colon and rectum than in the proximal colon. Conclusions: Patients with prostate cancer undergoing bilateral orchiectomy, prostatectomy, or other treatments, including antiandrogen therapy and radiation, may be at increased risk for colorectal adenocarcinoma.

Introduction activity.6,7 The incidence of cancer of the proximal The role of the influence of sex hormones in the colon is 10% to 20% higher in women than in men etiology of colorectal cancer (CRC) deserves atten- at all ages, whereas men have a higher incidence of tion, particularly regarding potential differences cancer in the distal colon and rectum.8,9 It is possible between the locations of such cancer. Based on separate that sex hormones could differently influence can- embryological origins, and thus potentially divergent cers of the proximal colon, distal colon, and rectum. causal pathways, the colorectal intestine is often Epidemiological studies in women have shown that divided into the proximal colon (cecum, ascending higher levels of the female sex hormones estrogen colon, and transverse colon), distal colon (descending and progesterone are associated with a lower risk of colon and sigmoid colon), and rectum. Accumulating developing CRC, particularly when exogenous steroid evidence from epidemiological studies has demon- hormones are used,10,11 but few studies have associ- strated the differences in risk-factor profiles between ated sex hormones with risk of colorectal adenocar- men and women of cancers in the colorectal subsites cinoma by subsite in men.12,13 with regard to dietary factors,1-4 obesity,5 and physical Previously, hormone therapy was a major treatment approach for prostate cancer, although hormone therapy does produce adverse events.14-16 Estrogen From the Department of Molecular Medicine and Surgery (YL, RL, therapy and androgen deprivation therapy (ADT; AM) and the Unit of Epidemiology at the Institute of Environmental Medicine (RL), Karolinska Institutet, Stockholm, Sweden, and the including orchiectomy and gonadotropin-releasing Department of Epidemiology and Biostatistics (YL, ML), Imperial hormone [GnRH] agonists) are 2 types of sex hormone College, London, England. therapies used in patients with prostate cancer. The- Address correspondence to Yunxia Lu, MD, PhD, Department of oretically, patients with prostate cancer treated with Molecular Medicine and Surgery, Karolinska Institutet, SE-171 76, Stockholm, Sweden. E-mail: [email protected] sex hormones may constitute an ideal natural human Submitted July 3, 2014; accepted September 26, 2014. model for evaluating the association between external The authors received financial support from the Swedish Research sex hormones and the risk of CRC. In 2010, a study Council and the Swedish Society of Medicine for this study. found that patients with prostate cancer receiving ADT

April 2015, Vol. 22, No. 2 Cancer Control 263 had an increased risk of CRC compared with patients Total Population Registry: The total population not receiving ADT.17 Due to confounding by indi- registry provides complete and continuously updat- cation, this association is still controversial. Further ed information on dates of emigration, immigration, studies comparing patients with prostate cancer and births, and deaths since 1961. Data of censored cohort the general population may provide more evidence. members who died or emigrated during follow-up were collected from this registry. However, these pa- Methods tients were censored from the date of first emigration Because of the nature of Swedish national health or exact date of death. registries (see below for a detailed description of each registry used), we were able to identify all of the Study Cohorts patients diagnosed with prostate cancer in Sweden The methods for treating prostate cancer changed in between 1961 and 2008 and followed them up to Sweden during the 1980s, with estrogen therapy being the development of a second adenocarcinoma in the the predominant treatment strategy prior to 1980.22 colon or rectum. The patients with prostate cancer Bilateral orchiectomy was also an available standard were grouped into 4 treatment cohorts. treatment for prostate cancer between the years 1950 The aim of this study was to determine whether and 1980 in many countries, but orchiectomy was prostate cancer treatments, particularly sex hormone not popular in Sweden during these years.23 Other therapy, were associated with risk for CRC. The study treatments, such as ADT (orchiectomy and GnRH an- was approved by the Regional Ethical Review Board alogues), radical prostatectomy, and radical radiother- in Stockholm, Sweden. apy, increased after the 1980s.22,24-26 The number of patients who did not receive treatment (active surveil- Data Sources lance and watchful waiting) was low.27 Therefore, the This study was based on data from 3 nationwide cohort was divided into mutually exclusive subcohorts Swedish population-based registries, ie, the cancer, based on the definition of sex hormone treatments patient, and total population registries. The personal following prostate cancer diagnosis (Fig). The cohort identity number, a unique 10-digit identification num- diagnosed prior to 1981 included persons diagnosed ber assigned to each resident of Sweden, was used to with prostate cancer before 1981; a second cohort of link all of the patients between the registries. patients included those diagnosed with prostate can- Cancer Registry: The cancer registry was initiat- cer between January 1, 1981, and December 31, 2008. ed in 1958 and includes the date of diagnosis, tumor The last subcohort was further divided into patients site, and histological type of all malignant tumors who underwent bilateral orchiectomy, patients who diagnosed in Sweden. All newly diagnosed tumors in underwent prostatectomy, and patients treated mainly the country must be reported to the cancer registry by by GnRH therapy, radiotherapy, or both assigned to a clinician and a pathologist or cytologist. The com- the “other treatment” group. pleteness of the registry approaches 98% to 100%,18 and 99% of all tumors are morphologically verified.19 Follow-Up The cohorts included all men registered with a first The follow-up of the cohort members started from diagnosis of prostate cancer. The years 1958 to 1960 the date prostate cancer was newly diagnosed were excluded to avoid inclusion of the prevalence of or the start of specific treatments in the cohort, cancer cases when the registry began. All patients with and patients were followed-up until any of the any diagnosis of other previous cancer were exclud- following end points: diagnosis of any cancer ed. Any subsequent cancer after prostate cancer was (except nonmelanoma skin cancer), emigration, identified throughout the cancer registry, including age 85 years, death, or end of the study period, the outcome of colorectal adenocarcinoma (proximal whichever came first. The end of the study period colon, distal colon, unspecified colon, and rectum). for the cohort before the 1980s was set to Patient Registry: The patient registry was initiat- December 31, 1980, and to December 31, 2008, ed in 1964 and covered 2 Swedish counties; starting in for all the other cohorts.22 The cutoff date 1987, the registry covered 100% of Sweden.20 This reg- (December 31, 1980) for the end of the first study pe- istry has achieved accuracy and surgical completeness riod or the beginning of the second study period was rates of 95% and 98%, respectively.21 Information on the arbitrarily selected based on the previously reported surgical procedures following the diagnosis of prostate changes of treatment in Sweden. cancer was collected from this registry. The Swedish Classification of Operations and Major Procedures has Statistical Analyses been included in the patient registry since 1964, and Person-years were calculated from the date of it was used to identify patients undergoing bilateral prostate cancer diagnosis (or specific treatment, orchiectomy and prostatectomy. eg, orchiectomy) to the date of the study’s end point.

264 Cancer Control April 2015, Vol. 22, No. 2 Diagnosis of Prostate Cancer (N = 149,743)

Estrogen-Exposed Cohort Nonexposed Estrogen Cohort Diagnosis of prostate cancer (1961–1980) Diagnosis of prostate cancer (1981–2008) (n = 33,373) (n = 116,370)

Bilateral Orchiectomy Prostatectomy Other Treatments (n = 21,917) (n = 16,521) (n = 77,932)

Fig. — Flow chart for the prostate cancer cohorts (1961–2008).

Standardized incidence ratios (SIRs) were estimated the specific cohorts is shown in Table 1. The cohort as a measure of relative risk when comparing the diagnosed between 1961 and 1980 was composed data from those diagnosed with prostate cancer and of 33,373 patients with prostate cancer who were the corresponding general male population. The ratio followed-up for 111,809 person-years until 1980, and was calculated based on the observed number divided the total cohort diagnosed with prostate cancer be- by the expected number of newly diagnosed cases of tween 1981 and 2008 included 116,370 patients with CRC. The expected number of cases was calculated prostate cancer and 489,733 person-years of follow-up by multiplying the observed number of person-years since 1981. During follow-up, we identified a total by the age- and calendar year–specific incidence rates of 1,698 cases of colorectal adenocarcinoma. Among with 5-year intervals of the entire Swedish male pop- them, 487 cases of adenocarcinoma were located in ulation. SIRs and 95% confidence intervals (CIs) were the proximal colon, 453 cases in the distal colon, calculated with the assumption that the number of 132 cases in an unspecified site of the colon, and cases followed a Poisson distribution. SIRs were sepa- 626 cases in the rectum. Age at entry and age at diag- rately estimated for cancers of the proximal colon, dis- nosis of colorectal adenocarcinoma were significantly tal colon, unspecified colon, and rectum in each of the different among the groups (all P values < .01). Those cohorts. We also performed sensitivity analyses with in the prostatectomy cohort were younger than pa- respect to the separate cohorts, which we assumed tients in the other cohorts. were exposed to various prostate cancer treatments. For the cohort diagnosed between 1961 and 1980, Estrogen Exposure we analyzed the cohort members who were followed Compared with the general population, the cohort from 1961 to 1975, with the end point being 5 years diagnosed between 1961 and 1980 was not associated earlier than the prior assumed cutoff date. For the with the overall development of colorectal adenocarci- 3 cohorts diagnosed after 1980, we performed noma (SIR 0.98; 95% CI: 0.85–1.12) or when analyzed analyses on patients who entered into the cohort by subsite (Table 2). Results suggested a statistically from 1986 through 2008, with the start point being nonsignificant decreased risk of adenocarcinoma of 5 years later than the prior assumed cutoff date. In the proximal colon, but these results were mainly ob- the main analysis, we excluded the first year of fol- served among patients with a short follow-up; howev- low-up after the diagnosis of prostate cancer because er, a nonsignificant trend of increased risk with years cancers diagnosed in the first year were particularly of follow-up (P value for trend = .09) in the proximal prone to detection bias and unlikely to be related to colon was identified. the treatment of prostate cancer. All tests were 2-sided with a significance Orchiectomy level of 0.05. Analyses were performed using the Between 1981 and 2008, a total of 21,917 individuals SAS statistical package, version 9.0 (SAS Institute, underwent bilateral orchiectomy and had an aver- Cary, North Carolina). age follow-up period of 3.4 years (Tables 1 and 3). An overall increased risk of colorectal ade- Results nocarcinoma was detected in patients with A total of 149,743 patients with prostate cancer were 1 to 9 years of follow-up, but this risk decrease after included in the final study cohort and information for 10 years of follow-up. A significantly increased risk

April 2015, Vol. 22, No. 2 Cancer Control 265 Table 1. — Basic Characteristics of the Prostate Cancer Treatment Groups Specific Treatment Cohorts Estrogen Exposed Nonexposed Estrogen (1981–2008) (1961–1980) Orchiectomy Prostatectomy Other Treatment No. of cohort members 33,373 21,917 16,521 77,932 Total person-y of follow-up 111,809 75,043 75,700 338,990 Average age at entry, y 71.9 ± 7.3 74.9 ± 6.4 63.8 ± 5.9 71.1 ± 7.3 Average follow-up, y 3.4 3.4 4.6 4.3 No. of colorectal adenocarcinoma cases 198 247 156 1097 No. of colorectal adenocarcinoma cases 74 97 16 332 at first year of follow-up Average age at diagnosis of colorectal adenocarcinoma, y 76.2 ± 6.2 77.8 ± 4.7 70.5 ± 6.1 76.4 ± 5.6 Person-y incidence (95% confidence intervala)b Colorectal adenocarcinoma 177 (152.4–201.8) 329 (288.1–370.2) 206 (173.7–238.4) 323 (304.5–342.8) Proximal colon adenocarcinoma 46 (33.1–58.1) 77 (57.4–97.2) 40 (25.4–53.8) 102 (91.9–113.4) Distal colon adenocarcinoma 53 (39.3–66.2) 85 (64.4–106.2) 63 (45.5–81.3) 83 (73.5–92.9) Rectal colon 59 (44.8–73.2) 124 (98.7–149.1) 94 (72.0–115.6) 117 (105.3–128.3) Unspecified colon adenocarcinoma 20 (11.5–27.9) 43 (27.9–57.4) 9 (2.4–16.1) 21 (16.1–25.8) aCalculation based on the first year of follow-up had been excluded. bIncidence counted as per 100,000 person-y.

Table 2. — SIRs and 95% CIs of Colorectal Adenocarcinoma in the Estrogen-Exposed Cohort Compared With the General Population (1961–1980) Colorectal Adenocarcinoma Adenocarcinoma Adenocarcinoma Unspecified Colorectal Adenocarcinoma of the Proximal Colon of the Distal Colon of the Rectum Adenocarcinoma Obs Exp SIR Obs Exp SIR Obs Exp SIR Obs Exp SIR Obs Exp SIR (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) 198 202 0.98 51 58 0.88 59 56 1.05 66 74 0.89 22 14 1.60 (0.85–1.12) (0.66–1.16) (0.80–1.35) (0.69–1.13) (1.00–2.42) Age at diagnosis of prostate cancer, y 41–64 25 29 0.86 3 8 0.39 8 8 1.00 11 12 0.94 3 2 1.69 (0.55–1.26) (0.08–1.13) (0.43–1.98) (0.47–1.69) (0.35–4.93) 65–74 104 105 0.99 29 30 0.98 34 30 1.15 29 39 0.74 12 7 1.77 (0.81–1.20) (0.66–1.41) (0.80–1.61) (0.50–1.06) (0.88–2.98) ≥75 69 68 1.02 19 20 0.93 17 19 0.90 26 24 1.10 7 5 1.41 (0.79–1.28) (0.56–1.45) (0.52–1.44) (0.72–1.61) (0.57–2.90) P value for trend 0.52 .31 .63 .043 .72 Follow-upa, y 1–4 129 135 0.96 29 39 0.74 40 39 1.04 44 48 0.92 16 10 1.67 (0.80–1.14) (0.50–1.06) (0.74–1.41) (0.67–1.24) (0.95–2.71) 5–9 59 54 1.10 17 15 1.14 16 14 1.12 20 21 0.95 6 3 1.76 (0.83–1.41) (0.66–1.82) (0.64–1.81) (0.58–1.46) (0.64–3.83) ≥10 10 14 0.74 5 4 1.37 3 3 0.88 2 6 0.35 0 0 0 (0.35–1.36) (0.44–3.19) (0.18–2.58) (0.04–1.27) P value for trend .99 .09 .98 .34 .49 aFirst year of follow-up was excluded. CI = confidence interval, Exp = expected, Obs = observed, SIRS = standardized incidence ratio. was observed for the overall rate of colorectal adeno- 16,521 patients who underwent prostatectomy carcinoma (SIR 1.30; 95% CI: 1.14–1.47) and for each between 1981 and 2008 and had an average fol- subsite except for the proximal colon. low-up period of 4.6 years (Tables 1 and 4). Sim- ilarly increased risks of colorectal adenocarcino- Prostatectomy ma overall and by specific subsites were observed The prostatectomy cohort was composed of among patients in the prostatectomy cohort.

266 Cancer Control April 2015, Vol. 22, No. 2 Table 3. — SIRs and 95% CIs of Colorectal Adenocarcinoma in the Prostate Cancer Cohort With Orchiectomy Compared With the General Population (1981–2008) Colorectal Adenocarcinoma Adenocarcinoma Adenocarcinoma Unspecified Colorectal Adenocarcinoma of the Proximal Colon of the Distal Colon of the Rectum Adenocarcinoma Obs Exp SIR Obs Exp SIR Obs Exp SIR Obs Exp SIR Obs Exp SIR (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) 247 190 1.30 58 55 1.06 64 44 1.44 93 74 1.26 32 17 1.91 (1.14–1.47) (0.80–1.36) (1.11–1.84) (1.01–1.54) (1.31–2.70) Age at diagnosis of prostate cancer, y 41–64 26 17 1.55 2 4 0.46 10 4 2.48 13 7 1.85 1 1 0.73 (1.01–2.27) (0.06–1.66) (1.19–4.57) (0.98–3.16) (0.02–4.06) 65–74 132 97 1.36 34 27 1.24 32 23 1.40 44 38 1.14 22 9 2.57 (1.14–1.61) (0.86–1.74) (0.96–1.98) (0.83–1.53) (1.61–3.89) ≥75 89 76 1.17 22 23 0.95 22 17 1.26 36 28 1.26 9 7 1.32 (0.94–1.44) (0.59–1.43) (0.79–1.91) (0.89–1.75) (0.60–2.50) P value for trend .15 .99 .14 .51 .49 Follow-upa, y 1–4 158 127 1.24 38 36 1.05 37 29 1.26 58 50 1.16 25 11 2.22 (1.06–1.45) (0.74–1.44) (0.89–1.73) (0.88–1.50) (1.44–3.28) 5–9 75 49 1.53 15 14 1.05 24 11 2.10 30 19 1.59 6 4 1.36 (1.20–1.91) (0.59–1.72) (1.34–3.12) (1.07–2.27) (0.50–2.95) ≥10 14 14 1.00 5 4 1.17 3 3 0.88 5 5 0.96 1 1 0.92 (0.55–1.68) (0.38–2.73) (0.18–2.57) (0.31–2.24) (0.02–5.11) P value for trend .77 .86 .47 .59 .18 aFirst year of follow-up was excluded. CI = confidence interval, Exp = expected, Obs = observed, SIRS = standardized incidence ratio.

Specifically, an increased SIR was found for the Discussion distal colon (1.44; 95% CI: 1.06–1.91) and rectum This nationwide Swedish cohort study identified an (1.36; 95% CI: 1.06–1.71). increased risk of colorectal adenocarcinoma for all anatomical locations among patients exposed to bilat- Other Treatments eral orchiectomy, ADT, radical prostatectomy, and ra- Among patients diagnosed with prostate cancer af- diotherapy. No clear association was detected among ter 1980, a total of 77,932 did undergo bilateral or- individuals exposed to estrogen. chiectomy or prostatectomy during the study period Treatment for prostate cancer diagnosed af- (see Table 1). Patients in the other treatment cohort ter 1980 may be associated with an increased risk had an overall increased risk of colorectal adeno- of colorectal adenocarcinoma, implying a possible carcinoma (SIR 1.37; 95% CI: 1.29–1.45) and for all connection to ADT, one of the most common treat- subsites (Table 5). Specifically, an increased risk of ments in Sweden used after 1980.27 The increased adenocarcinoma in the proximal colon was associated risk of colorectal adenocarcinoma found among with age (P value for trend < .01); a decreased risk patients undergoing bilateral orchiectomy could was observed in association with years of follow-up be related to an effect of surgical androgen depri- (P value for trend = .02; see Table 5). vation, whereas the increased risk seen in patients not undergoing orchiectomy or prostatectomy could Sensitivity Analyses be due to medically induced androgen deprivation Analyses of the cohort diagnosed between 1961 and (ie, GnRH therapy). Higher SIRs were observed in 1975 did not identify an association between estro- the sensitivity analyses for the prostatectomy and gen treatment and overall or subsite risk of CRC, a other treatment cohorts, particularly in those with finding consistent with the results from the originally short-term follow-up. The results also indicated that defined cohort (diagnosed between 1961 and 1980). other treatments such as GnRH therapy may play a In the orchiectomy cohort, which included patients role in the risk of CRC, although the probability value diagnosed between 1986 and 2008, increased SIRs for such a trend did not reach statistical significance. of CRC were found, in line with the results shown in Alternatively, a carcinogenic effect of radiotherapy Table 3. In the prostatectomy and other treatment cohorts, for prostate cancer may be present; however, this similarly increased and stronger SIRs were observed when represents exposure for which we have no data. Con- compared with the general male population. sistent results have also been found in a US study

April 2015, Vol. 22, No. 2 Cancer Control 267 Table 4. — SIRs and 95% CIs of Colorectal Adenocarcinoma in the Prostatectomy Cohort Compared With the General Population (1981–2008) Colorectal Adenocarcinoma Adenocarcinoma Adenocarcinoma Unspecified Colorectal Adenocarcinoma of the Proximal Colon of the Distal Colon of the Rectum Adenocarcinoma Obs Exp SIR Obs Exp SIR Obs Exp SIR Obs Exp SIR Obs Exp SIR (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) 156 127 1.22 30 35 0.86 48 33 1.44 71 52 1.36 7 7 1.04 (1.04–1.43) (0.58–1.23) (1.06–1.91) (1.06–1.71) (0.42–2.15) Age at diagnosis of prostate cancer, y 41–64 73 63 1.16 13 16 0.79 19 16 1.17 36 27 1.33 5 3 1.59 (0.91–1.46) (0.42–1.36) (0.70–1.82) (0.93–1.84) (0.52–3.71) 65–74 79 62 1.27 17 18 0.95 27 16 1.63 33 24 1.35 2 3 0.59 (1.00–1.58) (0.55–1.52) (1.07–2.37) (0.93–1.89) (0.07–2.13) ≥75 4 2 1.89 0 1 0.00 2 1 4.00 2 1 2.50 0 0 0.00 (0.51–4.83) (0.00–5.75) (0.48– (0.30–9.04) (0.00– 14.43) 20.85) P value for trend .41 .86 .12 .72 .19 Follow-upa, y 1–4 89 68 1.31 16 18 0.89 28 18 1.59 41 29 1.41 4 4 1.13 (1.05–1.61) (0.51–1.44) (1.05–2.29) (1.02–1.92) (0.31–2.90) 5–9 49 40 1.22 11 11 0.98 14 11 1.32 22 16 1.35 2 2 0.92 (0.90–1.61) (0.49–1.76) (0.72–2.21) (0.85–2.05) (0.11–3.34) ≥10 18 19 0.95 3 6 0.53 6 5 1.18 8 7 1.12 1 1 0.98 (0.56–1.50) (0.11–1.54) (0.43–2.57) (0.48–2.20) (0.02–5.48) P value for trend .24 .56 .44 .57 .84 aFirst year of follow-up was excluded. CI = confidence interval, Exp = expected, Obs = observed, SIRS = standardized incidence ratio.

Table 5. — SIRs and 95% CIs of Colorectal Adenocarcinoma in Other Treatment Cohorts Compared With the General Population (1981–2008)a A Colorectal Adenocarcinoma Adenocarcinoma Adenocarcinoma Unspecified Colorectal Adenocarcinoma of the Proximal Colon of the Distal Colon of the Rectum Adenocarcinoma Obs Exp SIR Obs Exp SIR Obs Exp SIR Obs Exp SIR Obs Exp SIR (95% CI) (95% CI) (95% CI) (95% CI) (95% CI) 1,097 800 1.37 348 234 1.49 282 203 1.39 396 309 1.28 71 54 1.31 (1.29–1.45) (1.33–1.65) (1.23–1.56) (1.16–1.41) (1.03–1.66) Age at diagnosis of prostate cancer, y 41–64 133 108 1.23 28 31 1.10 34 28 1.24 61 46 1.32 7 7 1.07 (1.03–1.45) (0.75–1.56) (0.86–1.73) (1.01–1.70) (0.43–2.21) 65–74 585 429 1.36 169 123 1.37 158 111 1.43 219 167 1.31 39 28 1.38 (1.26–1.48) (1.17–1.60) (1.21–1.67) (1.15–1.50) (0.98–1.88) ≥75 379 263 1.44 148 83 1.79 90 65 1.38 116 96 1.21 25 19 1.30 (1.30–1.59) (1.51–2.10) (1.11–1.70) (1.00–1.45) (0.84–1.92) P value for trend .12 .00 .72 .48 .81 Follow-upb, y 1–4 694 489 1.42 227 141 1.61 175 123 1.42 244 191 1.28 48 34 1.42 (1.32–1.53) (1.41–1.84) (1.22–1.64) (1.12–1.45) (1.05–1.88) 5–9 321 239 1.34 99 71 1.39 81 61 1.32 123 91 1.35 18 16 1.14 (1.20–1.50) (1.13–1.69) (1.05–1.64) (1.12–1.61) (0.68–1.81) ≥10 82 72 1.14 22 22 1.00 26 19 1.40 29 27 1.09 5 5 1.11 (0.91–1.42) (0.62–1.51) (0.92–2.06) (0.73–1.56) (0.36–2.58) P value for trend .07 .02 .76 .75 .40 aStudy cohorts excluded patients who underwent orchiectomy and prostatectomy. bFirst year of follow-up was excluded. CI = confidence interval, Exp = expected, Obs = observed, SIRS = standardized incidence ratio. that found a strong association between orchiecto- risk (increase in CRC incidence of 30%–40% in men my and colorectal adenocarcinoma.17 The elevated exposed to ADT) was noticeable even after adjust-

268 Cancer Control April 2015, Vol. 22, No. 2 ing for potential confounders such as diabetes, obe- instability — 2 forms of genetic instability in sity, and radiotherapy.17 Although that study had CRC — mostly occur in the proximal and distal colon, robust data regarding orchiectomy and GnRH ther- respectively, warranting the genetic basis.36 apy, it included bias from confounding by indication and its participants were older.17 Our study, which Limitations measured relative risk based on a comparison between Strengths of this study include its nationwide and the study cohort groups and the general population, population-based cohort design identified from further demonstrated the findings of the US study.17 national registries, its large prostate cancer cohort sizes, and its long and complete follow-up times; Estrogen and Androgen however, information on individual prostate cancer We did not find any association between estrogen treatment was limited. Data on bilateral orchiectomy, treatment and colorectal adenocarcinoma. Previous which is a common ADT, was available in the patient reports, which were based on observational data and registry from 1964 onward, but we had no detailed results from clinical trials, have suggested that women information on other hormonal treatments such as exposed to exogenous estrogen, including hormone estrogen or antiandrogen medications. We specified replacement therapy or oral contraceptives, have a the cohorts based on time periods and specific treat- decreased risk of CRC.12,27,28 However, the estrogen ments. The results of the sensitivity analyses showed hypothesis has seldom been evaluated in men. fair consistency with the reported results based on The relationship between estrogen and androgen the selected cutoff dates. is intriguing. Estrogens are produced with androgens Another concern is the introduction of pros- as precursors. Progesterone is the first important sex- tate-specific antigen (PSA), which could change ual steroid formed in the body. The androgens (dehy- the profiles of patients with prostate cancer.28 PSA droepiandrosterone, androstenedione, and testoster- screening was introduced in Sweden between 1995 one) arise thereafter, whereas the estrogens (estrone and 2000 and steadily increased from 1% to 10%.37 and estradiol) appear only during the final stage. Less aggressive prostate cancer might have been Estrogen and androgen may play a similar role in diagnosed after the introduction of PSA test- carcinogenesis among women and men, respectively. ing. However, this would have little influence Estrogen and androgen were postulated to prevent on our results because we stratified the analy- tumor growth by preventing insulin and insulin-like sis in all 3 subcohorts. Treatments for the first growth factor from binding to their receptors.29 By 2 cohorts were defined, whereas influences on the contrast, an increased risk of colorectal adenocarcino- third cohort could have been overlooked. Considering ma due to androgen deprivation could be possible as the lower proportion of PSA screening and the large the results from animal studies suggest that androgen size of this cohort, PSA screening should not engender may have a protective effect on the development of a substantial influence on the total results. CRC.30,31 Androgen receptors have also been found It is possible that CRC screening might influence more frequently in the normal — as opposed to can- the diagnosis of CRC in patients with prostate cancer. cerous — mucosa in the colon.31,32 Thus, orchiectomy In Sweden, CRC screening commenced in the 1980s or GnRH treatment would directly decrease androgen. but patients were limited; the officially organized screening of CRC began in Stockholm in 2008 and Specific Subsite in the rest of Sweden in 2014,38,39 so CRC screening Our results suggest differences in risk may exist be- would not have substantially affected the results of tween the proximal and distal colon. The decreasing our study. Furthermore, we did not find studies to in- trend of SIRs associated with adenocarcinoma in the dicate that colonoscopy was increased in the prostate proximal colon over follow-up years could indicate a cancer survivor group in Sweden. It may be possible differing association at that subsite. Select genetic or that patients receiving radiotherapy for prostate can- physiological mechanisms might explain this differ- cer may also have had bowel symptoms. Because ence in risk patterns of adenocarcinoma in the prox- SIRs were calculated based on a comparison of the imal and distal colon. The proximal colon originates prostate cancer cohort with the general population at from the midgut, but the distal colon and rectum origi- the same period, such an influence would be minor. nate from the hindgut. Research suggests that different The adverse events of hormone therapy are con- genetic pathways to CRC dominate the proximal cerning and may contribute to an increased risk of and the distal segments of the bowel.33 These genet- CRC,17 but such information cannot be specified in ic-dependent pathways are influenced by different the current study. Further adjustment of obesity and sex-related factors.34 Estrogen receptors are distribut- diabetes did not change the association of hormone ed differently in the proximal and distal colon.35 Fur- therapy and CRC in one study.17 thermore, chromosomal instability and microsatellite Lastly, no information was provided on what may

April 2015, Vol. 22, No. 2 Cancer Control 269 be considered to be confounding factors, including Am J Epidemiol. 1989;130(2):289-299. 21. Nilsson AC, Spetz CL, Carsjo K, et al. Reliability of the hospital registry. familial background and dietary habits. However, it is The diagnostic data are better than their reputation [in Swedish]. Lakartidnin- unlikely that these factors could explain our results, gen. 1994;91(7):598, 603-605. 22. Giertz G. Urology in Sweden: 1940-1990 [in Swedish]. Sydsven as it is likely they are equally distributed between Medicinhist Sallsk Arssk. 1996;(suppl 21):1-284. patients regardless of prostate cancer treatment. 23. Lindblad M, Ye W, Rubio C, et al. Estrogen and risk of gastric cancer: a protective effect in a nationwide cohort study of patients with prostate cancer Although the randomized clinical trial is the gold in Sweden. Cancer Epidemiol Biomarkers Prev. 2004;13(12):2203-2207. standard for this type of research, its use was not 24. McLeod DG. Hormonal therapy: historical perspective to future directions. Urology. 2003;61(2 suppl 1):3-7. feasible for our study due to ethical issues. 25. Adolfsson J, Garmo H, Varenhorst E, et al. Clinical characteristics and primary treatment of prostate cancer in Sweden between 1996 and 2005. Scand J Urol Nephrol. 2007;41(6):456-477. Conclusions 26. Cox RL, Crawford ED. Estrogens in the treatment of prostate cancer. Results from this population-based cohort study sug- J Urol. 1995;154(6):1991-1998. 27. Onkologiskt Centrum Norra Regionen. The National Prostate Cancer gest that androgen deprivation therapy, including Register (NPCR) in Sweden 2004 – 2008. Uppsala, Sweden: Onkologiskt bilateral orchiectomy, may increase the risk of col- Centrum; 2010. http://www.cancercentrum.se/Global/Diagnoser/prostatacan- cer/rapporter/prostataca_rapport_04_08.pdf. Accessed January 7, 2015. orectal adenocarcinoma in men, although confound- 28. Lilja H, Ulmert D, Vickers AJ. Prostate-specific antigen and prostate ers of radiotherapy or prostate cancer could not be cancer: prediction, detection and monitoring. Nat Rev Cancer. 2008;8:268-278. 29. Slattery ML, Ballard-Barbash R, Edwards S, et al. Body mass index excluded. Further studies are warranted to elucidate and colon cancer: an evaluation of the modifying effects of estrogen (United this potential association. States). Cancer Causes Control. 2003;14(1):75-84. 30. Chan AT, Giovannucci EL. Primary prevention of colorectal cancer. Gastroenterology. 2010;138(6):2029-2043e10. References 31. Izbicki JR, Hamilton SR, Wambach G, et al. Effects of androgen manip- ulations on chemically induced colonic tumours and on macroscopically normal 1. Larsson SC, Rafter J, Holmberg L, et al. Red meat consumption and colonic mucosa in male Sprague-Dawley rats. Br J Cancer. 1990;61(2):235-240. risk of cancers of the proximal colon, distal colon and rectum: the Swedish 32. Rao CV, Tokumo K, Rigotty J, et al. Chemoprevention of colon carcino- Mammography Cohort. Int J Cancer. 2005;113(5):829-834. genesis by dietary administration of piroxicam, alpha-difluoromethylornithine, 2. Koushik A, Hunter DJ, Spiegelman D, et al. Fruits, vegetables, and 16 alpha-fluoro-5-androsten-17-one, and ellagic acid individually and in com- colon cancer risk in a pooled analysis of 14 cohort studies. J Natl Cancer Inst. bination. Cancer Res. 1991;51(17):4528-4534. 2007;99(19):1471-1483. 33. Sugai T, Habano W, Jiao YF, et al. Analysis of molecular alterations 3. Kim MK, Sasaki S, Otani T, et al; Japan Public Health Center-based in left- and right-sided colorectal carcinomas reveals distinct pathways of Prospective Study Group. Dietary patterns and subsequent colorectal cancer carcinogenesis: proposal for new molecular profile of colorectal carcinomas. risk by subsite: a prospective cohort study. Int J Cancer. 2005;115(5):790-798. J Molec Diag. 2006;8(2):193-201. 4. Chao A, Thun MJ, Connell CJ, et al. Meat consumption and risk of 34. Breivik J, Lothe RA, Meling GI, et al. Different genetic pathways colorectal cancer. JAMA. 2005;293(2):172-182. to proximal and distal colorectal cancer influenced by sex-related factors. 5. Laake I, Thune I, Selmer R, et al. A prospective study of body mass Int J Cancer. 1997;74(6):664-669. index, weight change, and risk of cancer in the proximal and distal colon. 35. Jassam N, Bell SM, Speirs V, et al. Loss of expression of oestrogen Cancer Epidemiol Biomarkers Prev. 2010;19(6):1511-1522. receptor beta in colon cancer and its association with Dukes’ staging. Oncol 6. Boyle T, Fritschi L, Heyworth J, et al. Long-term sedentary work and the Rep. 2005;14(1):17-21. risk of subsite-specific colorectal cancer. Am J Epidemiol. 2011;173(10):1183-1191. 36. Lindblom A. Different mechanisms in the tumorigenesis of proximal 7. Moradi T, Gridley G, Björk J, et al. Occupational physical activity and and distal colon cancers. Curr Opin Oncol. 2001;13(1):63-69. risk for cancer of the colon and rectum in Sweden among men and women 37. Vickers AJ, Ulmert D, Sjoberg DD, et al. Strategy for detection of pros- by anatomic subsite. Eur J Cancer Prev. 2008;17(3):201-208. tate cancer based on relation between prostate specific antigen at age 40-55 8. McMichael AJ, Potter JD. Host factors in carcinogenesis: certain bile-ac- and long term risk of metastasis: case-control study. BMJ. 2013;346:f2023. id metabolic profiles that selectively increase the risk of proximal colon cancer. 38. Blom J, Kilpeläinen S, Hultcrantz R, et al. Five-year experience of J Natl Cancer Inst. 1985;75(2):185-191. organized colorectal cancer screening in a Swedish population - increased 9. Potter JD, Slattery ML, Bostick RM, et al. Colon cancer: a review of the compliance with age, female gender, and subsequent screening round. J Med epidemiology. Epidemiol Rev. 1993;15(2):499-545. Screen. 2014;21(3):144-150. 10. Chlebowski RT, Wactawski-Wende J, Ritenbaugh C, et al. Estrogen 39. Hakama M, Hoff G, Kronborg O, et al. Screening for colorectal cancer. plus progestin and colorectal cancer in postmenopausal women. N Engl J Acta Oncol. 2005;44(5):425-439. Med. 2004;350(10):991-1004. 11. Beral V, Banks E, Reeves G. Evidence from randomised trials on the long-term effects of hormone replacement therapy. Lancet. 2002;360(9337):942-944. 12. Troisi R, Schairer C, Chow WH, et al. Reproductive factors, oral con- traceptive use, and risk of colorectal cancer. Epidemiology. 1997;8(1):75-79. 13. Delellis Henderson K, Duan L, Sullivan-Halley J, et al. Menopausal hormone therapy use and risk of invasive colon cancer: the California Teachers Study. Am J Epidemiol. 2010;171(14):415-425. 14. Seible DM, Gu X, Hyatt AS, et al. Weight gain on androgen deprivation therapy: which patients are at highest risk? Urology. 2014;83(6):1316-1321. 15. James H III, Aleksic I, Bienz MN, et al. Comparison of fracture risk assessment tool score to bone mineral density for estimating fracture risk in patients with advanced prostate cancer on androgen deprivation therapy. Urology. 2014;84(1):164-168. 16. Schmidt-Hansen M, Hoskin P, Kirkbride P, et al. Hormone and radiotherapy versus hormone or radiotherapy alone for non-metastatic prostate cancer: a systematic review with meta-analyses. Clin Oncol (R Coll Radiol). 2014;26(10):e21-e46. 17. Gillessen S, Templeton A, Marra G, et al. Risk of colorectal cancer in men on long-term androgen deprivation therapy for prostate cancer. J Natl Cancer Inst. 2010;102(23):1760-1770. 18. Barlow L, Westergren K, Holmberg L, et al. The completeness of the Swedish Cancer Register: a sample survey for year 1998. Acta Oncologica. 2009;48(1):27-33. 19. Statistics Sweden. Official statistics of Sweden. http://www.scb.se/en_/ About-us/Official-Statistics-of-Sweden. Accessed January 7, 2015. 20. Naessén T, Parker R, Persson I, et al. Time trends in incidence rates of first hip fracture in the Uppsala Health Care Region, Sweden, 1965-1983.

270 Cancer Control April 2015, Vol. 22, No. 2 Ten Best Readings Relating to Molecular Biomarkers

Joosse SA, Gorges TM, Pantel K. Biology, detection, Reuss DE, Sahm F, Schrimpf D, et al. ATRX and and clinical implications of circulating tumor cells. IDH1-R132H immunohistochemistry with subse- EMBO Mol Med. 2014;7(1):1-11. quent copy number analysis and IDH sequencing In this review, the authors discuss the biology of as a basis for an “integrated” diagnostic approach tumor cell dissemination, technical advances, chal- for adult astrocytoma, oligodendroglioma and lenges, and potential clinical implications of detecting glioblastoma. Acta Neuropathol. 2015;129(1): and characterizing circulating tumor cells. 133-146. An algorithm based on a stepwise analysis with Kastelein F, Biermann K, Steyerberg EW, et al. initial immunohistochemistry is presented for ATRX Aberrant p53 protein expression is associated and IDH1-R132H followed by 1p/19q analysis fol- with an increased risk of neoplastic progres- lowed by IDH sequencing. IDH sequencing has re- sion in patients with Barrett’s oesophagus. Gut. duced the number of molecular analyses, leading to 2013;62(12):1676-1683. a better association with patient outcomes than the Aberrant p53 protein expression is associated World Health Organization classification of 2007. with an increased risk for neoplastic progression in patients with Barrett esophagus and may be a more Al-Zaid T, Somaiah N, Lazar AJ. Targeted thera- powerful predictor of neoplastic progression than a pies for sarcomas: new roles for the pathologist. histological diagnosis of low-grade dysplasia. Histopathology. 2014;64(1):119-133. This review enumerates molecular derangements Zhang L, Padron E, Lancet J. The molecular ba- and targeted agents in soft-tissue sarcoma. Select sar- sis and clinical significance of genetic mutations comas are highlighted to illustrate how pathologists identified in myelodysplastic syndromes. Leuk Res. can influence patient care through diagnosis, grading, 2015;39(1):6-17. and molecular characterizations. In this review, the authors describe novel key mutations in myelodysplastic syndromes and their Bridge JA. The role of cytogenetics and molecular significance in pathophysiology and clinical practice. diagnostics in the diagnosis of soft-tissue tumors. Mod Pathol. 2014;27(suppl 1):S80-S97. Cancer Genome Atlas Research Network. Integrat- This review addresses known recurrent or tu- ed genomic characterization of papillary thyroid mor-specific genetic events in soft-tissue tumors and carcinoma. Cell. 2014;159(3):676-690. discusses the molecular approaches commonly used The authors describe the genomic landscape of in clinical practice to identify them. Emphasis is 496 papillary thyroid carcinomas (PTCs). In this study, placed on the role of molecular pathology in soft-tis- they observed a low frequency of somatic alterations sue tumor management and the need for pathologists and extended the set of known PTC driver alterations to be familiar with these genetic events. to include EIF1AX, PPM1D, and CHEK2, along with diverse gene fusions. These discoveries reduced the Wei S, Siegal GP. Round cell tumors of bone: an fraction of PTC cases with unknown oncogenic driv- update on recent molecular genetic advances. ers. The authors propose reclassifying thyroid cancers Adv Anat Pathol. 2014;21(5):359-372. into molecular subtypes to reflect their underlying The authors discuss molecular genetic advances in signaling and differentiation properties. primary, nonhematological round cell tumors of bone.

Asa SL, Giordano TJ, LiVolsi VA. Implications of Dowsett M, Sestak I, Lopez-Knowles E, et al. Com- the TCGA genomic characterization of papillary parison of PAM50 risk of recurrence score with on- thyroid carcinoma for thyroid pathology: does fol- cotype DX and IHC4 for predicting risk of distant licular variant papillary thyroid carcinoma exist? recurrence after endocrine therapy. J Clin Oncol. Thyroid. 2015;25(1):1-2. 2013;31(22):2783-2790. The Cancer Genome Atlas study may simplify the The authors compare different methods for deter- classification of thyroid cancer, with the authors sug- mining risk of recurrence in estrogen receptor–pos- gesting refocused efforts on studying the complexity itive breast cancer. of the follicular variant of thyroid neoplasia.

April 2015, Vol. 22, No. 2 Cancer Control 271 Contact and General Information

Cancer Control: Journal of the Moffitt Cancer Center is To Cancel a Subscription: published by H. Lee Moffitt Cancer Center & Research Please send your cancellation request, including your Institute and is included in Index Medicus/MEDLINE and name and address, to the Editorial Coordinator, as EMBASE/Excerpta Medica, Thomson Reuters Science listed to the left. Citation Index Expanded (SciSearch) and Journal Citation Reports/Science Edition. Cancer Control currently has an impact factor of approximately 2.655. To Order an Individual, Institutional, This peer-reviewed journal contains articles on the or International Subscription: spectrum of actions and approaches needed to reduce For information on subscription rates, please contact the impact of human malignancy. the Editorial Coordinator. Cancer Control is sent at no charge to approximately 15,000* medical professionals, including oncologists To Request Permission to Reprint in all subspecialties, selected primary care physicians, Copyrighted Materials: medical researchers who specialize in oncology, and To request permission to reprint tables, figures, or others who have a professional interest in cancer control. materials copyrighted by Cancer Control, please *This includes approximately 1,000 oncologists from more contact the Editorial Coordinator. than 85 countries. To Access the Journal Online: To Inquire About Reprints: Most issues and supplements of Cancer Control For information on obtaining reprints and pricing in- are available at cancercontroljournal.org. formation, please contact the Editorial Coordinator.

To Change Your Mailing Address: To Submit an Article for Publication: Please provide your old address and your The editor welcomes submission of manuscripts new address to: pertaining to all phases of oncology care for possible Veronica Nemeth, Editorial Coordinator publication in Cancer Control. Articles are subject to Cancer Control Journal editorial evaluation and peer review. Author guide- Moffitt Cancer Center MBC-JRNL lines are available online at cancercontroljournal.org. 12902 Magnolia Drive Tampa, FL 33612 E-mail: [email protected] Follow us on Twitter Fax: 813-449-8680 @MoffittResearch #MoffittCCJ Phone: 813-745-1348

Celebrating More Than 20 Years of Publishing YOUR OPINION COUNTS! Access our readership survey at cancercontroljournal.org

Cancer Control 272 April 2015, Vol. 22, No. 2 Save the Date

4TH ANNUAL BMT WINTER WORKSHOP December 4, 2015 | Orlando, Florida

University of Central Florida Rosen College | 9907 Universal Blvd, Orlando, FL 32819

3:00-6:30 PM Workshop | 6:30-8:30 PM Reception

HOST Claudio Anasetti, MD Moffitt Cancer Center CO-CHAIR Marcel R.M. van den Brink, MD, PhD Memorial Sloan Kettering Cancer Center CO-CHAIR Edmund K. Waller, MD, PhD Winship Cancer Center | Emory University

Moffitt Cancer Center is pleased to host the 4th Annual BMT Winter Workshop, taking place in Orlando, Florida, prior to the 57th annual meeting of the American Society of Hematology. This informal meeting is formatted as previous workshops with numerous short talks on late-breaking research in BMT. Presenters will discuss recent unpublished clinical and preclinical data pertinent to hematopoietic stem cell transplantation. Registration will be available September 1, 2015. CONFERENCE CONTACTS [email protected] | [email protected] MOFFITT.org/BMTworkshop2015

Free registration, no travel reimbursement, no speaker fees, and no pharmaceutical support.

January 2015, Vol 22, No.1 Cancer Control 273 SAVE THE DATE

MOFFITT CANCER CENTER PRESENTS NEW FRONTIERS IN UROLOGIC ONCOLOGY Saturday, August 15, 2015 Marriott Waterside | Tampa, FL

CONFERENCE DIRECTOR: Julio Pow-Sang, MD

This 1-day conference designed

for urologists, medical oncologists,

radiation oncologists, and researchers

will provide the newest research

and clinical advances in urologic

oncology focusing on prostate,

bladder, and kidney cancers.

CONTACTS: [email protected] [email protected] PROVIDED BY