2019, Vol. 15, Number 2, 85–138

Oncology in Clinical Practice 2019, Vol. 15, Number 2, 85–138

2019, Vol. 15, Number 2 ISSN 2450–1654

Beata Hryciuk, Bartosz Szymanowski, Michał Bieńkowski, Adrian Perdyan, Aleksandra Korwat, Kamil Winnik, Barbara Radecka, Jolanta Żok, Natalia Cichowska, Katarzyna Sosińska-Mielcarek, Rafał Pęksa, Renata Duchnowska Consistency in biomarkers expression between matched tissue microarray cores from primary gallblader and ovarian cancers

Marcin Kaczor, Rafał Wójcik, Joanna Połowinczak-Przybyłek, Piotr Potemski Critical appraisal of clinical trials in oncology — part I

Monika Konopka-Filippow, Ewa Sierko, Marek Z. Wojtukiewicz Benefits and difficulties during brain radiotherapy planning with hippocampus sparing

Ewa Cedrych, Ida Cedrych Neratinib in adjuvant treatment of patients with HER2-positive breast cancer — less is more?

Katarzyna Kozak, Piotr Rutkowski Why do we need a new BRAF-MEK inhibitor combination in melanoma?

Małgorzata Flis, Paweł Krawczyk, Izabella Drogoń, Katarzyna Kurek, Robert Kieszko, Janusz Milanowski The effectiveness of chemotherapy in small cell lung cancer patients with BRCA2 gene mutation and Schwartz-Bartter syndrome

Piotr Tomczak, Zuzanna Synowiec Nivolumab in the treatment of advanced renal cell carcinoma

Aneta Lebiedzińska, Dawid Sigorski, Maciej Michalak, Zygmunt Kozielec, Anna Doboszyńska, Dariusz Zadrożny, Paweł Różanowski Complete pathological remission after palliative therapy with sorafenib in hepatocellular carcinoma — case report

Kamila Kaźmierczak, Joanna Kufel-Grabowska, Tomasz Kozłowski, Błażej Nowakowski When to say “no” to a patient with an ovarian tumour and in poor general condition?

Maciej Kawecki Current literature review

ONCOLOGY IN CLINICAL PRACTICE Official Journal of the Polish Society of Clinical Oncology https://journals.viamedica.pl/oncology_in_clinical_practice

Editor-in-Chief dr med. Aleksandra Łacko prof. dr hab. med. Maciej Krzakowski prof. Ruggero De Maria (Rome, Italy) dr hab. med. Radosław Mądry Deputy Editors dr med. Janusz Meder prof. dr hab. med. Andrzej Kawecki dr hab. med. Sergiusz Nawrocki prof. dr hab. med. Piotr Potemski prof. dr hab. med. Włodzimierz Olszewski prof. dr hab. med. Piotr Rutkowski prof. dr hab. med. Maria Podolak-Dawidziak prof. dr hab. med. Krzysztof Składowski dr med. Barbara Radecka prof. dr hab. med. Piotr Wysocki prof. dr hab. med. Tadeusz Robak prof. dr hab. med. Kazimierz Roszkowski Scientific Board prof. dr hab. med. Piotr Rutkowski dr Edita Baltruskeviciene (Vilnius, Lithuania) dr hab. med. Ewa Sierko prof. Tomasz M. Beer (Portland, USA) dr Ryszard Szydło (London, UK) dr med. Joanna Didkowska prof. dr hab. med. Jerzy Walecki dr hab. med. Renata Duchnowska dr med. Beata Jagielska prof. dr hab. med. Jan Walewski dr med. Jerzy Jarosz prof. dr hab. med. Krzysztof Warzocha prof. dr hab. med. Jacek Jassem prof. dr hab. med. Marek Wojtukiewicz prof. dr hab. med. Arkadiusz Jeziorski dr Agnieszka Wozniak (Leuven, Belgium) prof. dr hab. med. Jan Kornafel prof. Christoph Zielinski (Vienna, Austria) prof. dr hab. med. Jan Kulpa prof. dr hab. med. Radzisław Kordek Managing Editor dr hab. med. Maria Litwiniuk Izabela Siemaszko

Opinions presented in the articles do not necessarily represent the opinions of the Editors

Oncology in Clinical Practice (ISSN 2450–1654) is published six times a year by VM Media sp. z o.o. VM Group sp.k. ul. Świętokrzyska 73, 80–180 Gdańsk, Poland Phone: (+48 58) 320 94 94, fax: (+48 58) 320 94 60 e-mail: [email protected], http://www.viamedica.pl, wap.viamedica.pl

Editorial Address Klinika Nowotworów Płuca i Klatki Piersiowej Centrum Onkologii — Instytut im. Marii Skłodowskiej-Curie ul. Roentgena 5, 02–781 Warszawa, Poland Phone: (+48 22) 546 21 69 e-mail: [email protected]

Advertising For details on media opportunities within this journal please contact the advertising sales department, ul. Świętokrzyska 73, 80–180 Gdańsk, Poland, phone: (+48 58) 326 78 20; e-mail: [email protected]

The Editors accept no responsibility for the advertisement contents.

All rights reserved, including translation into foreign languages. No part of this periodical, either text or illustration, may be used in any form whatsoever. It is particularly forbidden for any part of this material to be copied or translated into a mechanical or electronic language and also to be recorded in whatever form, stored in any kind of retrieval system or transmitted, whether in an electronic or mechanical form or with the aid of photocopying, microfilm, recording, scanning or in any other form, without the prior written permission of the publisher. The rights of the publisher are protected by national copyright laws and by international conventions, and their violation will be punishable by penal sanctions. Legal note: http://czasopisma.viamedica.pl/owpk/about/legalNote

Indexed in Index Copernicus (ICV 2018 = 84.45), Ulrich’s Periodicals Directory and CAS. According to the statement of the Polish Ministry of Science and Higher Education publication in the journal has been awarded with 20 points. Editorial policies and author guidelines are published on journal website: http://journals.viamedica.pl/oncology_in_clinical_practice

The journal “Oncology in Clinical Practice” is financed under Contract No. 790/P-DUNdem/2019 by the funds of the Minister of Science and Higher Education for the science promotion activities.

Case Report Contest Policies

This policy defines the scope, requirements and regulations regarding The Krzysztof Krzemieniecki Award for the best case report published in “Oncology in Clinical Practice” (OCP) Third Edition. 1. The aim of the contest is to encourage submission of quality case reports related to oncological practice and to promote them in the scientific deliberations. 2. All respective manuscripts submitted to OCP between June 1st, 2018 and May 31st, 2019 and accepted for publication will qualify. 3. Manuscripts should be prepared in line with Authors’ guidelines and should be submitted only through the manuscript system available at Journal’s website: https://journals.viamedica.pl/oncology_in_clinical_practice 4. All submitted manuscripts will be evaluated during the peer review process and authors will be informed about their qualification for publication in OCP. Accepted papers will be evaluated by the Contest Committee based upon fulfillment of the Contest criteria as well as practical significance, originality, applicability and addressing of current/critical concerns. 5. The first author of the winning paper will be eligible for a prize of gross 1000,00 Euro gross (one thousand euro). 6. Results will be announced during the XXII National Congress of The Polish Society of Clinical Oncology which will take place between August 29th and 31st, 2019 and subsequently at the Journal website. 7. Winner will be notified via email. 8. Contest Committee may exclude a paper from participation in case of potential conflict of interest or ask submitting author for adequate clarifications. 9. The Sponsor at any stage and in any respect, will not participate in the evaluation of entries and selection of a winning paper. 10. The award amount shall be paid based on the copyright transfer agreement to the paper. 11. These Regulations are the sole and exclusive document defining the principles and conditions for the Contest. In all matters not regulated, decisions are made by The Organizer.

Contest Organizer: VM Media sp. z o.o. VM Group sp. k., seated at 73 Swietokrzyska Street, 80-180 Gdansk, Poland (Register of Entrepreneurs kept by the District Court for Gdansk, Commercial Division VII of the National Court Register under KRS No 0000266430, VAT Reg. No PL 583-28-39-187).

Patronage ONCOLOGY IN CLINICAL PRACTICE Official Journal of the Polish Society of Clinical Oncology https://journals.viamedica.pl/oncology_in_clinical_practice 2019, Vol. 15, Number 2

ORIGINAL ARTICLE Consistency in biomarkers expression between matched tissue microarray cores from primary gallblader and ovarian cancers Beata Hryciuk, Bartosz Szymanowski, Michał Bieńkowski, Adrian Perdyan, Aleksandra Korwat, Kamil Winnik, Barbara Radecka, Jolanta Żok, Natalia Cichowska, Katarzyna Sosińska-Mielcarek, Rafał Pęksa, Renata Duchnowska...... 85

REVIEW ARTICLES Critical appraisal of clinical trials in oncology — part I Marcin Kaczor, Rafał Wójcik, Joanna Połowinczak-Przybyłek, Piotr Potemski...... 89

Benefits and difficulties during brain radiotherapy planning with hippocampus sparing Monika Konopka-Filippow, Ewa Sierko, Marek Z. Wojtukiewicz...... 104

Neratinib in adjuvant treatment of patients with HER2-positive breast cancer — less is more? Ewa Cedrych, Ida Cedrych...... 111

Why do we need a new BRAF-MEK inhibitor combination in melanoma? Katarzyna Kozak, Piotr Rutkowski...... 115

CASE REPORTS The effectiveness of chemotherapy in small cell lung cancer patients with BRCA2 gene mutation and Schwartz-Bartter syndrome Małgorzata Flis, Paweł Krawczyk, Izabella Drogoń, Katarzyna Kurek, Robert Kieszko, Janusz Milanowski...... 120

Nivolumab in the treatment of advanced renal cell carcinoma Piotr Tomczak, Zuzanna Synowiec...... 124

Complete pathological remission after palliative therapy with sorafenib in hepatocellular carcinoma — case report Aneta Lebiedzińska, Dawid Sigorski, Maciej Michalak, Zygmunt Kozielec, Anna Doboszyńska, Dariusz Zadrożny, Paweł Różanowski ...... 127

When to say “no” to a patient with an ovarian tumour and in poor general condition? Kamila Kaźmierczak, Joanna Kufel-Grabowska, Tomasz Kozłowski, Błażej Nowakowski ...... 132

Current literature review Maciej Kawecki...... 135

ORIGINAL ARTICLE

Beata Hryciuk1*, Bartosz Szymanowski2*, Michał Bieńkowski3, Adrian Perdyan4, Aleksandra Korwat3, Kamil Winnik5, Barbara Radecka6, Jolanta Żok7, Natalia Cichowska8, Katarzyna Sosińska-Mielcarek8, Rafał Pęksa3, Renata Duchnowska2* 1Mazovian Centre for Lung Diseases and Tuberculosis, Division III, Otwock, Poland 2Oncology Clinic Military Institute of Medicine, Warsaw, Poland 3Chair and Department of Pathomorphology, Medical University of Gdansk, Poland 4Medical Faculty, Medical University of Gdansk, Poland 5Department of Pathomorphology, of Janusz Korczak Provincial Specialist Hospital in Slupsk, Poland 6Institute of Medicine, Opole University, Opole, Poland 7Provincial Centre of Oncology in Gdansk, Poland 8Department of Oncology and Radiotherapy, Medical University of Gdansk, Poland *On co-authorship rights

Consistency in biomarkers expression between matched tissue microarray cores from primary gallblader and ovarian cancers

Address for correspondence: ABSTRACT Dr hab. n. med. Renata Duchnowska, Introduction. Tissue microarray (TMA) technique has been widely used, especially in immunohistochemical assays prof. nadzw. WIM of new prognostic and predictive markers. The main objections raised by its opponents are the small amount of Wojskowy Instytut Medyczny CSK MON sampled material and the associated risk of inadequate assessment of analysed expression, resulting from the ul. Szaserów 128, 04–141 Warszawa potential heterogeneity of tumour tissue. e-mail: [email protected] Material and methods. This study evaluated the compatibility of biomarker expression in two independent tissue

Oncology in Clinical Practice cores, 1.5 mm in diameter, obtained by TMA technique from patients with gallbladder cancer (ERb, cytoPgR, HER2, 2019, Vol. 15, No. 2, 85–88 CTGF) and ovarian cancer (PTEN, BCL2, PIK3CA, IGF1R). Comparison of the expression of individual biomark- DOI: 10.5603/OCP.2019.0011 ers between cores was performed using the intraclass correlation coefficient (ICC), assuming a kappa < 0.4 as Translation: dr n. med. Dariusz Stencel a weak, ≥ 0.4 as sufficient,≥ 0.6 as good, and ≥ 0.75 as optimal correlation, and Kendall’s tau test — ICC package. Copyright © 2019 Via Medica Results. Evaluation of biomarker expression in the primary tumour was performed in 60 patients with gallbladder ISSN 2450–1654 cancer and in 64 patients with high-grade serous ovarian cancer. Additionally, in patients with follicular cancer, the expression of the tested markers was assessed in the epithelium free from neoplastic malignancy. In both tumours, a good or sufficient level of homogeneity was observed in the expression of the analysed biomarkers between tissue cores. The correlation coefficient for the expression of individual markers in gallbladder cancer and adhering healthy tissue was: 0.68 (95% CI: 0.53–0.79)/0.62 (95% CI: 0.39–0.78) for ERb, 0.44 (95% CI: 0.23–0.61)/0.77 (95% CI: 0.61–0.87) for cytoPgR, 0.77 (95% CI: 0.65–0.85)/0.66 (95% CI: 0.44–0.80) for HER2, and 0.68 (95% CI: 0.53–0.79)/0.62 (95% CI: 0.39–0.78) for CTGF. In patients with ovarian cancer, the correlation coefficient within the primary tumour was 0.82 (95% CI: 0.71–0.89) for PTEN, 0.84 (95% CI: 0.75–0.90) for BCL2, 0.71 (95% CI: 0.56–0.81) for PIK3CA, and 0.77 (95% CI: 0.65–0.85) for IGF1R. Conclusions. Tissue microarray technique allows reliable assessment of the expression of tissue biomarkers within the primary tumour of gallbladder cancer and ovarian cancer. Key words: tissue microarrays, biomarkers, gallbladder cancer, ovarian cancer

Oncol Clin Pract 2019; 15, 2: 85–88

85 Oncology in clinical practice 2019, Vol. 15, No. 2

Introduction encoded by the PTEN suppressor gene (phosphatase and tensin homolog deleted on chromosome 10) on the The technique of tissue microarray (TMA) was long arm of chromosome 10, proteins belonging to the first described in the 1980s [1]. In the following years, BCL2 family (B-cell CLL/lymphoma 2), protein of the a modified method has been widely used, especially in catalytic subunit a phosphatidyl inositol 3-kinase (PI3K- immunohistochemical studies on new prognostic and CA), and insulin-like growth factor-1 receptor (IGF1R). predictive markers [2, 3]. It enables tissue material from tens or even hundreds of patients to be placed on a single Preparation of tissue microarrays microscope slide. In the first stage, the pathologist makes a microscopic evaluation of the whole specimen stained In the analysed group, sections stained with haema- with haematoxylin and eosin to determine the most toxylin and eosin were subjected to histopathological representative necrosis-free tumour area for further reassessment, which allowed verification of the diagnosis analysis. In the second stage, from a paraffin tissue block and determination of the most representative fragments (the so-called “donor”) containing a formalin-fixed of cancer and healthy tissues. Selected samples together fragment of the tumour, a small, cylindrical core with with the corresponding paraffin blocks were used to a diameter of 0.6 to 2 mm is collected using a special determine the tumour areas from which the sections for needle. This core is then placed in a pre-prepared hole tissue microarray were taken using a 1.5 mm diameter located in another paraffin block called the “recipient”. needle. Biopsy specimens of tumour-containing frag- To increase the representativeness of the material ments were placed in previously prepared, tissue-free being tested and to reduce the risk of tissue loss in the paraffin blocks — “recipients”. Tissue microarrays staining process, at least two cores are usually taken were performed using a Manual Tissue Arrayer I by for each case. In addition, a map is created containing Beecher Instruments (MTAI, K7 BioSystems). Two information about the location of the material, which fragments (biopsies) of primary tumours were collected allows it to be quickly identified in the block. After in both groups, and in the gallbladder cancer project, completion of the material collection process, sections additionally, excisions from adjacent healthy tissues. Im- are obtained for examination using the microtome; one munohistochemistry was performed on tissue sections of microscopic slide usually contains of 50 to 150 cases microarrays with a thickness of 4 μm. Table 1 presents [4]. The main objection raised by the opponents of this a list of the antibodies used in the study along with method is the small amount of material tested and the the methodology of performing immunohistochemical associated risk of inadequate assessment of analysed staining. biomarker expression resulting from the potential heterogeneity of tumour tissue. Data on the reliability Statistical analysis of TMA in gallbladder and ovarian cancer are scarce. This study evaluated the compatibility of biomarker Statistical analysis was performed using the statistical expression between two tissue cores obtained by TMA environment R, version 3.4.3 [5] on the basis of data in both tumours. contained in a specially prepared database. A comparison of the expression of individual biomarkers between the “tissue cores” was performed using the intraclass Material and methods correlation coefficient (ICC), assuming kappa < 0.4 as weak, ≥ 0.4 as sufficient, ≥ 0.6 as good and ≥ 0.75 as opti- Characteristics of the assessed biomarkers (proteins) mal correlation, and Kendall tau test — ICC package [6].

The analysis included patients in whom the expression of a panel of tissue biomarkers was examined as part Results of two retrospective clinical studies. Proteins for immunohistochemical analysis were selected on the basis of In the gallbladder cancer project, biomarker expres- available literature, taking into account the availability sion was evaluated in tissue material from cholecystec- of antibodies and technical feasibility of assessment on tomy in 60 patients treated between 2004 and 2016 in archived formalin-fixed paraffin-embedded (FFPE) four oncology centres in Poland: The Military Institute tissue. In the project concerning gallbladder cancer, of Medicine in Warsaw, the University Clinical Centre of the expressions of following receptors were analysed: the Medical University of Gdansk in Gdansk, Professor steroid hormones receptors: estrogen a (ERa) and b Tadeusz Koszarowski Opole Oncology Centre in Opole, (ERb), progesterone (PgR), human epidermal growth and Janusz Korczak Provincial Specialist Hospital in factor 2 (HER2), and connective tissue growth factor Slupsk. In the ovarian cancer project, the analysis was (CTGF). In turn, in ovarian cancer, the expression of carried out in the primary tumour, in the postoperative the following proteins was determined: human protein material in 64 patients diagnosed with high-grade serous

86 Beata Hryciuk et al., Consistency in biomarkers expression between matched tissue microarray cores from primary gallblader and ovarian cancers

Table 1. Antibodies tested and immunohistochemical methods

Antibody Manufacturer Concentration Epitope Exposure Control Assessment Catalogue No. recovering time tissue method ERa DAKO; anti-human; RU HIER; DAKO PT-link, 20’ Breast cancer Semiquantitative rabbit clone EP1 high pH ERb Abcam; anti-human; 1:70 HIER; DAKO PT-link, Night Breast cancer Semiquantitative rabbit clone EPR3778; high pH incubation ab133467 PgR DAKO; anti-human; RU HIER; DAKO PT-link, 20’+linker Breast cancer Semiquantitative mouse clone 636 high pH mouse 15’ HER2 Ventana; rabbit clone RU Epitope recovering in 20’ Breast cancer Semiquantitative 4B5 the machine CTGR Santa Cruz, 1:100 HIER, DAKO PT-link, 60’ Smooth muscles Semiquantitative California; high pH goat sc-14939 PTEN DAKO; clone 6H2.1 1:50 HIER, DAKO PT-link, 30’ Placenta Semiquantitative high pH BCL2 DAKO monoclonal RU HIER, DAKO PT-link, 20’ Lymph node Semiquantitative mouse clone 124 high pH PIK3CA Cell signalling Rabbit 1:50 HIER, DAKO PT-link, 60’ Breast cancer Semiquantitative monoclonal low pH IGF1R Roche Rabbit RU Epitope recovering in 30’ Placenta Semiquantitative Monoclonal (G11) the machine

cancer, treated surgically between 2010 and 2016 at the Table 2. Compatibility analysis for ERb, cytoPgR, HER2, Military Institute of Medicine in Warsaw. and CTGF expression between tissue cores in gallbladder In both tumours, a good or sufficient level of homo- cancer and adherent healthy tissue (intraclass correlation geneity was observed in the expression of the analysed coefficient [ICC], assuming kappa: < 0.4 as weak, ≥ 0.4 biomarkers between tissue cores. ERa expression was as sufficient,≥ 0.6 as good, and ≥ 0.75 as optimal correlation, not demonstrated in gallbladder and healthy tissue. The and Kendall tau test — ICC package) correlation coefficient for the expression of other bio- HER2 markers in gallbladder carcinoma and adhering healthy In total 0.74 (95% CI: 0.64–0.82) tissue was: 0.68 (95% CI: 0.53–0.79)/0.62 (95% CI: 0.39–0.78) for ERb, 0.44 (95% CI: 0.23–0.61) 0.77 (95% Gallbladder cancer 0.77 (95% CI: 0.65–0.85) CI: 0.61–0.87) for cytoplasmic PgR, 0.77 (95% CI: Healthy tissue 0.66 (95% CI: 0.44–0.80) 0,65–0.85)/0.66 (95% CI: 0.44–0.80) for HER2, and cytoPgR 0.68 (95% CI: 0.53–0.79)/0.62 (95% CI: 0.39–0.78) for In total 0.80 (95% CI: 0.73–0.86) CTGF. In patients with ovarian cancer, the correlation Gallbladder cancer 0.44 (95% CI: 0.23–0.61) coefficient within the primary tumour was 0.82 (95% Healthy tissue 0.77 (95% CI: 0.61–0.87) CI: 0.71–0.89) for PTEN, 0.84 (95% CI: 0.75–0.90) for BCL2, 0.71 (95% CI: 0.56–0.81) for PIK3CA, and CTGF 0.77 (95% CI: 0.65–0.85) for IGF1R (Table 2 and 3). In total 0.66 (95% CI: 0.55–0.76) Gallbladder cancer 0.68 (95% CI: 0.53–0.79) Healthy tissue 0.62 (95% CI: 0.39–0.78) Discussion ERb In total 0.66 (95% CI: 0.55–0.76) Neoplasms are heterogeneous in nature, which means that there may be significant genotype differences Gallbladder cancer 0.68 (95% CI: 0.53–0.79) in the primary tumour or its distant lesions, resulting Healthy tissue 0.62 (95% CI: 0.39–0.78) from the selection of cell clones [7–9]. Therefore, the heterogeneity of tumours is spatial and temporal. In turn, in diagnostics and qualifications for treatment, mine reliable prognostic and predictive factors — bio- especially molecularly targeted, there is a need to deter- markers. Undoubtedly, intra-tumour heterogeneity in

87 Oncology in clinical practice 2019, Vol. 15, No. 2

Table 3. Compatibility analysis for PTEN, BCL2, PIK3CA, and cancer, the tissue microarray technique is a reliable IGF1R expression between tissue cores in ovarian cancer diagnostic method. (intraclass correlation coefficient [ICC], assuming kappa: < 0.4 as weak, ≥ 0.4 as sufficient,≥ 0.6 as good, and ≥ 0.75 as optimal correlation, and Kendall tau test — ICC package) References PTEN 1. Battifora H. The multitumor (sausage) tissue block: novel method for In total 0.82 (95% CI: 0.71–0.89) immunohistochemical antibody testing. Lab Invest. 1986; 55(2): 244–248, indexed in Pubmed: 3525985. BCL2 2. Hewitt SM. Tissue microarrays as a tool in the discovery and validation In total 0.84 (95% CI: 0.75–0.90) of predictive biomarkers. Methods Mol Biol. 2012; 823: 201–214, doi: 10.1007/978-1-60327-216-2_13, indexed in Pubmed: 22081347. PIK3CA 3. Kononen J, Bubendorf L, Kallioniemi A, et al. Tissue microarrays for high-throughput molecular profiling of tumor specimens. Nat Med. 1998; In total 0.71 (95% CI: 0.56–0.81) 4(7): 844–847, indexed in Pubmed: 9662379. IGF1R 4. Camp RL, Neumeister V, Rimm DL. A decade of tissue microarrays: progress in the discovery and validation of cancer biomarkers. J Clin In total 0.77 (95% CI: 0.65–0.85) Oncol. 2008; 26(34): 5630–5637, doi: 10.1200/JCO.2008.17.3567, indexed in Pubmed: 18936473. 5. Team RC (2015) R: A Language and Environment for Statistical Com- puting (R Foundation for Statistical Computing, Vienna, 2015). http:// www.R-project.org. neoplastic disease can lead to erroneous conclusions and 6. Wolak ME, Fairbairn DJ, Paulsen YR. Guidelines for estimating repeat- hinder the development of personalised medicine [7–9]. ability. Methods in Ecology and Evolution. 2012; 3: 129–137. For this reason, validation of diagnostic methods used 7. Gerlinger M, Rowan AJ, Horswell S, et al. Intratumor heterogeneity and branched evolution revealed by multiregion sequencing. N Engl J Med. in scientific research is very important. The technique of 2012; 366(10): 883–892, doi: 10.1056/NEJMoa1113205, indexed in tissue microarray, due to the gathering of material from Pubmed: 22397650. 8. Gerlinger M, Horswell S, Larkin J, et al. Genomic architecture and different patients on one slide, significantly shortens the evolution of clear cell renal cell carcinomas defined by multiregion time of staining and evaluation, saves tissue material sequencing. Nat Genet. 2014; 46(3): 225–233, doi: 10.1038/ng.2891, indexed in Pubmed: 24487277. and the amount of reagents used, and allows testing in 9. Seoane J, De Mattos-Arruda L. The challenge of intratumour hetero- uniform conditions and with the same dilutions of the geneity in precision medicine. J Intern Med. 2014; 276(1): 41–51, doi: antibodies used. On the other hand, the evaluation of 10.1111/joim.12240, indexed in Pubmed: 24661605. 10. Camp RL, Charette LA, Rimm DL. Validation of tissue microarray technol- such small fragments of tissue raises doubts as to their rep- ogy in breast carcinoma. Lab Invest. 2000; 80(12): 1943–1949, indexed resentativeness in relation to the whole tumour. Previous in Pubmed: 11140706. 11. Zhang D, Salto-Tellez M, Putti TC, et al. Reliability of tissue microarrays in studies on this issue, carried out in various cancers, indi- detecting protein expression and gene amplification in breast cancer. Mod cate high consistency of results evaluated in microarrays Pathol. 2003; 16(1): 79–84, doi: 10.1097/01.MP.0000047307.96344.93, indexed in Pubmed: 12527717. and in full tumour sections [10–18]. In individual studies, 12. Fonseca FP, de Andrade BA, Rangel AL, et al. Tissue microarray is the discrepancy in the number of cores needed to obtain a reliable method for immunohistochemical analysis of pleomorphic an acceptable sample representation could be due to the adenoma. Oral Surg Oral Med Oral Pathol Oral Radiol. 2014; 117(1): 81–88, doi: 10.1016/j.oooo.2013.08.029, indexed in Pubmed: 24332331. heterogeneity of the expression of antigens in tumours 13. Khouja MH, Baekelandt M, Sarab A, et al. Limitations of tissue microarrays [14, 16, 17, 19]. In a breast cancer study it was found that compared with whole tissue sections in survival analysis. Oncol Lett. 2010; 1(5): 827–831, doi: 10.3892/ol_00000145, indexed in Pubmed: 22966388. one or two TMA cores in each case yielded results that 14. Griffin MC, Robinson RA, Trask DK. Validation of tissue microarrays were 95% similar to those obtained using tumour sections using p53 immunohistochemical studies of squamous cell carcinoma of the larynx. Mod Pathol. 2003; 16(12): 1181–1188, doi: 10.1097/01. [10]. However, most validation studies have shown that MP.0000097284.40421.D6, indexed in Pubmed: 14681317. analysis of two to three cores with a diameter of 0.6 mm 15. Jourdan F, Sebbagh N, Comperat E, et al. Tissue microarray technol- gives higher compliance rates than using one core [10, ogy: validation in colorectal carcinoma and analysis of p53, hMLH1, and hMSH2 immunohistochemical expression. Virchows Arch. 2003; 14–16]. Therefore, two cores, 1.5 mm in diameter, were 443(2): 115–121, doi: 10.1007/s00428-003-0833-z, indexed in Pubmed: used in this work. High homogeneity in the expression of 12802583. 16. Gomaa W, Ke Y, Fujii H, et al. Tissue microarray of head and neck the analysed biomarkers with the use of tissue microarray squamous carcinoma: validation of the methodology for the study technology in tumours has been demonstrated, which until of cutaneous fatty acid-binding protein, vascular endothelial growth factor, involucrin and Ki-67. Virchows Arch. 2005; 447(4): 701–709, doi: now have not been the subject of a similar assessment. The 10.1007/s00428-005-0002-7, indexed in Pubmed: 16012850. reliability and usefulness of this method in the diagnosis 17. Su Y, Shrubsole MJ, Ness RM, et al. Immunohistochemical expressions of of other cancers requires similar research. Ki-67, cyclin D1, beta-catenin, cyclooxygenase-2, and epidermal growth factor receptor in human colorectal adenoma: a validation study of tissue microarrays. Cancer Epidemiol Biomarkers Prev. 2006; 15(9): 1719–1726, doi: 10.1158/1055-9965.EPI-05-0946, indexed in Pubmed: 16985035. 18. Rosen DG, Huang X, Deavers MT, et al. Validation of tissue microarray Conclusions technology in ovarian carcinoma. Mod Pathol. 2004; 17(7): 790–797, doi: 10.1038/modpathol.3800120, indexed in Pubmed: 15073602. 19. Leversha MA, Fielding P, Watson S, et al. Expression of p53, pRB, and p16 In immunohistochemical studies on new prognostic in lung tumours: a validation study on tissue microarrays. J Pathol. 2003; and predictive biomarkers in gallbladder and ovarian 200(5): 610–619, doi: 10.1002/path.1374, indexed in Pubmed: 12898597.

88 REVIEW ARTICLE

Marcin Kaczor1, 2, Rafał Wójcik2, Joanna Połowinczak-Przybyłek3, Piotr Potemski3 1Jagiellonian University Medical College, Krakow, Poland 2Aestimo, Krakow, Poland 3The Department of Chemotherapy, Copernicus Memorial Multidisciplinary Centre for Oncology and Traumatology, Lodz; Chemotherapy Clinic, Medical University of Lodz, Poland

Critical appraisal of clinical trials in oncology — part I

ABSTRACT Address for correspondence: The main concept of evidence-based medicine is that all therapeutic decisions should be based on results from Dr n. med. Marcin Kaczor relevant, credible, and up-to-date clinical trials. Availability of a publication presenting a description of a clinical trial II Katedra Chorób Wewnętrznych conducted with reliable methods and its high-quality results seems to be an ideal situation from the practitioner’s im. prof. Andrzeja Szczeklika point of view. However, reading only the abstract or just the author’s conclusions may not always be sufficient Uniwersytet Jagielloński to make the right clinical decision. For this purpose, several aspects of the clinical trial should be put under as- Collegium Medicum w Krakowie sessment, namely the methodology, its quality, internal and external credibility, clinical and statistical significance, e-mail: [email protected] as well as consistency of the results. The ability to perform the proper assessment of clinical trials may prove to be very helpful for practicing oncologists, especially in the case of new, emerging therapies, specific clinical Oncology in Clinical Practice situations, or when salvage treatment is necessary. It is also worth emphasising that the outcome assessment 2019, Vol. 15, No. 2, 89–103 DOI: 10.5603/OCP.2018.0057 in oncology trials is specific, mainly due to the role of the survival analysis, which is relatively difficult to interpret. Translation: dr n. med. Dariusz Stencel In this paper we tried to present in a clear and intelligible way the theoretical basis and subsequent steps in the Copyright © 2019 Via Medica critical appraisal of methods and results of clinical trials in oncology. ISSN 2450–1654 Key words: oncology, randomised clinical trial, critical appraisal, statistical analysis, survival analysis

Oncol Clin Pract 2019; 15, 2: 89–103

Introduction be made, to find out whether the results of a clinical trial recognized as internally reliable can be extrapolated Decisions regarding the choice of treatment are to the population subjected to treatment under real made based on correctly performed and reliable clinical clinical practice, and whether similar clinical effects trials. The results of clinical trials are used to develop could be expected in these circumstances (patients’ the current guidelines for clinical practice in accord- characteristics, additional medical procedures, appro- ance with the principles of evidence-based medicine priate comparator, compliance of study participants). (EBM). To assess whether the conclusions from the Finally, clinical significance of the results should be study are appropriate, first of all it should be critically assessed to answer the question of whether the magni- analysed for internal credibility. In order to do this, it tude of the observed effect indicates significant clinical should be assessed whether the study has been carried benefit (taking into account the prognosis in a given out correctly (an appropriate methodology ensuring patient population) and whether it really should lead reliable and undistorted inference and proper statistical to a change in clinical practice [1]. analysis) and whether there is internal consistency of The individual elements of critical appraisal of conclusions in a range of individual endpoints. External clinical trials are discussed below. In addition, taking consistency assessment can be also helpful, determining into account the specifics of clinical trials in oncology, whether a similar effect was observed in other clinical the analysis of “time to event” endpoints is presented trials. Then an assessment of external credibility should in more detail.

89 Oncology in clinical practice 2019, Vol. 15, No. 2

Clinical trial methodology unequal distribution to the studied groups is used, e.g. in a 2:1 ratio, which may increase the amount of informa- Correctly designed and conducted, blinded, ran- tion about a new therapy, especially regarding safety as domised clinical trials (RCTs) provide evidence with the well as recruitment capacity (patients are more willing highest level of credibility [2]. These are experimental to participate in the study due to a greater chance of tests that assess at least two therapeutic interventions, receiving experimental therapy), but it adversely affects and their use in patients is strictly controlled according to the statistical power and requires a higher sample size a previously developed study protocol. In oncology these compared to allocation with a 1:1 ratio [3]. studies usually have the character of trials with parallel groups. In some populations of oncological patients it Allocation concealment and blinding is difficult to carry out a randomised trial, which may be due to the low prevalence of some cancers or small With random assignment of patients to study arms numbers of patients in specific clinical stages or treat- the process of allocation concealment is very impor- ment lines. It these cases it is necessary to conduct the tant, to prevent access to information about the group study without a control group (single-arm). However, to which the patient was assigned — which is possible the methodological quality of such studies is initially with use of central randomisation, performed regard- lower than that of randomised trials. The same applies less of the individuals participating in the study. Ad- to cohort studies, which include a control group, but, ditionally, allocation concealment allows elimination due to the lack of randomisation, the non-random dis- of influence of the researcher on patient assignment tribution of disturbing factors is a burden, and inference to particular groups, thereby reducing the risk of selec- about the observed differences in the effectiveness of tion bias. The second step that ensures greater cred- therapy is limited [1]. ibility is the introduction of blinding; therefore, the patient (single-blinded) or the patient and investigator Randomisation (double-blinded) or patient, investigator, and team analysing the results (triple-blinded) are not aware of Randomisation (random allocation of patients to which intervention is received by each patient. It pro- respective groups) is used to obtain as similar as possible vides higher credibility of the study due to elimination or almost identical baseline clinical and demographic of some confounders — a terminally ill patient, who characteristics of patients, which, with an appropriately knows that he/she was assigned to a placebo group large population, ensures balanced distribution of all po- instead of an active intervention group, may present tential, as well as unknown, confounding factors. There- much worse results than a patient who is unaware of the fore, the randomisation procedure cannot be carried out study assignment [4, 5]. In the case of medicines blind- on the basis of simple assumptions, such as a medical ing is ensured through their preparation in the same history number or date of birth, because it allows the pre- form (e.g. in visually identical vials), and for different diction of which group a patient will be allocated (this is routes of administration or collation of different treat- called pseudorandomisation). Randomisation methods ment methods an additional important role is played by providing full randomness, i.e. unpredictability, include proper masking (dummy) of intervention, e.g. simulta- those in which the lists of random numbers are created neous administration of two interventions that differ with use of a computer or special tables (such a method by administration routes, but in each study arm a dif- is called simple randomisation). In the case of a small ferent intervention is replaced with a placebo. In some target number of patients (sample size) in the study the cases, e.g. different medical procedures it is difficult to probability of unbalanced number and distribution of ensure blinding or it is associated with high burden to patients’ characteristics in individual groups is higher; the patients. It should be remembered that the lack of in this case more complex randomisation methods can blinding significantly affects primarily the evaluation be used. Examples of these include: block randomisa- of the subjective endpoints, independently assessed by tion (patients are assigned to individual interventions in patients (PRO, patient-reported outcome; e.g. scoring blocks, or groups with a specific sequence of subsequent of symptom severity, quality of life) or safety analysis, patients allocation), stratified randomisation (independ- but does not disturb unambiguously objective endpoints, ent stratification in any previously defined layer, such such as death (and hence survival outcomes) [3]. During as country origin, gender, or type of previously used endpoint evaluation with use of pathological or imag- treatment, especially when differences in the effective- ing examination, or standardised criteria (e.g. RECIST ness of assessed intervention between these subgroups — Response Evaluation Criteria in Solid Tumours) the are expected), or adaptive randomisation (in which the risk of systematic error is ambiguous. On the other side, probability of allocation to a given group changes during in oncology studies, despite the blinding of imaging the study, allowing the control of distribution of indi- tests assessing a progression (response to treatment), vidual features in particular groups) [3]. In some studies they are centrally confirmed by an independent and

90 Marcin Kaczor et al., Critical appraisal of clinical trials in oncology — part I blinded committee. However, there is a risk, especially in —— selective presentation of results (reporting bias) placebo-controlled trial with crossover after disease pro- — whether the study protocol is available, and gression, that a lack of blinding will result in performing the publication presents the results for all prede- imaging examinations faster than planned, even upon fined endpoints; mild symptoms. There are also clinical trials in which —— other factors (other bias) — whether no other po- blinding of all researchers or patients is not required, tential sources of reduced reliability of presented but instead analysts evaluate the endpoints — these are results were found (such as incorrect study design referred to as PROBE (prospective, randomised, open, or the allegation of dishonesty). blinded-endpoint evaluation). It is worth noting that currently multicentre trials are preferred with appropriate representation of different Evaluation of study quality geographical regions [8], although they are associated with the risks of lowering the standardisation of the The simplest assessment of study credibility is pos- interventions used as well as the results [1]. sible with the use of the Jadad five-point scale [6]. It assesses whether the study was described as randomised, Defining the studied population whether double blinding was used, and provides information on how many patients discontinued the study The target study population should be described in and for what reasons. Additional points can be granted details and defined based on inclusion criteria. They or deducted depending on whether randomisation and are analysed to conduct an external credibility assess- blinding were or were not performed correctly. However, ment, e.g. determining the characteristics of patients for this scale allows only for very general assessment of study whom the conclusions of the study may be generalised. quality and does not take into account other factors that Too narrow and detailed inclusion criteria may limit could result in systematic error (bias). A more compre- the possibilities of recruiting patients to the study and hensive method is use of the Cochrane Collaboration the possibility of generalisation of conclusions, but too recommendations [7], according to which the following general inclusion criteria can cause dispersion of the as- aspects are assessed: sessed effect in subgroups with different characteristics, —— selection bias — whether the correct method of making it difficult to randomly distribute confounders randomisation and allocation concealment was used; and preventing subgroup analysis [3]. —— blinding of patients and medical staff (performance bias); Defining the comparator —— blinding of assessment of results (detection bias) — whether investigators were blinded or whether Another key element is the choice of a proper the authors of the publication justified that the lack comparator (control group), which determines the pos- of such blinding does not affect the assessment of sibility of further extrapolation of results on the target a given endpoint; in the case of assessment of end- population and the study’s external reliability. The points with different susceptibility to bias resulting optimal and desirable comparator is the current clinical from the lack of blinding, it is necessary to carry out practice, consistent with widely accepted recommenda- the evaluation for each of them separately; tions and guidelines [9]. However, placebo is often used —— incompleteness of results and loss of patients from in the control group. This is justified when new therapy the study (attrition bias) — the low risk of this type of is an add-on treatment to the current standard (then error is when the data lost does not interfere with the placebo is used only for blinding, and it's the current prac- assessment of endpoints, investigators have applied tice that is in fact the comparator) or when there is no the right method of imputation of missing data (e.g. other therapeutic option available in real-life conditions LOCF [last observation carried forward], in which except symptomatic treatment, e.g. when the evaluated for patients lost from observation, the individual intervention is the very last treatment line. A compari- values of assessed endpoints recorded during the last son with placebo is usually aimed at demonstrating the control visit are imputed for each subsequent time superiority of the new treatment. The choice of an active point until the end of the study), and the percentage intervention as a comparator always brings additional of patients excluded is not different between the challenges, also in the context of sample size, but use groups; in practice, it is assumed that if more than of placebo would be simply unethical. In the case of 10% of patients have been lost from the study, the comparison with active treatment, testing of the non-in- risk of systematic error resulting from data incom- feriority hypothesis may be considered [3]. The rationale pleteness is high, unless the frequency of individual for the selection of active intervention as a comparator causes of exclusion is similar and the percentage of should also be assessed in the context of changing clini- patients lost to follow-up is small; cal recommendations, especially in the case of clinical

91 Oncology in clinical practice 2019, Vol. 15, No. 2 trials planned a few years before. In oncology, in view predictor, although so far the conclusions presented of diversified chemotherapy regimens and treatment by many authors are ambiguous [10]. There are other methods, the investigator’s choice of therapy is often commonly used endpoints such as objective response accepted as a comparator. In such a situation, it should rate (ORR) based on imaging tests for solid tumours be assessed which interventions were used in the com- or haematological remission for haematological malig- parator group and how they were distributed, especially nancies together with the time of duration of response when symptomatic treatment is an option, and whether (DoR). As the alternative to PFS time to disease pro- they reflect clinical practice and possible extrapolation gression (TTP) is sometimes used. It differs from PFS in of conclusions (external credibility). that, that it comprises only events of progression, while observation of patients who died before its occurrence Defining the endpoints are censored at the time of death. Related endpoints, although much less frequently used in the assessment of Endpoints (outcomes) should be accurately defined palliative care effectiveness, include time to treatment in the study protocol, with the specification of primary failure (TTF) and time to next treatment (TTNT). (for which sample size and statistical power are esti- Considering the diversity of evaluated endpoints, mated) and additional/secondary endpoints. Evaluation the internal coherence of presented results should be of clinically relevant endpoints, such as overall survival highlighted, i.e. demonstrating a significant impact of (OS) and quality of life (general and aimed at evalua- the studied intervention on ORR, PFS, and then on tion of symptoms associated with a given type of cancer) OS. However, the following should always be remem- is also desirable. The possibilities to assess the impact bered: the differentiation of studied populations in of interventions on overall survival will depend on the terms of type and stage of cancer, prognosis and time type of cancer and its clinical stage. This analysis will of expected survival, and even the type of intervention undoubtedly be difficult in the case of assessment of used, e.g. demonstration of the impact of immuno- early stages of therapy with curative intent (e.g. neo- therapy on OS, in the absence of effects on PFS due to and adjuvant treatment), when the expected further pseudoprogression [11]. Regarding studies in oncology, survival could last for decades, and additional effects particular attention should be paid to safety assessment, of subsequent treatment lines, implemented after later including undesirable or fatal adverse reactions, which recurrences or progression, will have an impact on the in turn should include toxicity specific to the interven- observed differences in survival. In such cases, surrogate tion. Finally, the benefit-risk ratio should be evaluated, endpoints may include disease-free survival (DFS), taking into account the prognosis in a specific patient event-free survival (EFS), relapse-free survival (RFS), population [12, 13]. e.g. the time since the date of inclusion to the study to the date of occurrence the first documented clinical event Information about planned statistical analysis or death (whichever occurs earlier), for therapies used in the early stages of cancer, or progression-free survival The scope and type of statistical analysis in a cor- (PFS), e.g. the time from the date of randomisation rectly performed clinical trial should be predefined as until the date of progression or death, in the advanced part of a previously accepted protocol, together with stages. Clinical events included in PFS/DFS definition predefined matching factors and subgroup analysis. are observed earlier than death, therefore the observa- The initial estimation of sample size (statistical tion period necessary to show a statistically significant power of the study) is one of the key elements of the difference between the interventions is usually shorter statistical analysis. It allows assessment of whether the than for OS. Hence, the PFS assessment is preferred for sample is large enough to confirm or exclude differ- example when high clinical needs exist (no other effec- ences between interventions. The assessment of the tive treatment available), because the registration of the sample size refers to the main (primary) endpoint (or drug in a given indication can be obtained much faster endpoints). It requires the determination of the expected (even by several years) than if it would be necessary to frequency of events in the control group, the magnitude wait for OS outcome. In addition, the observed differ- of the intervention effect that the study is aimed to detect ences in PFS are not affected by successive treatment (an alternative hypothesis), the assumption of the ability lines and possible cross-over because further treatment to detect the real effect (statistical power of the study), is not usually introduced before disease progression. and the selection of the statistical significance level. Progression-free survival is assessed in the majority of In oncological studies with a long observation period studies with the treatment of advanced cancer stages; the expected discontinuation rate should also be taken nevertheless, it is considered as a surrogate endpoint. into account. Because statistical power depends on the There are many publications assessing the correlation number of patients experiencing a given event during between PFS and OS regarding PFS usefulness as OS observation, in oncological studies it is often assumed

92 Marcin Kaczor et al., Critical appraisal of clinical trials in oncology — part I that patients will be observed until the occurrence of an severity, previous treatment, and other factors that may expected number of events (e.g. deaths or deaths and affect the effectiveness of the assessed therapy — their disease progressions in evaluation of OS or PFS) [3]. scope and type depend on the type of cancer and should In the statistical analysis the researchers may adopt also be adapted to disease severity. The analysis of these different analytical approaches — the hypothesis of parameters can be used to assess the correctness of ran- superiority is tested most often, especially in the early domisation and eliminate the influence of confounders stages of clinical trials and in placebo-controlled tri- (analysis of such a table can refer only to those known als. The second approach is based on the assessment of factors, but at the same time it could be assumed that whether the intervention is not inferior in terms of clini- random assignment to groups with appropriate sample cal efficacy to currently used methods (non-inferiority), sizes also ensures equal distribution of other, unknown especially with a better safety profile. In this case, there prognostic factors). It is necessary to distinguish sub- is a need to establish a clinically acceptable variability groups defined for randomisation with stratification and of effectiveness in terms of primary endpoint, and if the subgroups, within which predefined analysis or possible appropriate confidence interval (CI) for the difference unplanned post-hoc analysis will be performed. This between interventions does not exceed the set level, information is also helpful in determining the external the intervention is considered to be no worse than the validity of study results. It allows also the assessment control. The use of the non-inferiority approach reduces of whether the analysed population is close to the one the required sample size [14, 15]. There are also studies in which the evaluated intervention is to be applied [8]. assessing the equivalence of interventions, where accept- Study outcomes in the form of categorical (nominal) able variability is assumed in both directions, but they variables are usually presented as numbers and percent- are rarely used to assess clinical endpoints — laboratory ages, while continuous variables are usually presented by parameters or pharmacokinetics are used instead. The means of a measure of central tendency and dispersion population included in the analysis is also important — usually mean and standard deviation (SD) values, — it may vary depending on the assessed endpoint. and in the case of variables that present the normal The ITT (intention-to-treat) population is included in distribution, median and range, possibly interquartile the analysis of the results of all randomised patients, range (IQR), are used (see below). In addition, some regardless of whether they received an assigned inter- continuous variables can be transformed into ordinal vention and regardless of how long they remained in the variables (e.g. the percentage of patients above a given observation (this usually applies to the assessment of OS age). Referring to the previously mentioned assessment or PFS). Sometimes a modified ITT (mITT) population of the accuracy of randomisation, it should be checked is defined, i.e. randomised patients who have received that there are no significant differences in baseline at least one dose of the study drug — a safety analysis characteristics between the groups — the authors should is usually performed in this population. Population PP provide P-values in the table or declare no significant (per-protocol) refers to patients who additionally did not differences in the publication text [8]. discontinue the treatment, did not violate the protocol, The study should also include detailed informa- and for whom a complete set of information is avail- tion (usually on the appropriate diagram) on patient able — it is often used to compare the effectiveness of flow from the screening period (i.e. from consent to interventions in non-inferiority trials [14]. ITT analysis participation in the study to inclusion) until a possible is more conservative because it tends to underestimate additional follow-up period. As was already mentioned, the beneficial clinical effect, whereas PP analysis allows this is an important element of assessment of study reli- a comparison of therapeutic options in conditions of ability — the size of the loss of patients from observation a complete observation. If the results obtained in ITT should be assessed, as well as how it can affect reliable and PP analysis clearly differ, this may indicate reduced analysis of results, and the occurrence of differences reliability of the study. The evaluation of objective between groups. response rate is often carried out in the population of The next step involves quantifying the differences patients for whom additional imaging results are avail- between interventions, exposing the uncertainty of able, i.e. there is a possibility to assess the progression. these estimates by means of confidence intervals, and evaluating the strength of evidence, i.e. confirming by means of P-value (statistical significance test) that the Evaluation of results observed difference is true and not by chance [8, 16–18]. Because, for obvious reasons, it is not possible to test The analysis of results of a clinical study begins with all patients in the considered clinical conditions, a sam- a detailed assessment of the description of the popula- ple should be selected (a group included into the clini- tion included and tables with baseline characteristics, cal trial), and, based on observed effects, conclusions which should include basic demographic data, disease should be drawn with some approximation about the

93 Oncology in clinical practice 2019, Vol. 15, No. 2 real overall effectiveness of treatment. In the language and a 2.5% probability that it is above this range (such of statistics, it is said that, based on the selected meas- a confidence interval results from the assumption ure of the effect determined in a random sample from of P-value < 0.05). The accuracy of the estimation this population, a parameter in the general population increases with the sample size: the larger the study, could be estimated, which in this approach comprises the more accurate the estimate and the narrower the all possible results of the experiment. confidence interval for the assessed parameter. With At the beginning a research hypothesis should be many repeated tests for effect measurement CI can precisely formulated and then tested with statistical be calculated in each of these tests, and 95% of them methods for its acceptance or rejection. By default, should contain a true value. The designated CI may it is assumed that evaluated interventions influence also be used to assess the statistical significance of the the health effect in a similar way (so-called null result if the entire interval indicates a coherent effect, hypothesis), and the observed differences are a result i.e. it does not contain a value indicating no differences of random variation derived from the limitations of (0 in the case of difference in continuous variables or the experiment (e.g. the group of patients being too probabilities and 1 when considering the hazard or pro small). An alternative hypothesis is that the observed bability ratio in both groups). The intervals that exclude differences are true and not just a random observa- these values indicate significant differences between tion. The role of statistics is to indicate which of these compared groups. hypotheses is more likely. Usually, a statistical evaluation will assess the three But how will we know that the groups do not differ types of variables in clinical studies: dichotomous (bina- from each other? It could be intuitively said that the ry) (e.g. response to treatment/no response), continuous lack of differences between these groups will be surely (e.g. average body weight), and time-to-event variables confirmed by the same percentage of patients with a re- used in survival analysis (e.g. OS). sponse in the intervention group as in the control group. However, if there are any differences, the probability is Evaluation of dichotomous variables estimated — designated as a P-value — to obtain a difference in treatment at least as high as that observed (in One of the most common types of variables is the both directions, i.e. in favour or not of the intervention number of patients in whom the assessed event occurred being analysed) in a situation in which a null hypothesis or not. Usually it is expressed in the form of numbers and were real. If the probability of the lack of differences percentages. It should always be remembered that this is between groups is below the statistical significance the number of cases with the first event being assessed, threshold of 5% adopted in the biomedical sciences which is not a problem if they are unique (e.g. death) (P < 0.05), it is assumed that these differences exist or rare. However, if they can be repeated many times, and are not the result of chance, so the null hypothesis such as febrile neutropaenia, assessment of the overall is rejected with conclusions of significant differences number of events can be more informative, preferably between the groups. In other words, this means that calculated per observation period (incidence rate per the probability of obtaining at least such a difference patient-year). The number of events can therefore be as demonstrated is less than 0.05. Thus, the lower the analysed as a continuous or dichotomous variable. P-value for a given estimate, the stronger the evidence Let us assume that patients in the study were against the hypothesis of the lack of differences and randomly assigned to two groups of 100 patients, one the greater the conviction about the effectiveness of receiving the active drug and the other placebo. After the intervention. Obviously, the statistical significance a year of observation, a clinical response was observed of the result indicates only that the observed reliance in 80 patients in the group with active treatment (80%) is more likely than would result from a simple random and only 40 (40%) patients in the control group. The case, but it does not mean that the observed effect is authors of this illustrative publication presented four real. Critical appraisal should also take into account parameters, reflecting the differences between both internal reliability and the influence of confounders groups in the frequency of response rate (it should be related to study methodology and conduction (including remembered that the event probability is 0.8 in the in- randomisation, blinding, and loss of patients). In addi- tervention group and 0.4 in the control group): tion, it is important to distinguish the difference between —— RB = 2.00 (95% CI: 1.54–2.59); P < 0.0001; statistical and clinical significance and to further assess —— OR = 6.00 (95% CI: 3.19–11.29); P < 0.0001; the magnitude of the observed effect in the context of —— RD = 0.40 (95% CI: 0.28–0.52); P < 0.0001; prognosis in a specific population. —— NNT = 3 (95% CI: 2–4). The uncertainty of estimates can be assessed by The question is how many times the probability (risk) analysing the 95% confidence interval (CI), assuming of a given event is higher in the intervention group com- there is a 2.5% probability that the real effect is below pared to the control group; the answer is then a relative

94 Marcin Kaczor et al., Critical appraisal of clinical trials in oncology — part I parameter that in the case of a negative event is called difference between probabilities in particular groups. In relative risk (RR), while in case of a positive event it the presented example this difference is 0.40, and it is called relative benefit (RB). If the frequencies of could be concluded that the probability increases by events (probabilities of their occurrence) are the same 40 percentage points in the intervention group in rela- in both groups, their ratio will be 1 — this is therefore tion to the control group. In practice, this means that for a neutral value, indicating lack of differences between every 100 patients receiving intervention a response to interventions. Values greater than 1 indicate an increase treatment will be recorded in an additional 40 patients in the probability in the intervention group, while less compared to the control treatment. In the presented than 1 — the probability being lower. In the presented example a confidence interval constructed for the example the RB value is 2, so the probability of response calculated risk difference and P-value is also provided. is two times higher after using the study drug than with Hence, the range (0.28–0.52) does not contain the value the placebo. It is also shown that the confidence interval 0, so it could be assumed that the observed differences constructed for this value ranges from 1.54 to 2.59 and between groups are statistically significant. does not contain the value 1, so it could be concluded This relationship can be also reversed, and the ques- that the differences are statistically significant, which is tion could be asked, for how many treated patients one also confirmed by the quoted P-value (< 0.0001). more event will occur. Such a parameter is called the Instead of probability, the so-called chance of occur- number needed to treat (NNT) in respect of beneficial rence of a given event in each group can be calculated. effect of treatment, and number needed to harm (NNH) Odds are defined as the ratio of the number of patients when the event is unfavourable. The ratio shows that this in whom the event was observed to the number of pa- number will be 1/0.40 (NNT = 1/RD), i.e. 2.5; because tients without such an event and hence determines how it refers to a number of patients, the result should be many times an event could be more frequently observed rounded up to the total number, so it shows that pro- than could not. In the presented example, 80 patients viding three patients with an intervention instead of responded in the intervention group and 20 did not, so control for a given time (year), one additional response the odds is 80/20 = 4 (it could be said that the chance of could be expected [16–18]. Interpreting the results for a response in this group is like 4 to 1). On the other side, dichotomous variables, in particular the values of ab- in the control group, the chance of obtaining a response solute parameters, the observation period for a given was much lower and amounted to 40/60 = 0.67. Then, endpoint should be taken into account. For example, by analogy with the relative benefit, the ratio of these NNT obtained in studies of different duration may not odds values in the analysed groups could be calculated. be directly comparable because the NNT value may vary Such a parameter is called the odds ratio (OR), and in with the observation period. the presented case it was shown that the chance of get- ting a response was six times higher in the intervention Evaluation of continuous variables group than in the control group, which was statistically significant, as shown by the P-value and confidence In clinical trials, the parameters determining the interval not containing a neutral value of 1. Although severity of disease symptoms, laboratory tests, or quality the RR/RB values are more intuitive in interpreta- of life on a certain scale are often assessed. In each of tion, the publications often present calculation results these situations, the obtained results have a continu- in the form of ORs, which are a natural result of the ous nature, i.e. they take any value expressed in a real statistical methods commonly used in the assessment number from a given range. For example, patients may of dichotomic endpoints (logistic regression, often also be asked to indicate their well-being on a scale from 0 to taking into account adjusting factors). It is worth noting 100, from the worst to the best. We also have continu- that in the case of very high frequencies of any event ous results when measuring body mass, height, blood in both compared arms, the OR calculation may have pressure, average white blood cell count, haemoglobin an advantage over RR, which in such situations will be concentration, etc. close to 1 and will not accurately illustrate the actual Such results for the studied groups of patients are difference between the groups. usually summarised by presenting the central measure In addition to the relative parameters presented — mean or median value — as a reminder, the mean above, absolute parameters can also be estimated, which (arithmetic) is the sum of the results obtained for each are considered more informative, because they addition- patient in the group divided by the number of patients ally show the real frequency of events — whether they in this group, while the median is the middle value that are extremely rare or occur in a significant percentage divides a group of patients into half (i.e. half of the of the population. Risk difference (RD) or absolute patients have a score below the median value and the risk reduction (ARR) — in the presented case it could other half, above). The set of results of a given effect be called absolute benefit increase (ABI) — is a simple expressed in a continuous variable is also characterised

95 Oncology in clinical practice 2019, Vol. 15, No. 2 by a certain variability that can be represented graphi- earlier, statistical significance can be assessed based on cally as a spread of observed values around the mean. confidence intervals, where “0” is a value indicating no The parameter indicating the magnitude of this variation differences between groups. is the standard deviation (SD) — lower values indicate The parameter most often presented in the study a small spread of results around the mean, while higher is the mean difference (MD) between the analysed values indicate a large variability and large differences groups. Assessment of the results should be carried out between the results and the mean value. carefully because the authors of the study can present As in the case of event frequency analysis, also in them in several ways. For example, when assessing the the analysis of continuous data, the overall effect is quality of life, the average score can be determined at assessed based on a sample from the general popula- the end of the observation period in both groups, and tion. Because such sampling results in different mean then the difference in mean values between them can be values distributed around the mean value in this general calculated (it is important always to make sure that no population, an additional measure of this distribution significant differences in the measurements were initially can be introduced, precisely defining the distribution of observed). It is also possible to assess the mean score the mean values from samples around the mean value change during treatment with respect to the baseline in the general population; this parameter is called the and to calculate the mean differences for such changes standard error (SE) and is equal to the standard de- — this approach is used more often because it allows viation in the sample divided by the square root of the assessment of the effect of treatment with matching in sample size. Standard error decreases with increased regard to an already existing effect. sample size, and the lower its value, the better the ap- In clinical trials the least square mean (LSM) is proximation of the true value in the general population also commonly used; this is simply an average adjusted by the given sample. for additional factors. For example, in a given group of If the study results are presented in the form of medi- people the average age can be calculated by summing ans, the range in which the observed results are found is the years of life of each person and dividing this amount also usually given. As already mentioned, the “median” by the number of persons in the sample; a simple mean is a median value, i.e. which divides a series of data into value is then obtained. However, if there are many older two equal parts. However, we can determine several such women in this group, it could lead to overestimation “aliquots” of the data set, depending on the adopted of the average — in this case, the average age can be criteria — in general they are called quantiles, and the calculated first among women, then among men, and median is a special case of such a quantile. The set can only averaging the age value for both groups the aver- be also divided into four parts — then the “aliquots” age value in the whole cohort could be obtained, with are called quartiles (it is worth noting that the median matching for gender [16–18]. is also the second quartile of the set), and in the case of dividing the set into 100 equal parts — percentiles (the Evaluation of “time-to-event” variables — survival median is the 50th percentile of the set). Sometimes the analysis authors present median values along with the so-called interquartile range (IQR), i.e. the distance between the Analysis of time-to-event data (e.g. death from any first and third quartiles. cause in OS analysis, death or tumour progression in The reasoning when assessing the statistical sig- PFS analysis) is associated with several problems. In nificance of differences between groups for continuous general, the survival analysis is performed because variables is analogous to that carried out when describ- during a sufficiently long period of observation the ing the difference in the frequency of events in two clinical events (progression/death) will occur in all or groups — in general, average values of a given param- almost all patients. In this case, estimation of a sim- eter should be determined in the analysed groups, and ple parameter like RR will be useless (probabilities then their difference and the confidence intervals or of events will be close to 100% in both groups). Ad- P-value should be calculated. In case of variables with ditionally, in the long-term observation, apart from a normal distribution (also after the appropriate data the cases marked as “with an event” or “without an transformation), Student’s t- or ANOVA test is usually event”, there will also be patients who will be lost to used to assess the differences, and in other cases, one of follow-up during the study, whose state will remain the non-parametric tests is used, e.g. U Mann-Whitney. unknown, or at statistical analysis they are still in Because in patients with higher values of a given param- observation but their future status is hard to predict. eter major changes can be expected during the test, an Finally, taking into account the different recruitment analysis of covariance (ANCOVA) is also used, which periods and the dates of inclusion of patients in the compares the mean values adjusted with the baseline study, the observation will include patients with dif- values. As in the case of the risk difference described ferent periods of study [18–22].

96 Marcin Kaczor et al., Critical appraisal of clinical trials in oncology — part I

The solution to these limitations is the survival 12 months of treatment). Interpreting the HR value analysis, considering not only the occurrence of an presented in the study, it is assumed that this ratio is event, but also the time to its occurrence, and ena- approximately constant at any time point during the bling the cutting-off (censoring) of patients lost to observation (assuming proportionality of hazards), i.e. follow-up or with unknown further fate at the time of if, for example, the HR value is 0.61, it is assumed that data analysis. It should be noted that the term “survival for patients in the intervention group the risk of death analysis” is not reserved exclusively for the assessment is approximately 39% lower than in the control group at of overall survival (i.e. time to death) but applies also each time point during the follow-up. This relationship to all time-to-event endpoints (e.g. time to response, can also be presented as the average prolongation of progression-free survival). survival time by 64% (1/0.61 = 1.64) in the interven- The simplest form of survival analysis is plotting tion group compared to control [18–22]. It should be of the Kaplan-Meier curves, based on which the noted that HR values cannot easily be translated into probabilities of survival to a given time point and the absolute differences in survival time — for example, in median survival could be assessed. Then, by means of a population with low mortality, a 30% reduction of the an appropriate statistical test (usually a log-rank test), risk of death (i.e. HR = 0.70) may be accompanied by an a comparative assessment of time differences to the increase in the average survival time of 12 months, while occurrence of an event between groups takes place. the same relative reduction in death risk (HR = 0.70) More advanced analysis, allowing us to take into ac- in a high-mortality population may be associated with count the matching factors (independent explanatory) a much lower absolute effect (e.g. three months). affecting the survival time, is carried out using regres- The simplest way to confirm the assumption about sion models, most often the Cox proportional hazard the proportionality of hazards is visual analysis of the model (it should be remembered that Kaplan-Meier course of Kaplan-Meier curves, assessing whether the curves, as well as median time-to-event are still not difference between them is approximately constant adjusted in this situation). In general, the analysis of and persists over time. Small deviations (decrease or the magnitude and direction of differences in survival is increase in time differences in the course of curves) are based on the assessment of hazard ratio (HR), median acceptable (Fig. 1). With sufficiently long observation, survival time (until an event), and survival probability especially in the final lines of treatment of advanced can- at a specific time point (e.g. 12-month survival). The cers, when events occur in all patients (with very mature HR value summarises the relative differences in survival data, when only a few patients remain in observation), between groups over the entire observation period. The after the initial period of maintaining differences in the assessment of differences in survival, consisting only of course of curves their convergence can be observed (Fig. a simple comparison of median survival time in groups, 2A). The opposite situation may occur in a population is not sufficient to depict differences in the horizon of with low risk of death and expected long-term survival, the entire study and can be quite misleading, especially when the curves can reach a flat course (plateau) due in the case of lack of hazard proportionality (this issue to very few deaths (Fig. 2D). Sometimes, at the very is discussed later in this article). beginning of the observation, the curves intersect, which The Kaplan-Meier curve is drawn based on the may happen because in the intervention group in the results of the study for individual patients, e.g. whether initial period the risk of complications may increase and when death occurred. However, it should be remem- (especially if there is a significant difference between the bered that some patients “fall out” from the study for observed procedures — e.g. surgery with chemotherapy other reasons, and information about how long they live vs. conservative treatment and chemotherapy), and the is lost. So, at the time of statistical analysis, it includes expected clinical benefit is only observed in the further patients who live, patients who have died, and those who period when the curves separate (Fig. 2B). In general, if left the study some time ago and it is not known whether the variability of the course of curves during the obser- they are still alive, which is called censored observation. vation relates to the magnitude of the effect, but not its The risk (hazard) determines the probability of an direction, it could be considered that the deviation from event occurring at a given time, assuming that the event the assumption of hazard proportionality is insignificant, has not occurred so far. The ratio of the hazard values and the presented interpretation of HR could remain. estimated for the intervention and control group at However, if there is a significant change in the direction a given time is called the hazard ratio (HR). Conceptu- of action (Fig. 2C), the calculated HR value cannot be ally, in a simplified interpretation the HR is close to the interpreted because it changes significantly over time. RR, but it should be remembered that the HR includes One solution is an attempt to perform subgroup analysis data from the entire period of observation for survival in order to detect the cause of differences in effective- in the study and censored cases, while the RR is car- ness over time, with some objections related to such an ried out at a predefined time point (e.g. deaths after analysis (see below) [18–22].

97 Oncology in clinical practice 2019, Vol. 15, No. 2

Rate of overall � 1 6 Hazard ratio (HR) 100% survival at 24 mo 0.61 (95% CI: � A: 79.7% � 0.48; 0.77) 5 (95% CI: 74.7%; 84.6%) 90% P < 0.0001 B: 69.5% 80% (95% CI: 63.8%; 75.2%) 70% 3 60% �

50%

40% � 4 30% A 20% Median OS in the B group 10% 34,7 mo B (95% CI: 32.3; 37.2) Median OS in the A group 43.4 mo (95% CI: 40.8; 45.9) 0% 0 3 6 9 12 15 18 21 24 27 30 33 36 39 42 45 48 51 54 2 � No. at Risk: Months A 298 280 267 248 231 215 199 186 170 146 129 114 96 81 68 52 32 16 B 301 275 257 233 218 205 192 172 152 133 118 97 82 65 47 17 8 5

Figure 1. The Kaplan-Meier chart is located on the plane bordered by the Y-axis describing the probability of overall survival and the X-axis, on which the time from the beginning of treatment (observation) in the study is presented. While constructing the Kaplan-Meier curve, the subsequent time intervals take into account the number of patients with the possibility of measurement at the beginning of the interval (at-risk), the number of patients with the event, and the number of patients lost to observation. The graph shows the cumulative probability at a given time. At the start of observation ( ) all patients are alive (OS = 100%). In fact, this is not the same moment in time (date of randomisation) for each of the participants, because they enter the study in different centres at different times. Then a decrease in the likelihood of overall survival in both arms over time is observed; however, it is always higher in group A (despite some variability in the size of differences between groups, the direction of the effect is consistent and it could be concluded that the deviation from the assumption of proportionality is slight — see text). Below the graph the number of patients in the observation at regular intervals should be given (at-risk) at the beginning of the given time interval ( ). If at the end of the graph they are small (< 10% of the baseline value), the inference from the curves is limited in this section. Looking at the Kaplan-Meier graph, values from both curves can be compared horizontally, looking for a difference in time when the cumulative probability of survival reaches 50% — ( ) and ( ), and they provide median survival values for groups A and B, respectively (median OS in the A group, median OS in the B group), and the median difference is 8.7 months. Differences in survival can also be assessed vertically, comparing the survival values at a given time point. In our example, the two-year survival rate (rate of overall survival at 24 months) is 79.7% in the intervention group and 69.5% in the control group ( ). This is the cumulative probability for this period; often its value is given with the confidence interval (which allows us to assess the accuracy of the estimation), and in the text a statistical evaluation of the result will usually find (P-value for differences in cumulative survival at this time point). Concluding the differences in survival throughout the observation period gives the relative hazard ratio ( ); it can be seen that at a given time point the risk of an event (death) is lower in the group with intervention, and the result is statistically significant (looking at both the confidence interval and the given P-value)

It should be remembered that HR is a relative value During the RCT, especially with a long period of that allows the assessment of the statistical significance observation, preliminary (interim) analyses carried of observed differences in survival, but, as mentioned, out by independent researchers in an unblinded but when making a therapeutic decision it is also necessary confidential manner are necessary. An independent, to assess the clinical relevance, also in relation to prog- committee with no affiliation to the study may decide to nosis in a given population. The absolute impact of the prematurely terminate the study, e.g. due to safety con- intervention compared to the control can be assessed by cerns or the spectacular effect of the new intervention analysing the differences in the medians or by comparing (in this situation the decision to stop should be assessed the probability of survival at a given time (e.g. an annual carefully, because the clear effect is often overestimated or two-year survival). in the short period of observation and the differences

98 Marcin Kaczor et al., Critical appraisal of clinical trials in oncology — part I

100% A. Kaplan-Meier curves converging at the end of the observation period 90% 80% � 70% � 1 60% Intervention Control 50% 40% 30% � � 2 Survival probability (%) 20% 10% 0% 0 2 4 6 8 10 12 14 16 18 20 22 Follow-up period (months)

Figure 2 A. Kaplan-Meier curves converging at the end of the observation period. From the very beginning of the observation, there are differences in survival ( ). The curves clearly “separate”, but at the end of the observation period ( ) the cumulative probability of death is practically the same in both arms of the study. Such a situation may occur, for example, at the terminal stages of cancer, where ultimately, regardless of the treatment used, an event (death) will occur in almost all patients. This is a permissible deviation from the assumption about the proportionality of hazards. Convergence of curves in the final observation period may also be caused by a high percentage of censored observations (i.e. a small number of at-risk patients), as a consequence of which the Kaplan-Meier estimation in the “tail” of the curve is impaired

100% B. Kaplan-Meier curves intersecting in the initial period of observation 90% �1 80% 70% 60% Intervention 50% Control 40% 2� 30% � Survival probability (%) 20% 10% 0% 0 2 4 6 8 10 12 14 16 18 20 22 Follow-up period (months)

Figure 2 B. Early intersection of the Kaplan-Meier curves. In the initial period of observation, the curves intersect, and for a short period the survival is lower in the group with intervention compared to the control — e.g. when the intervention is associated with higher risk of complications ( ), but then there are clear differences in survival in further observation ( ). In this situation, it is also a permissible deviation from the assumption about the proportionality of hazards

between interventions become less visible in the longer Finding significant OS differences between interven- period of observation) [14]. In statistical protocols of tions according to the assumed statistical power may oncology studies further interim analyses are also prede- enable patients after disease progression to move from fined (when the statistical power calculation depends on a control to an intervention group (cross-over, treat- the occurrence of a given number of events). Because of ment switching), which acts in a conservative direction, repeatedly testing the hypothesis, the risk of accidentally overestimating the effectiveness of control intervention observed “statistically significant” results increases (the and reducing the estimated effect of the study drug. greater, the more pre-planned interim analyses, e.g. ac- Obviously, this only matters for OS assessment, and cording to the O’Brien-Fleming criteria); it is worth noting PFS assessment itself is unaffected. In this case, it is that significant results will not refer to P-values of < 0.05, possible to use appropriate methods of correction of but assume a much lower threshold, e.g. < 0.001. the cross-over impact on OS, the simplest of which is

99 Oncology in clinical practice 2019, Vol. 15, No. 2

100% C. Kaplan-Meier curves intersecting in the later period of observation 90% � 1 80% 70% Intervention 60% Control 50% � � 2 40% 30% Survival probability (%) 20% 10% 0% 0 2 4 6 8 10 12 14 16 18 20 22 Follow-up period (months)

Figure 2 C. Kaplan-Meier curves intersecting in the later period of observation. The initial advantage of intervention ( ) disappears unexpectedly during the observation ( ), and the survival remains higher in the comparator arm until the end of the observation period. In this case, the proportionality of hazards criterion is not met, and no differences can be indicated between the groups. Probably there was an unknown confounder in the study that reversed the inference, and a detailed subgroups analysis is needed to identify it

100% D. Kaplan-Meier curves for an observation period that is too short 90% 80% 70% 60% 50% Intervention Control 40% 30% Survival probability (%) 20% 10% 0% 0 2 4 6 8 10 12 14 16 18 20 22 Follow-up period (months)

Figure 2 D. Kaplan-Meier curves for an observation period that was too short. Mortality persists at a relatively low level, and it is difficult to conclude about the fate of patients and differences between groups (assessment in the early stages of cancer, with expected long-term follow-up). Despite the initial differences with a longer observation period, when more observed events accumulate, the course of these curves may approach any of the previously described situations the censoring of observations in patients changing the modify the central measure of the estimate. However, treatment [24, 25]. taking into account the characteristics not related to the prognosis, although even including stratification factors Adjusted results and subgroup analysis (e.g. geographical location), does not significantly affect the results. Post-hoc selection of matching factors (not Analysis of results with adjustment to baseline defined in the statistical plan) may raise the suspicion characteristics within logistic regression (when OR of purposeful data selection to achieve the desired ef- values are presented as adjusted or multivariate) or fect; in this case it should always be expected to display Cox proportional hazard model (in the case of survival adjusted and not adjusted results [23]. analysis) can be presented as primary or sensitivity Subgroup analysis should also be predefined in the analysis. It is worth noting that the adjustment should statistical plan. Considering the diversity of the general primarily include prognostic factors and — unless it study population, it allows the assessment of whether usually affects the accuracy of the estimate — can the general results refer to all patients or whether there

100 Marcin Kaczor et al., Critical appraisal of clinical trials in oncology — part I are differences in the effectiveness of the interven- Summary tion. If it is the case, it should be remembered that reliable evaluation of statistical differences leads to Translating the results of clinical trials into clinical loss of statistical power. On the other hand, multiple daily practice is the established method of EBM. For this defined subgroups and repeated testing increase the purpose, however, many elements should be assessed, risk of completely random occurrence of statistically which together provide evidence of the reliability of the significant results. Above all, statistical significance or study and significance of its results (Fig. 3). It is not easy, lack thereof in one of the subgroups is not sufficient especially since most of the publications are written in to conclude real differences in the effectiveness of English, and the authors often assume in advance that intervention. Usually a coherent effect is observed in the recipient is fluent in terms of statistics and detailed individual subgroups, although with some variability of explanations are unnecessary. In addition to assessment the central measure, and in smaller groups the confi- of the methodology and the reliability of the clinical dence intervals become wider and, in some cases, may study, it must be ensured that the population being exceed the value of 1 (loss of statistical significance). evaluated is representative, i.e. it has characteristics In this situation attention should be paid to the sig- similar to those for which a therapeutic decision is to nificance of the interaction test (statistical analysis of be made, and if there are discrepancies (e.g. different whether the impact of interventions on an observed age of patients or presence of comorbidities), what is result depends on other factors) to assess whether there their meaning. Then, if the comparator in the study is a difference in the effectiveness of the intervention is not widely used or not available but has a similar in a given subgroup (remembering, however, the pos- mechanism of action to the current treatment stand- sibility of obtaining false results in repeated testing). ard, it should be assessed whether there is evidence of Subgroup analysis may also be helpful in seeking a nar- similar effectiveness, which would allow transferral of rowing target population with no significant result in the inference from the study to clinical practice in this the general cohort. However, it should be remembered aspect. The sample size of the study is significant — if that subgroup analysis is more exploratory and serves it was small and it was not due to the low prevalence to create further hypotheses rather than make final of the disease, then it should be assessed whether the conclusions [23]. study had statistical power to indicate the differences.

Methodology Primary results Additional analysis Others • Baseline characteristics • Statistical significance, hazard • Intention-to-treat, per protocol • Subsequent treatment • Allocation, intervention, con- ratio for progression-free sur- analysis • The authors’ conclusions trol vival, overall survival — P-value • Subgroup analysis — unambiguous or conservative • Comparator and local clinical and confidence interval — convergence of inference • Consistency of results with practice • Course of Kaplan-Meier curves with the ITT population other tests (external coherence) • Research hypothesis • Median OS and PFS • Statistical significance in sub- • The possibility of transferring • Randomization method, blind, • Other end points populations (interaction test) conclusions to clinical practice open label, unmasked — convergence of inference • Interim analysis, cross-over (external credibility) • Drop-out, discontinuation with primary endpoints • Internal credibility — internal consistency (time to treatment discontinuation, objective response, complete response) • Quality of life, safety, adverse event

Figure 3. Aspects to which attention should be paid, making a critical evaluation of the methodology and results of a clinical trial in oncology

101 Oncology in clinical practice 2019, Vol. 15, No. 2

It is also necessary to pay attention to possible differenc- extremely important. In oncology, treatment with higher es in baseline characteristics between groups — if they efficacy is often associated with increased toxicity; this were significant, they may indicate a selection bias. An situation may be acceptable with clear clinical profit, incorrect randomisation method can cause imbalanced such as prolonged survival. distribution of confounders. In the absence of blinding, In conclusion, the evaluation of a clinical trial con- the assessment of differences in subjective endpoints is sists of many elements discussed briefly in this paper. subject to the limitations. In the case of significant dif- The authors hope, however, that they have addressed the ferences in the lost patient rate, it is important to know most important aspects of the evaluation of clinical trial whether this could be related to the treatment applied. results and the terminology used, and that the article It is also worth checking the type of the research hypo managed to show the complexity of the interpretation thesis tested. In oncology the primary endpoint will be process. In the second part of the work, examples of the survival analysis — PFS and OS. If consistent, statis- clinical trials will be presented along with an assessment tically significant, and clinically relevant differences are of their credibility and impact on clinical practice. observed in favour of the intervention, there are strong premises about the superiority of the evaluated therapy. If, however, the significance of the results was observed References only for PFS, it should be decided whether a further (final) assessment of the OS is planned, in which, with 1. Govani SM, Higgins PD. How to read a clinical trial paper: a lesson in basic trial statistics. Gastroenterol Hepatol (N Y). 2012; 8(4): more matured data, the result could reach statistical 241–248. PubMed PMID: 22723755; PubMed Central PMCID. significance. For some cancers, especially when assessing PMCPMC3380258. 2. Bothwell LE, Greene JA, Podolsky SH, et al. Assessing the Gold their early stages, it may be difficult to show differences Standard — Lessons from the History of RCTs. N Engl J Med. 2016; in survival due to the expected follow-up, sometimes 374(22): 2175–2181, doi: 10.1056/NEJMms1604593, indexed in Pubmed: 27248626. even decades. Such a long observation is an additional 3. Pocock SJ, Clayton TC, Stone GW. Design of Major Randomized Tri- challenge, because during this time patients may be als: Part 3 of a 4-Part Series on Statistics for Clinical Trials. J Am Coll Cardiol. 2015; 66(24): 2757–2766, doi: 10.1016/j.jacc.2015.10.036, subjected, for example, to many different lines of further indexed in Pubmed: 26700838. treatment, and the evaluation of ultimate survival dif- 4. Wood L, Egger M, Gluud LL, et al. Empirical evidence of bias in treat- ferences is limited (in this situation, such endpoints as ment effect estimates in controlled trials with different interventions and outcomes: meta-epidemiological study. BMJ. 2008; 336(7644): DFS or pathological response are of higher importance). 601–605, doi: 10.1136/bmj.39465.451748.AD, indexed in Pubmed: It is important whether the possibility of changing the 18316340. 5. Poolman RW, Struijs PAA, Krips R, et al. Reporting of outcomes in treatment after the progression (cross-over) was al- orthopaedic randomized trials: does blinding of outcome assessors lowed, which could lead to OS overestimation in the matter? J Bone Joint Surg Am. 2007; 89(3): 550–558, doi: 10.2106/JB- JS.F.00683, indexed in Pubmed: 17332104. control group. When HR for both OS and PFS didn't 6. Jadad A, Moore R, Carroll D, et al. Assessing the quality of reports of reach the significance level, the question should be asked randomized clinical trials: Is blinding necessary? Controlled Clinical Trials. 1996; 17(1): 1–12, doi: 10.1016/0197-2456(95)00134-4. whether this is not due to the lack of study statistical 7. Higgins JPT GSe. Cochrane Handbook for Systematic Reviews of power, immaturity of published results (interim analy- Interventions Version 5.2.0 [updated June 2017]. The Cochrane Collaboration, 2017. Available from: http://handbook.cochrane.org. sis), or high lost-patient rate. If none of these factors 8. Pocock SJ, McMurray JJV, Collier TJ. Making Sense of Statistics in is relevant, there are probably no differences between Clinical Trial Reports: Part 1 of a 4-Part Series on Statistics for Clinical Trials. J Am Coll Cardiol. 2015; 66(22): 2536–2549, doi: 10.1016/j. the interventions. Assessing the consistency of results jacc.2015.10.014, indexed in Pubmed: 26653629. with other publications for a similar population may 9. Health Technology Assessment Guidelines. The Agency for Health be very helpful. If PFS/OS medians differ from similar Technology Assessment and Tariff System. 2016. Version 3.0:[Avail- able from: http://www.aotm.gov.pl/www/wp-content/uploads/wyty- studies, the characteristics of the population should be czne_hta/2016/20161104_HTA_Guidelines_AOTMiT.pdf. carefully analysed. A valuable source of information 10. Li L, Pan Z. Progression-Free Survival and Time to Progression as Real Surrogate End Points for Overall Survival in Advanced Breast Cancer: is also subgroup analysis. Sometimes the result for the A Meta-Analysis of 37 Trials. Clin Breast Cancer. 2018; 18(1): 63–70, doi: subpopulation becomes statistically significant, despite 10.1016/j.clbc.2017.07.015, indexed in Pubmed: 28818493. 11. Panasiuk AWR, Budasz-Świderska M, Kaczor M. Cancer immuno- the lack of significance in the total population — this therapy in second-line treatment of non-small cell lung cancer — is may be a premise of higher treatment effectiveness only there a need to change the approach to the assessment of clinical in a specific subgroup, but on the other hand, one should benefits? Journal of Helath Policy & Outcomes Research. 2017; 2(2): 65–77, doi: 10.7365/JHPOR.2018.1.9. bear in mind the exploratory nature of such analysis. If 12. Wilson MK, Collyar D, Chingos DT, et al. Outcomes and endpoints in a similar trend in the results is observed depending on cancer trials: bridging the divide. Lancet Oncol. 2015; 16(1): e43–e52, doi: 10.1016/S1470-2045(14)70380-8, indexed in Pubmed: 25638556. the presence or absence of a given criterion, but one 13. Wilson MK, Karakasis K, Oza AM. Outcomes and endpoints in trials of of the subgroups lacks relevance, it is worth checking cancer treatment: the past, present, and future. Lancet Oncol. 2015; 16(1): e32–e42, doi: 10.1016/S1470-2045(14)70375-4, indexed in whether the size of this subgroup is not too small, as Pubmed: 25638553. well as checking the result of the interaction test. If any 14. Pocock SJ, Clayton TC, Stone GW. Challenging Issues in Clinical Trial therapeutic options are available in the analysed indica- Design: Part 4 of a 4-Part Series on Statistics for Clinical Trials. J Am Coll Cardiol. 2015; 66(25): 2886–2898, doi: 10.1016/j.jacc.2015.10.051, tion, the safety issues of the therapy being evaluated are indexed in Pubmed: 26718676.

102 Marcin Kaczor et al., Critical appraisal of clinical trials in oncology — part I

15. Mauri L, D’Agostino RB. Challenges in the Design and Interpretation 21. Fendler WCJ, Młynarski W. Methods of survival analysis applied in of Noninferiority Trials. N Engl J Med. 2017; 377(14): 1357–1367, doi: oncology — assumptions, methods and common pitfalls. Onkologia 10.1056/NEJMra1510063, indexed in Pubmed: 28976859. w Praktyce Klinicznej. 2011; 7(2): 89–101. 16. West CP, Dupras DM. 5 ways statistics can fool you — tips for prac- 22. Barraclough H, Simms L, Govindan R. Biostatistics primer: what a clini- ticing clinicians. Vaccine. 2013; 31(12): 1550–1552, doi: 10.1016/j. cian ought to know: hazard ratios. J Thorac Oncol. 2011; 6(6): 978–982, vaccine.2012.11.086, indexed in Pubmed: 23246309. doi: 10.1097/JTO.0b013e31821b10ab, indexed in Pubmed: 21623277. 17. Elwood JM. Interpreting clinical trial results: seven steps to under- 23. Pocock SJ, McMurray JJV, Collier TJ. Statistical Controversies in Re- standing. Can Med Assoc J. 1980; 123(5): 343–345, indexed in porting of Clinical Trials: Part 2 of a 4-Part Series on Statistics for Clinical Pubmed: 7260774. Trials. J Am Coll Cardiol. 2015; 66(23): 2648–2662, doi: 10.1016/j. 18. Case LD, Kimmick G, Paskett ED, et al. Interpreting measures of treat- jacc.2015.10.023, indexed in Pubmed: 26670066. ment effect in cancer clinical trials. Oncologist. 2002; 7(3): 181–187, 24. Panasiuk AHM, Pawlik D, Prząda-Machno P, et al. Approach to indexed in Pubmed: 12065789. uncertainty in health technology assessment in a Central and East- 19. Pocock SJ, Travison TG, Wruck LM. How to interpret figures ern European country: appraisal of cancer drugs by a Polish HTA in reports of clinical trials. BMJ. 2008; 336(7654): 1166–1169, agency in presence of high crossover rates in clinical trials. Journal doi: 10.1136/bmj.39561.548924.94, indexed in Pubmed: of Helath Policy & Outcomes Research. 2016; 10(2): 17–34, doi: 18497415. 10.7365/JHPOR.2016.2.2. 20. Goel MK, Khanna P, Kishore J. Understanding survival analysis: 25. Haslam A, Prasad V. When is crossover desirable in cancer drug trials Kaplan-Meier estimate. Int J Ayurveda Res. 2010; 1(4): 274–278, doi: and when is it problematic? Ann Oncol. 2018; 29(5): 1079–1081, doi: 10.4103/0974-7788.76794, indexed in Pubmed: 21455458. 10.1093/annonc/mdy116, indexed in Pubmed: 29648572.

103 REVIEW ARTICLE

Monika Konopka-Filippow1, 2, Ewa Sierko1, 2, Marek Z. Wojtukiewicz1 1Oncology Clinic, Medical University of Bialystok, Poland 2Department of Radiotherapy, Białystok Centre of Oncology, Poland

Benefits and difficulties during brain radiotherapy planning with hippocampus sparing

Address for correspondence: ABSTRACT Dr hab. n. med. Ewa Sierko Radiotherapy (RT) is frequently used in the treatment of primary and secondary brain tumours, as well as in Klinika Onkologii prophylactic cranial irradiation (PCI). The hippocampus plays a key function in the process of remembering, Uniwersytet Medyczny w Białymstoku relaying information from short-term to long-term memory as consolidation, and spatial orientation. Sparing the ul. Ogrodowa 12, 15–027 Białystok hippocampus during brain radiotherapy aims to prevent hippocampal-dependent cognitive function deterioration. Phone: +48 85 66 46 827 This procedure requires a good knowledge of the brain’s radiological anatomy and use of modern radiotherapy e-mail: [email protected] techniques. Oncology in Clinical Practice This article presents the validity of hippocampus sparing during brain radiotherapy, the potential benefits of using 2019, Vol. 15, No. 2, 104–110 this procedure, available clinical premises regarding patient qualification, and technical difficulties in the brain’s DOI: 10.5603/OCP.2019.0019 RT planning with hippocampus avoidance. Translation: dr n. med. Dariusz Stencel Key words: hippocampus sparing, brain radiotherapy, cognitive function Copyright © 2019 Via Medica ISSN 2450–1654 Oncol Clin Pract 2019; 15, 2: 104–110

Hippocampus damaging factors such as ischaemia, stress, and ionising radiation [4]. An analysis of brain magnetic resonance The hippocampus was first described by Arantius in imaging (MRI) of 58 patients with nasopharyngeal car- 1587 as a grey matter brain zone resembling a creature cinoma performed three and six months after completed from Greek mythology drawing the chariot of the god of brain RT revealed atrophy of the hippocampal area [5]. the sea, Poseidon. It consists of the head with the appear- Depopulation of NSCs by apoptosis, which occurs after ance of a horse’s head and a curved body like a sea wave. the activation of the damaging factor, appears already Hence the name of this organ is derived from hippo after 12 hours and leads to the manifestation of deficits (horse) and campi (turn) [1]. It is difficult to depict the in cognitive functions for which the hippocampus cor- shape of the hippocampus in a two-dimensional plane responds, and in particular to disorders in memorising due to its long, curved body (Fig. 1A, 1B). and reproduction of information from working memory Anatomically, the hippocampus is an even organ [6–8]. It was shown that irradiation of the hippocampus located in telencephalon region, in the temporal lobes area with doses close to 30 Gy and higher, given in of the cerebral cortex of the right and left hemispheres conventional fractionation, causes a decrease in NSCs of the brain. Within the hippocampus, in the vicinity of proliferation by 93–96% after 48 hours [7]. Deficits in the dentate gyrus, there is a cluster of neural stem cells cognitive functions appear about two months after the (NSCs) grouped in two niches: the subventricular zone activation of the damaging factor, and the peak of inten- (SVZ) and the subgranular zone (SGZ) [2, 3] (Fig. 2). sity falls around the fourth month [9–12]. Importantly, These NSCs are responsible for the key functions of this the consequences of NSC apoptosis are irreversible and structure. It is worth noting that they are very sensitive to usually progressive over time.

104 Monika Konopka-Filippow et al., Benefits and difficulties during brain radiotherapy planning with hippocampus sparing

A B

Figure 1. The anatomical shape of the hippocampi (A) and a sketch of the creation from Greek mythology (B) (author: M. Filippow)

Hippocampi, dentate gyrus

Figure 2. Localisation of neural stem cell (NSC) compartment in the region of dentate gyrus of hippocampi (author: M. Filippow)

The rationale of hippocampus tive functions based on a previously unknown mechanism, contouring during brain CT planning presumably as a paraneoplastic effect [16]. The progress in oncological treatment is reflected both in the quality The majority of patients with malignant neoplasms of therapy and in its effectiveness, which translates into within the brain manifest cognitive dysfunction even be- longer survival time. Recently, attention has been paid to fore the implementation of any causative treatment. They the patient’s quality of life after the use of anti-cancer treat- may result from the presence of malignant disease within ment, and attempts are being made to reduce the negative the brain, its progression, the use of supportive treatment effects of the therapy. Irradiation of the brain, particularly (e.g. opioids, steroids), comorbidities, or advanced age the area of the hippocampus, may lead to further cognitive [13, 14]. Sudden/acute deterioration of cognitive functions deficits, which as a result significantly affects patients’ quality may appear just after the brain’s RT due to the presence of life. Described cognitive functions relate to the thought of brain metastases accompanied by extensive oedema processes used to process information coming from the zones around changes [15]. In turn, it was demonstrated outside world into the mind and contain basic aspects such that patients with small-cell lung cancer (SCLC) without as memory, attention, and association and complex ones, metastases in the brain may show deterioration of cogni- which include thinking and imagination [17].

105 Oncology in clinical practice 2019, Vol. 15, No. 2

The most frequently described deficits of cognitive cognitive deficits within the first four months of treat- functions after brain CT are losses in short-term memory ment versus the state before treatment, as compared to and less frequently in delayed memory, and problems RT without the cover of this structure [10, 12]. Another with information recall and learning [18, 19]. There study, using the HVLT test, showed a smaller loss of cog- are also described verbal memory disorders, necessary nitive functions in the field of learning and short-term to understand reading text, as well as inhibition of the memory in patients with 1–3 BM, who underwent only higher cognitive processes necessary to behave in new stereotactic brain radiotherapy (SRT), in comparison to and difficult situations [20, 21]. It is worth noting that patients with WBRT [12]. In turn, the AVLT auditory any deterioration of cognitive functions in oncological test revealed a decrease in verbal memory 6–8 weeks patients significantly affects the quality of life after the after completion of WBRT in patients with BM, in com- completion of anticancer treatment and can contribute parison to baseline status [12]. In the phase III RTOG to the deepening of lowered mood, and even the occur- 0214 study conducted in a group of patients with stage rence of depressive episodes [22, 23]. III non-small cell lung cancer (NSCLC) subject to PCI, It should be noted that the hippocampus is a very a marked decrease in cognitive function was seen after rare location of cancer metastases [24, 25]. Researchers three months of brain RT evaluated with the MMSE test from the University of Wisconsin documented that only [27]. Research is ongoing to find an objective biomarker 3.3% of intracranial metastatic lesions were located up used alone or in fusion with MRI to detect early damage to 5 mm from the hippocampus, and over 86% were in the hippocampal region [5]. located at least 15 mm from this structure [26]. In another retrospective study analysing the location of 697 intracranial metastases, only 2.2% of lesions were Hippocampus contouring techniques found in the direct location of the hippocampus, and in patients with oligometastatic disease (one to three Manual contouring of the hippocampus brain metastases) the rate of hippocampal metastases was below 1% [25]. This is the most popular technique in the daily practice of a radiotherapist; however, it requires a good knowledge of anatomy in the planned area. Correct Clinical assessment of impaired contouring of the hippocampus is the most important cognitive functions process during the preparation of an irradiation plan with a procedure for the protection of this structure. In Evaluation of the cognitive deficit after brain CT any case of contouring of the hippocampus, it is neces- is methodically difficult and ambiguous. Until now, sary to fuse locational computed tomography images researchers have used subjective methods in the form with a current T2-weighted MRI brain examination of psychological tests, e.g. MMSE (mini-mental stage at a scan density of min. every 1.5 mm [1]. The atlas examination), HVLT (Hopkins verbal learning test), created by the RTOG group is an assistance in the pro- and AVLT (auditory verbal learning test) [10, 12, 27]. cess of contouring the hippocampus during brain RT An example is the RTOG 0914 study conducted in planning [29]. The necessity of training in contouring a group of 445 patients with brain metastases (BM), is emphasised, which allows practice of the technique who underwent whole brain radiotherapy (WBRT), of contouring of the most important area within the which proved that both hypofractionated (30 Gy/10 frac- hippocampus: the dentate gyrus [30]. It was shown that tions) and conventional (40 Gy/20 fractions) RT lead without “contouring learning” of this structure there are to a significant reduction in cognitive function, and the large discrepancies in the exact location of this region results of the MMSE test revealed a marked deteriora- between radiotherapists, and hence inconsequence in tion in cognitive functions in both groups two and three planning the brain RT and suboptimal results of treat- months after completion of RT [9]. Another multicentre ment. The biggest discrepancies during manual contour- phase III trial based on 401 patients with BM treated ing of the hippocampus occur in the area of the horn of with WBRT (30 Gy/10 fractions) revealed a significant the anterior lateral ventricle, while the smallest are in decrease in cognitive functions assessed on the basis of the area of the brain stem [31]. the verbal fluency test (COWA, controlled oral word association) four months after RT, and then their Automatic contouring of the hippocampus improvement 15 months after completion of RT [28]. Preliminary results of the phase II RTOG 0933 study, Automatic methods of brain segmentation are usu- using Hopkins’ verbal learning test (HVLT) showed ally based on MRI images obtained on 1.5 T, 3 T, and that the use of hippocampal sparing in patients with even 7 T cameras, for better imaging and contrast of BM during WBRT resulted in lower intensity of early individual structures [32]. The first group includes pro-

106 Monika Konopka-Filippow et al., Benefits and difficulties during brain radiotherapy planning with hippocampus sparing grams based on atlases, i.e. Atlas-Based Segmentation, sparing may be considered in patients with radical ra- Multi-Atlas-Based Methods [33, 34]. The second group dio-chemotherapy or in patients undergoing palliative of methods are modern computer programs analysing chemotherapy, who have achieved a clinical response voxels, such as the Auto-Context Model (ACM) [35]. within the chest after this treatment with no disease The automatic program for contouring the hippocampus progression [46, 47]. The most frequently recommended during RT planning has advantages and disadvantag- PCI regimen is whole brain irradiation to a total dose es. The advantages undoubtedly include the minimum of 25 Gy in 10 fractions of 2.5 Gy [48]. It has also been contribution of the “radiotherapist’s hand” and the ac- demonstrated that PCI may contribute to the impair- curacy of contouring on MRI images. However, it should ment of cognitive functions as a result of post-radiation be remembered that such contouring should always depopulation of NSCs within the hippocampus [49, 50]. be verified and approved by a radiotherapist, which is Research results indicate that the use of a cover of both connected with the requirement of his/her knowledge hippocampi could reduce or even prevent cognitive not only of the radiological anatomy of the brain, but complications after PCI [49, 51]. also of the influence of various pathologies on its mor- Patients with secondary brain tumours (BM) consti- phology [36, 37]. Automatic atlases for contouring of tute the most controversial group in terms of application structures are becoming more common; however, most of hippocampal protection procedure, due to the short- radiotherapy departments still do not have such software est expected overall survival, and therefore a relatively that is optimally integrated with the RT planning system. short time of expected potential benefit. On the other Importantly, this is also associated with the additional hand, the majority of research on the hippocampus pro- costs of purchasing such software. tection procedure during cerebral irradiation concerns It is worth noting that the brain often has various pa- patients with BM. Unfortunately, the current results of thologies that disturb the anatomy of its structures, such the study do not allow us to clearly define the eligibi as microcalcifications, the number of which increases lity criteria for the hippocampus protection procedure with age, states after strokes, seizures, after brain infec- during brain RT in the above group of patients [52–54]. tions or Alzheimer’s disease [38–40]. In such cases, the There is a need to select specific criteria for the automatically contoured structures of the hippocampus qualification of patients for hippocampus protection may turn out to be incorrect. procedure during brain RT and to develop practical recommendations during this procedure within brain RT.

Clinical situations in which hippocampal protection should be “Protective” doses of ionising radiation considered during RT planning in the area of the hippocampus

In clinical practice, an appropriate group of patients In current research it has been shown that even should be selected in which there is a need for avoiding small doses of ionising radiation cause radiation-induced of the hippocampus during brain RT planning. The inflammation of the areas of neurogenesis within the hip- above procedure is a technical challenge regarding pocampus [6, 7]. In the phase II RTOG 0933 study a dose of application of highly specialised RT techniques. It can ionising radiation was initially proposed that should not be be considered for patients with primary cerebral tu- exceeded in the hippocampal area during PCI and WBRT mours, where the use of intensity-modulated radiation planning (Table 1) [54]. The above recommendations may therapy (IMRT) allows us to reduce the dose within the prevent deterioration in terms of memory, especially short- hippocampus by 56.8% in relation to the classical 3D term memory and verbal memory, or the reproduction of technique, i.e. from 36.6 Gy to approx. 15 Gy in the case freshly-stored information after application of RT to the of irradiation of part of the brain in the aforementioned cerebral region [58]. The proposed doses refer to conven- group of patients [41]. In certain clinical situations, i.e. tional radiotherapy in which the fractional dose oscillates in the presence of an extensive oedema or central tu- between 2 and 3 Gy. The problem arises in the case of mour location, especially around the brain stem, many hypofractionated RT and in particular stereotactic RT, authors suggest shielding only one hippocampus — on although there are newer reports of “protective” doses in the opposite side of the tumour site [20, 42, 43]. It is the hippocampus region in such cases [59, 60]. worth adding that in children primary brain tumours are diagnosed much more frequently than in adults, and the procedure of hippocampal protection in these cases Summary has a special clinical value during RT planning [44, 45]. Patients with SCLC represent another population. Brain radiotherapy is a recognised method of on- Elective brain radiotherapy — PCI with hippocampal cological treatment in patients with primary and meta-

107 Oncology in clinical practice 2019, Vol. 15, No. 2

Table 1. Ionising radiation doses (Gray — Gy) recommended for hippocampal sparing brain irradiation (WBRT, PCI) procedure, according to the RTOG 0933 study compared to other authors. The ranges of standards are given in brackets

RTOG 0933 Gondi et al. Nevelsky et al. Wang et al. Krayenbuehl Zhao et al. [54, 58] [55] [53] [37] et al. [56] [57] PTV D98% (Gy) ≥ 25 25.7 25.8 ≥ 25* (25.4–25.9) (25.0–27.1) PTV V95% (%) 96.9 96.9 96.4 (96.1–97.5) (96.0–97.5) (95.2–97.8) PTV D2 (Gy) ≤ 37.5 37.2 35.1 33.5 (36.9–37.6) (34.8–35.6) (32.8–34.6) PTV V30Gy (%) = 90 92.1 92 (92.0–92.0) < 23.75 (90.5–93.2) Hippocampus D100% (Gy) ≤ 9 8.4 (7.7–8.9) 9.3 (8.3–10.0) 8.1 (7.8–8.5) ≤ 9

Hippocampus Dmean (Gy) 7.3 (7.2–7.6) 7.3 (6.0–7.9)

Hippocampus Dmax (Gy) ≤ 16 15.3 14.3 16 (14.6–16.9) 14.1 ≤ 16 (14.3–15.9) (13.5–15.4) (12.0–15.3)

Lens Dmax (Gy) 3.8 (3.1–4.3) 5.8 (4.5–6.5) 4.6 (3.7–5.6) Crossing of the optic nerves ≤ 37.5 36.2 34.7 32.9

(optic chiasm) Dmax (Gy) (33.9–37.2) (33.1–36.8) (31.7–35.1)

Optic nerve Dmax (Gy) ≤ 37.5 32.5 32.0 33.1 (28.3–35.7) (23.7–36.1) (32.5–33.8)

*PCI-PTV — planning target volume with 3 mm margin excluding the hippocampal region (hippocampus expanded by 5 mm); PCI — prophylactic cranial irradiation; WBRT — whole brain radiation therapy; PTV — planning target volume; Dmax — maximal point dose; Dmin — minimal point dose; Dmean — mean point dose; D100% — dose to 100% of the volume; D98% — dose to 98% of the volume; V95% — volume covered by 95% of the prescribed dose; D2% — dose to 2% of the volume; V30Gy — volume covered by 30% of the prescribed dose static cerebral lesions, although cognitive impairment pocampal avoidance in cranial radiotherapy. Radiat Oncol. 2014; 9: 139, doi: 10.1186/1748-717X-9-139, indexed in Pubmed: 24935286. appearing after this treatment may contribute to the 3. Kier EL, Kim JH, Fulbright RK, et al. Embryology of the human fetal deterioration of patients’ quality of life. Occurrence of hippocampus: MR imaging, anatomy, and histology. AJNR Am J Neuroradiol. 1997; 18(3): 525–532, indexed in Pubmed: 9090416. these complications is associated with post-radiation 4. Barazzuol L, Rickett N, Ju L, et al. Low levels of endogenous or damage to the hippocampus, a structure particularly X-ray-induced DNA double-strand breaks activate apoptosis in sensitive to ionising radiation, and especially the NSCs adult neural stem cells. J Cell Sci. 2015; 128(19): 3597–3606, doi: 10.1242/jcs.171223, indexed in Pubmed: 26303202. within it. Cognitive deficits mainly concern problems 5. Lv X, He H, Yang Y, et al. Radiation-induced hippocampal atrophy with memorising and reproducing information, and in patients with nasopharyngeal carcinoma early after radiotherapy: a longitudinal MR-based hippocampal subfield analysis. Brain Imaging problems with short-term, delayed, and verbal memory. Behav. 2018 [Epub ahead of print], doi: 10.1007/s11682-018-9931-z, Application of the hippocampus protection procedure indexed in Pubmed: 30054872. 6. Monje ML, Mizumatsu S, Fike JR, et al. Irradiation induces neural during brain RT may significantly reduce or even pre- precursor-cell dysfunction. Nat Med. 2002; 8(9): 955–962, doi: vent the above complications. 10.1038/nm749, indexed in Pubmed: 12161748. Modern RT techniques provide the ability to 7. Mizumatsu S, Monje ML, Morhardt DR, et al. Extreme sensitivity of adult neurogenesis to low doses of X-irradiation. Cancer Res. 2003; protect the hippocampus during brain RT, although 63(14): 4021–4027, indexed in Pubmed: 12874001. there is a need for further research to establish clinical 8. Tofilon PJ, Fike JR. The radioresponse of the central nervous system: a dynamic process. Radiat Res. 2000; 153(4): 357–370, indexed in indications, qualify the appropriate group of patients, Pubmed: 10798963. and develop technical recommendations for the imple- 9. Murray KJ, Scott C, Zachariah B, et al. Importance of the mini-mental status examination in the treatment of patients with brain metastases: mentation of this procedure, which could translate into a report from the Radiation Therapy Oncology Group protocol 91-04. clinical benefits and improve the quality of radiotherapy Int J Radiat Oncol Biol Phys. 2000; 48(1): 59–64, indexed in Pubmed: (quality assurance). 10924972. 10. Chang EL, Wefel JS, Hess KR, et al. Neurocognition in patients with brain metastases treated with radiosurgery or radiosurgery plus whole- -brain irradiation: a randomised controlled trial. Lancet Oncol. 2009; 10(11): 1037–1044, doi: 10.1016/S1470-2045(09)70263-3, indexed References in Pubmed: 19801201. 11. Li J, Bentzen SM, Renschler M, et al. Regression after whole-brain 1. Chera BS, Amdur RJ, Patel P, et al. A radiation oncologist’s guide to radiation therapy for brain metastases correlates with survival and im- contouring the hippocampus. Am J Clin Oncol. 2009; 32(1): 20–22, doi: proved neurocognitive function. J Clin Oncol. 2007; 25(10): 1260–1266, 10.1097/COC.0b013e318178e4e8, indexed in Pubmed: 19194118. doi: 10.1200/JCO.2006.09.2536, indexed in Pubmed: 17401015. 2. Kazda T, Jancalek R, Pospisil P, et al. Why and how to spare the 12. Welzel G, Fleckenstein K, Schaefer J, et al. Memory function before hippocampus during brain radiotherapy: the developing role of hip- and after whole brain radiotherapy in patients with and without brain

108 Monika Konopka-Filippow et al., Benefits and difficulties during brain radiotherapy planning with hippocampus sparing

metastases. Int J Radiat Oncol Biol Phys. 2008; 72(5): 1311–1318, 32. Kim M, Wu G, Li W, et al. Automatic hippocampus segmentation of doi: 10.1016/j.ijrobp.2008.03.009, indexed in Pubmed: 18448270. 7.0 Tesla MR images by combining multiple atlases and auto-context 13. Scoccianti S, Ricardi U. Treatment of brain metastases: review of models. Neuroimage. 2013; 83: 335–345, doi: 10.1016/j.neuroima- phase III randomized controlled trials. Radiother Oncol. 2012; 102(2): ge.2013.06.006, indexed in Pubmed: 23769921. 168–179, doi: 10.1016/j.radonc.2011.08.041, indexed in Pubmed: 33. Artaechevarria X, Munoz-Barrutia A, Ortiz-de-Solorzano C. Combina- 21996522. tion strategies in multi-atlas image segmentation: application to brain 14. Oğurel T, Oğurel R, Özer MA, et al. Mini-mental state exam versus Mont- MR data. IEEE Trans Med Imaging. 2009; 28(8): 1266–1277, doi: real Cognitive Assessment in patients with diabetic retinopathy. Niger 10.1109/TMI.2009.2014372, indexed in Pubmed: 19228554. J Clin Pract. 2015; 18(6): 786–789, doi: 10.4103/1119-3077.163274, 34. Heckemann RA, Hajnal JV, Aljabar P, et al. Automatic anatomical indexed in Pubmed: 26289518. brain MRI segmentation combining label propagation and decision 15. Welzel G, Fleckenstein K, Mai SK, et al. Acute neurocognitive fusion. Neuroimage. 2006; 33(1): 115–126, doi: 10.1016/j.neuroima- impairment during cranial radiation therapy in patients with intra- ge.2006.05.061, indexed in Pubmed: 16860573. cranial tumors. Strahlenther Onkol. 2008; 184(12): 647–654, doi: 35. Tu Z, Bai X. Auto-context and its application to high-level vision tasks 10.1007/s00066-008-1830-6, indexed in Pubmed: 19107345. and 3D brain image segmentation. IEEE Trans Pattern Anal Mach Intell. 16. Kanard A, Frytak S, Jatoi A. Cognitive dysfunction in patients with 2010; 32(10): 1744–1757, doi: 10.1109/TPAMI.2009.186, indexed in small-cell lung cancer: incidence, causes, and suggestions on Pubmed: 20724753. management. J Support Oncol. 2004; 2(2): 127–140, indexed in 36. Dekeyzer S, De Kock I, Nikoubashman O, et al. “Unforgettable” Pubmed: 15328816. — a pictorial essay on anatomy and pathology of the hippocampus. In- 17. Falkowski A, Kurcz I. Procesy poznawcze. JS, Psychologia ogólna. sights Imaging. 2017; 8(2): 199–212, doi: 10.1007/s13244-016-0541-2, Podręcznik akademicki, t. 2. GWP, Gdańsk 2006. indexed in Pubmed: 28108955. 18. Meyers CA, Smith JA, Bezjak A, et al. Neurocognitive function and 37. Wang S, Zheng D, Zhang C, et al. Automatic planning on hippocampal progression in patients with brain metastases treated with whole-brain avoidance whole-brain radiotherapy. Med Dosim. 2017; 42(1): 63–68, radiation and motexafin gadolinium: results of a randomized phase III doi: 10.1016/j.meddos.2016.12.002, indexed in Pubmed: 28237294. trial. J Clin Oncol. 2004; 22(1): 157–165, doi: 10.1200/JCO.2004.05.128, 38. Becerril-Villanueva E, Ponce-Regalado MD, Pérez-Sánchez G, et indexed in Pubmed: 14701778. al. Chronic infection with Mycobacterium lepraemurium induces 19. Chang EL, Wefel JS, Hess KR, et al. Neurocognition in patients with alterations in the hippocampus associated with memory loss. Sci brain metastases treated with radiosurgery or radiosurgery plus whole- Rep. 2018; 8(1): 9063, doi: 10.1038/s41598-018-27352-x, indexed in -brain irradiation: a randomised controlled trial. Lancet Oncol. 2009; Pubmed: 29899533. 10(11): 1037–1044, doi: 10.1016/S1470-2045(09)70263-3, indexed 39. Yuede CM, Timson BF, Hettinger JC, et al. Interactions between stress in Pubmed: 19801201. and physical activity on Alzheimer’s disease pathology. Neurobiol 20. Meyers CA, Brown PD. Role and relevance of neurocognitive as- Stress. 2018; 8: 158–171, doi: 10.1016/j.ynstr.2018.02.004, indexed sessment in clinical trials of patients with CNS tumors. J Clin Oncol. in Pubmed: 29888311. 2006; 24(8): 1305–1309, doi: 10.1200/JCO.2005.04.6086, indexed in 40. Peters MEM, Kockelkoren R, de Brouwer EJM, et al. Histological valida- Pubmed: 16525186. tion of calcifications in the human hippocampus as seen on computed 21. Meyers CA, Smith JA, Bezjak A, et al. Neurocognitive function and tomography. PLoS One. 2018; 13(5): e0197073, doi: 10.1371/journal. progression in patients with brain metastases treated with whole-brain pone.0197073, indexed in Pubmed: 29750809. radiation and motexafin gadolinium: results of a randomized phase III 41. Marsh JC, Godbole R, Diaz AZ, et al. Sparing of the hippocampus, lim- trial. J Clin Oncol. 2004; 22(1): 157–165, doi: 10.1200/JCO.2004.05.128, bic circuit and neural stem cell compartment during partial brain radio- indexed in Pubmed: 14701778. therapy for glioma: a dosimetric feasibility study. J Med Imaging Radiat 22. Jehn CF, Becker B, Flath B, et al. Neurocognitive function, brain-de- Oncol. 2011; 55(4): 442–449, doi: 10.1111/j.1754-9485.2011.02282.x, rived neurotrophic factor (BDNF) and IL-6 levels in cancer patients indexed in Pubmed: 21843181. with depression. J Neuroimmunol. 2015; 287: 88–92, doi: 10.1016/j. 42. Canyilmaz E, Uslu GD, Colak F, et al. Comparison of dose distributions jneuroim.2015.08.012, indexed in Pubmed: 26439967. hippocampus in high grade gliomas irradiation with linac-based imrt 23. Petty F, Noyes R. Depression secondary to cancer. Biol Psychiatry. and volumetric arc therapy: a dosimetric study. Springerplus. 2015; 4: 1981; 16(12): 1203–1220, indexed in Pubmed: 6984347. 114, doi: 10.1186/s40064-015-0894-x, indexed in Pubmed: 25815244. 24. Regine WF, Schmitt FA, Scott CB, et al. Feasibility of neurocognitive 43. Chan JL, Lee SW, Fraass BA, et al. Survival and failure patterns of outcome evaluations in patients with brain metastases in a multi- high-grade gliomas after three-dimensional conformal radiotherapy. J -institutional cooperative group setting: results of Radiation Therapy Clin Oncol. 2002; 20(6): 1635–1642, doi: 10.1200/JCO.2002.20.6.1635, Oncology Group trial BR-0018. Int J Radiat Oncol Biol Phys. 2004; indexed in Pubmed: 11896114. 58(5): 1346–1352, doi: 10.1016/j.ijrobp.2003.09.023, indexed in 44. Marsh J, Godbole R, Diaz A, et al. Feasibility of cognitive sparing Pubmed: 15050309. approaches in children with intracranial tumors requiring partial brain 25. Marsh JC, Herskovic AM, Gielda BT, et al. Intracranial metastatic radiotherapy: A dosimetric study using tomotherapy. journal of Cancer disease spares the limbic circuit: a review of 697 metastatic lesions Therapeutics and Research. 2012; 1(1): 1, doi: 10.7243/2049-7962-1-1. in 107 patients. Int J Radiat Oncol Biol Phys. 2010; 76(2): 504–512, 45. Brodin NP, Munck af Rosenschöld P, Blomstrand M, et al. Hippo- doi: 10.1016/j.ijrobp.2009.02.038, indexed in Pubmed: 20117288. campal sparing radiotherapy for pediatric medulloblastoma: impact 26. Ghia A, Tomé WA, Thomas S, et al. Distribution of brain metastases in of treatment margins and treatment technique. Neuro Oncol. 2014; relation to the hippocampus: implications for neurocognitive functional 16(4): 594–602, doi: 10.1093/neuonc/not225, indexed in Pubmed: preservation. Int J Radiat Oncol Biol Phys. 2007; 68(4): 971–977, doi: 24327585. 10.1016/j.ijrobp.2007.02.016, indexed in Pubmed: 17446005. 46. Lester JF, MacBeth FR, Coles B. Prophylactic cranial irradiation for 27. Sun A, Bae K, Gore EM, et al. Phase III trial of prophylactic cranial preventing brain metastases in patients undergoing radical treatment irradiation compared with observation in patients with locally advanced for non-small-cell lung cancer: a Cochrane Review. Int J Radiat Oncol non-small-cell lung cancer: neurocognitive and quality-of-life analysis. J Biol Phys. 2005; 63(3): 690–694, doi: 10.1016/j.ijrobp.2005.03.030, Clin Oncol. 2011; 29(3): 279–286, doi: 10.1200/JCO.2010.29.6053, indexed in Pubmed: 15913909. indexed in Pubmed: 21135267. 47. Stuschke M, Pöttgen C. Prophylactic cranial irradiation as a compo- 28. Meyers CA, Smith JA, Bezjak A, et al. Neurocognitive function and nent of intensified initial treatment of locally advanced non-small cell progression in patients with brain metastases treated with whole-brain lung cancer. Lung Cancer. 2003; 42 Suppl 1: S53–S56, indexed in radiation and motexafin gadolinium: results of a randomized phase III Pubmed: 14611915. trial. J Clin Oncol. 2004; 22(1): 157–165, doi: 10.1200/JCO.2004.05.128, 48. Le Péchoux C, Dunant A, Senan S, et al. Prophylactic Cranial Irradiation indexed in Pubmed: 14701778. (PCI) Collaborative Group. Standard-dose versus higher-dose prophy- 29. www.rtog.org/CoreLab/ContouringAtlases/HippocampalSparing.aspx. lactic cranial irradiation (PCI) in patients with limited-stage small-cell 30. Cacao E, Cucinotta FA. Modeling Heavy-Ion Impairment of Hippo- lung cancer in complete remission after chemotherapy and thoracic campal Neurogenesis after Acute and Fractionated Irradiation. Radiat radiotherapy (PCI 99-01, EORTC 22003-08004, RTOG 0212, and IFCT Res. 2016; 186(6): 624–637, doi: 10.1667/RR14569.1, indexed in 99-01): a randomised clinical trial. Lancet Oncol. 2009; 10(5): 467–474, Pubmed: 27925861. doi: 10.1016/S1470-2045(09)70101-9, indexed in Pubmed: 19386548. 31. Konopka-Filippow M, Sierko E, Hempel D, et al. Learning curve 49. Redmond KJ, Hales RK, Anderson-Keightly H, et al. Prospective Study and interobserver variability in contouring hippocampus under of Hippocampal-Sparing Prophylactic Cranial Irradiation in Limited- the guidelines of Radiation Therapy Oncology Group 0933 hippo- -Stage Small Cell Lung Cancer. Int J Radiat Oncol Biol Phys. 2017; campal sparing atlas recommendations. Radiother Oncol. 2018; 98(3): 603–611, doi: 10.1016/j.ijrobp.2017.03.009, indexed in Pubmed: 127: S1238. 28581401.

109 Oncology in clinical practice 2019, Vol. 15, No. 2

50. Tarnawski R, Michalecki L, Blamek S, et al. Feasibility of reducing the J Radiat Oncol Biol Phys. 2010; 78(4): 1244–1252, doi: 10.1016/j. irradiation dose in regions of active neurogenesis for prophylactic ijrobp.2010.01.039, indexed in Pubmed: 20598457. cranial irradiation in patients with small-cell lung cancer. Neoplasma. 56. Krayenbuehl J, Di Martino M, Guckenberger M, et al. Improved 2011; 58(6): 507–515, indexed in Pubmed: 21895404. plan quality with automated radiotherapy planning for whole brain with 51. Marsh JC, Gielda BT, Herskovic AM, et al. Cognitive Sparing during the hippocampus sparing: a comparison to the RTOG 0933 trial. Radiat Administration of Whole Brain Radiotherapy and Prophylactic Cranial Oncol. 2017; 12(1): 161, doi: 10.1186/s13014-017-0896-7, indexed Irradiation: Current Concepts and Approaches. J Oncol. 2010; 2010: in Pubmed: 28969706. 198208, doi: 10.1155/2010/198208, indexed in Pubmed: 20671962. 57. Zhao L, Shen Y, Guo JD, et al. Analyses of distribution and dosimetry 52. Tallet AV, Azria D, Barlesi F, et al. Neurocognitive function impairment of brain metastases in small cell lung cancer with relation to the neural after whole brain radiotherapy for brain metastases: actual assessment. stem cell regions: feasibility of sparing the hippocampus in prophylactic Radiat Oncol. 2012; 7: 77, doi: 10.1186/1748-717X-7-77, indexed in cranial irradiation. Radiat Oncol. 2017; 12(1): 118, doi: 10.1186/s13014- Pubmed: 22640600. 017-0855-3, indexed in Pubmed: 28709456. 53. Nevelsky A, Ieumwananonthachai N, Kaidar-Person O, et al. Hippo- 58. Gondi V, Mehta MP, Pugh S, et al. Memory Preservation With Conformal campal-sparing whole-brain radiotherapy using Elekta equipment. Avoidance of the Hippocampus During Whole-Brain Radiation Therapy J Appl Clin Med Phys. 2013; 14(3): 4205, doi: 10.1120/jacmp. for Patients With Brain Metastases: Primary Endpoint Results of RTOG v14i3.4205, indexed in Pubmed: 23652251. 0933. Int J Radiat Oncol Biol Phys. 2013; 87(5): 1186, doi: 10.1016/j. 54. Gondi V, Pugh SL, Tome WA, et al. Preservation of memory with ijrobp.2013.10.005. conformal avoidance of the hippocampal neural stem-cell compart- 59. Daniela Falco M, Giancaterino S, D’Andrea M, et al. Hippocampal ment during whole-brain radiotherapy for brain metastases (RTOG sparing approach in fractionated stereotactic brain VMAT radio therapy: 0933): a phase II multi-institutional trial. J Clin Oncol. 2014; 32(34): A retrospective feasibility analysis. J Appl Clin Med Phys. 2018; 19(1): 3810–3816, doi: 10.1200/JCO.2014.57.2909, indexed in Pubmed: 86–93, doi: 10.1002/acm2.12216, indexed in Pubmed: 29125239. 25349290. 60. Di Carlo C, Trignani M, Caravatta L, et al. Hippocampal sparing in 55. Gondi V, Tolakanahalli R, Mehta MP, et al. Hippocampal-sparing whole- stereotactic radiotherapy for brain metastases: To contour or not -brain radiotherapy: a “how-to” technique using helical tomotherapy contour the hippocampus? Cancer Radiother. 2018; 22(2): 120–125, and linear accelerator-based intensity-modulated radiotherapy. Int doi: 10.1016/j.canrad.2017.08.113, indexed in Pubmed: 29576492.

110 REVIEW ARTICLE

Ewa Cedrych1, Ida Cedrych2 1Department of Oncology, Uniwersyteckie Centrum Kliniczne im. prof. K. Gibińskiego, Medical University of Silesia, Katowice, Poland 2University Hospital Reykjavik, Iceland

Neratinib in adjuvant treatment of patients with HER2-positive breast cancer — less is more?

Address for correspondence: ABSTRACT Lek. Ewa Cedrych Neratinib is a new small molecule aimed at HER2 receptor. It has recently been approved in the United States of Oddział Onkologii, Uniwersyteckie America and Europe for adjuvant treatment of patients with early, HER2-positive breast cancer, who underwent Centrum Kliniczne im. prof. K. Gibińskiego surgical resection followed by at least one year of adjuvant trastuzumab treatment. Despite initial enthusiasm, Śląskiego Uniwersytetu Medycznego several factors limit the implementation of neratinib in clinical practice. These include: modest reduction of recur- w Katowicach rence rate; limited data regarding the effect on overall survival; and a significant rate of adverse events. Thus, e-mail: [email protected] neratinib should be considered mainly in patients with high-risk HER2-positive breast cancer, because its clinical

Oncology in Clinical Practice benefit might outweigh the side effects in this population. In the following article, we discuss the controversies 2019, Vol. 15, No. 2, 111–114 regarding the pivotal phase III trial that eventually led to neratinib approval. DOI: 10.5603/OCP.2019.0014 Key words: neratinib, breast cancer, HER2 Translation: lek. Maciej Kawecki Copyright © 2019 Via Medica ISSN 2450–1654 Oncol Clin Pract 2019; 15, 2: 111–114

Introduction to the impairment of cell proliferation. In a phase II trial published in 2010, neratinib was evaluated in Recent years have brought significant improvement patients with metastatic, HER2-positive breast cancer. in systemic treatment of breast cancer. Recognition of The patients were divided into two cohorts — the first the biological mechanisms responsible for breast cancer included patients with a confirmed progression during development and growth have led to the introduction of trastuzumab treatment, and the second only patients personalised treatment that bases on the immunological naïve to trastuzumab. Neratinib was used at a dose of phenotype of cancer cells. Systemic treatment arma- 240 mg per day. The primary end point was the rate mentarium for breast cancer in the USA includes over of progression-free survival (PFS) after 16 weeks of 30 agents nearly half of which are molecularly targeted treatment. The rate of PFS was 56% in the first cohort drugs. Neratinib is one of the most recent additions to and 78% in the second cohort, with a median PFS of this list and acts as a molecularly targeted drug [1]. 22.3 and 39.6 weeks, respectively. The response rate was 24% in patients who progressed on trastuzumab and 56% in trastuzumab-naïve patients. The most common Phase II clinical trials treatment-related adverse events were diarrhoea, nausea, vomiting, and fatigue. Diarrhoea usually occurred Neratinib is a small molecule oral tyrosine kinase in the first week of treatment and was present in 93% inhibitor. It irreversibly blocks human epidermal growth of patients, with grade 3 and 4 intensity (according factor receptor 2 (ERBB2, HER2) and epidermal to National Cancer Institute Common Terminology growth factor receptor (EGFR) [2]. The mechanism of Criteria for Adverse Events v.3.0, CTCAE) in 21% of action relies on the suppression of ErbB and autophos- patients. Grade 3 and 4 diarrhoea was more prevalent in phorylation of EGFR family receptor proteins, leading the cohort of patients with prior trastuzumab exposure

111 Oncology in clinical practice 2019, Vol. 15, No. 2

(30%) compared to the cohort of trastuzumab-naïve ruptions for no longer than three weeks were allowed. patients (13%). Dose reduction was required in nearly No diarrhoea prophylaxis was included; treatment was one-third of patients in cohort one (29%), in contrast initiated if necessary. Besides routine clinical and ra- to only 4% of patients who needed dose reductions in diological assessment, the trial included quality-of-life cohort two [3]. The next phase II trial, the results of evaluation, using EuroQoL Five Dimensions (EQ5D) which were published three years later, aimed at proving and Functional Assessment of Cancer Therapy-Breast non-inferiority of neratinib monotherapy to a combina- (FACT-B) questionnaires. Quality of life evaluation was tion of lapatinib and capecitabine in patients with locally undertaken every three months. The primary end-point advanced or metastatic HER2-positive breast cancer was invasive disease-free survival (iDFS) assessed after prior trastuzumab treatment. Patients in the experi- 24 months after randomisation. Secondary end-points mental arm received neratinib 240 mg per day continu- were: DFS including incidence of preinvasive breast ously (n = 117), and patients in the control arm received cancer, distant recurrence-free survival, cumulated in- lapatinib 1250 mg per day continuously and capecitabine cidence of central nervous metastases incidence, overall 2000 mg/m2 daily on days 1–14 of every 21 day cycle survival, and treatment safety. (n = 116). The trial failed to show non-inferiority of In February 2010 the protocol was modified, nar- neratinib. Median PFS in patients receiving neratinib rowing the inclusion criteria to patients with stage II was 4.5 months versus 6.8 months in patients receiving and III disease and to those who finished trastuzumab lapatinib and capecitabine, with a median overall sur- therapy within a year. The amendment was justified by vival (OS) of 19.7 and 23.6 months, respectively. The the results of two other published clinical trials (BCIRG response rate was lower in the experimental arm (29%) 006), which showed excellent survival parameters of compared to the control arm (41%) (p = 0.067), as was patients with HER2 overexpressed breast cancer with- the rate of clinical benefit (44% vs. 64%, respectively) out involvement of local lymph nodes, who received (p = 0.003). Patients receiving neratinib experienced adjuvant trastuzumab. Additionally, most of the recur- more diarrhoea (85%) compared to patients receiving rences occurred within one year of completing adjuvant lapatinib and capecitabine (68%) (p = 0.002) [4]. Several treatment [8]. In October 2011 subsequent amendments other phase I/II trials evaluated neratinib combined with were implemented. The trial finished recruitment after different cytotoxic agents (vinorelbine, capecitabine, only 2842 patients from 3850 initially planned, and the and paclitaxel) in the treatment of patients with breast observation period was limited to two years instead of cancer, with a modest signs of activity [5–7]. a previously accounted five years. Those amendments, as described with published results, were due to the sponsor’s doubts regarding safety [9]. Moreover, a previous Phase III clinical trial sponsor withdrew financial support for the trial, which was continued by a small pharmaceutical company with Results of a prospective, randomised, multicentre, limited experience in oncology. double-blinded, placebo-controlled trial evaluat- The next amendment was implemented in January ing neratinib in adjuvant treatment of patients with 2014. The decision revived some of the initial study HER2-positive breast cancer, who underwent surgery assumptions and prolonged the period of observa- followed by at least one-year of trastuzumab treatment, tion, again, to five years, with an additional efficiency were published in 2016 in „The Lancet Oncology”. The analysis in the whole study population after 24 months trial included patients from 495 oncological centres from randomisation. The aforementioned analysis was from all continents. Initially, the trial included patients undertaken in June 2014. In the neratinib arm, there over 18 years old with stage I–III HER2 breast cancer, were 70 events of invasive breast cancer in comparison who finished 12-months of adjuvant trastuzumab treat- to 109 events in the placebo arm (HR 0.67; 95% CI ment within the last two years. Patients whose tumours 0.50–0.91; p = 0.0091). The difference was even more expressed oestrogen receptors were recommended to clear in a subgroup of patients with positive hormone receive endocrine therapy simultaneously. Hormonal receptors (HR 0.51; 95% CI 0.53–0.77; p = 0.0013). No receptor expression was evaluated locally and was not difference in iDFS was seen in patients without expres- verified by a central laboratory. No homogenous method sion of hormonal receptors. There were no differences of receptor evaluation was required. Inclusion criteria regarding the metastasis-free survival (HR 0.75; 95% included only typical factors: adequate performance sta- CI 0.53–1.04; p = 0.089) or the distant recurrence-free tus (ECOG 0–1); no clinical or laboratory contraindica- survival (HR 0.71; 95% CI 0.50–1.0; p = 0.054). The cu- tions arising from liver, kidneys, or heart; no diagnosis of mulative incidence of central nervous system metastases mental diseases; and no difficulties with oral ingestion. was similar in both arms (0.91% [95% CI 0.49–1.59] and Neratinib was administered at a daily dose of 240 mg 1.25% [95% CI 0.75–1.99]; p = 0.44). Quality of life as- continuously for 12 months. Dose reductions or inter- sessment showed impaired scores in the neratinib group

112 Ewa Cedrych, Ida Cedrych, Breast cancer treatment with neratinib during the first month of treatment, with a subsequent ing diarrhoea (22 vs. one event/s), vomiting (12 vs. one diminishment of difference between both arms. event/s) and dehydration (nine vs. one event/s). Grade The restitution of a five-year period of observation 2, 3, and 4 diarrhoea usually emerged during the first required renewal of informed consent, which was ob- week of treatment, with an increased risk during the tained from 2117 patients (74.4% of the primary group). subsequent few weeks. The presented trial did not Data regarding patients who refused re-consent were include diarrhoea prophylaxis. However, conclusions censored at the date of last control. Results from the arising from the trial suggest that this kind of prophylaxis five-year observation were published in „The Lancet should be encouraged in patients receiving neratinib. Oncology” [10] and were comparable to the results The prophylaxis should be initiated concurrently with published after the 24-month analysis. The median neratinib and continued for the first two treatment cycles treatment time was 353 days in the neratinib group and and as clinically indicated thereafter [10, 11]. Quality of 360 days in the placebo group. Among patients with life assessment is an essential factor during evaluation of hormonal-receptor-positive cancer, 93% of patients in novel drugs, especially when applied as part of adjuvant the experimental arm and 94% of patients in the placebo treatment. Results regarding quality of life in the Ex- arm received simultaneous hormonal treatment, but at teNet trial were published as a conference paper during the time of final analysis only 52% and 47%, respectively, the European Society for Medical Oncology Congress continued endocrine treatment. The difference between in 2017. The presented results included evaluation of arms might be attributed to the higher rate of disease health-related quality of life (HRQoL) assessment with recurrence in the control arm, which forced hormonal two properly validated questionnaires: EuroQol Five therapy withdrawal. Long-term analysis showed a sig- Dimensions (EQ5D) and Functional Assessment of nificantly lower rate of invasive breast cancer in the Cancer Therapy-Breast (FACT-B). The assessment was neratinib group compared to the control group (116 and undertaken every three months for 12 months. Similar 163 events, respectively; HR 0.73; 95% CI 0.57–0.92; patterns of HRQoL changes were detected in both p = 0.0083). The rate of five-year recurrence-free FACT-B and EQ-5D questionnaires: the quality of life survival was 90.2% (95% CI 88.3–91.8) in the neratinib scores in the neratinib group worsened in the first month arm and 87.7% (95% CI 85.7–89.4) in the placebo arm. of treatment but improved thereafter [11]. There were no significant differences between rates and medians of distant recurrence-free survival as well as cumulative five-year risk of central nervous system Discussion metastasis development. Again, the greatest difference in recurrence-free survival was seen in a sub-group of Neratinib used as part of adjuvant treatment in patients with a cancer expressing hormonal receptors patients with early-stage HER2-positive breast cancer (HR 0.60; 95% CI 0.43–0.83; p = 0.063), and no differ- improves invasive disease-free survival and non-invasive ence was seen in patients without receptor expression disease-free survival, especially in patients with tumours (HR 0.95; 95% CI 0.66–1.35). No long-term toxicities expressing hormonal receptors. No improvement with attributed to neratinib were observed, including rates neratinib was seen regarding distant metastasis-free of cardio-vascular diseases and rates of secondary ma- survival and rate of central nervous system metastasis lignancies. During the five-year observation, there were incidence. Additionally, we currently lack data re- 121 incidences of death, 102 of which can be attributed garding survival parameters. Better results obtained to cancer and 19 to other causes. The results regarding with neratinib in hormonal-positive patients might be overall survival have not been published yet because explained by an interaction of HER2 and oestrogen such analysis is planned for the third quarter of 2019 [10]. receptors. Inhibition of the former leads to an overex- pression of the latter, sensitising cancer cells to hormonal treatment. It should be emphasised that improved Adverse events effectiveness in the hormonal-positive population has not been observed in a trial with other HER2 inhibi- The most common adverse event during neratinib tors: trastuzumab, pertuzumab, or lapatinib. A possible treatment is diarrhoea, with all grade incidents present explanation might be that neratinib itself can interact in over 70% of patients (33% grade 2, 40% grade 3, and with oestrogen receptors. However, as the central as- less than 1% grade 4 diarrhoea). Treatment emergent sessment evaluated only HER2 receptor status and not adverse events were responsible for a 28% rate of treat- hormonal receptor expression, the differences between ment withdrawal in the neratinib group (compared to multiple local standards might bias correct interpreta- 2% in the placebo group) and a 31% rate of dose ad- tion of data. Several controversies arise due to the few justment (compared to only 2% patients in the placebo protocol amendments undertaken during the trial and group). Serious adverse events were reported in 7% and due to the change of sponsorship. The decision regard- 6% of patients, respectively, with the most common being limitation of the number of participants correlated

113 Oncology in clinical practice 2019, Vol. 15, No. 2 with a primary analysis of a different small molecule advancements, has shortened the cycle of drug develop- HER2 inhibitor — lapatinib — used as a monotherapy ment and allows rapid introduction of novel compounds in adjuvant treatment of HER2-positive breast cancer. to the market. The inferiority of lapatinib compared to trastuzumab Still, it should be remembered that the daily prac- led to an early trial suspension, and patients receiv- ticing oncologist, who meets patients and their needs, ing lapatinib were offered to continue their treatment is the one responsible for the final recommendation of with trastuzumab [12]. Because the neratinib trial was certain treatment for a certain patient. Knowledge about modified at the same time that the lapatinib trial results the full process the development of a novel drug offers were made available, the sponsor might have limited a valuable insight into its clinical utility and supports recruitment only to reduce costs due to a high risk of precise and accurate therapeutic decisions [15]. obtaining negative results. Despite the lack of overall survival data, on 17 June 2017 the Food and Drug Ad- ministration (FDA) approved neratinib as an adjuvant References treatment in adult early-stage HER-2 positive breast cancer patients who received at least 12 months of tras- 1. National Institutes of Health. National Cancer Institute 2018. Neratinib Maleate. https://www.cancer.gov/about-cancer/treatment/drugs/nera- tuzumab treatment [13]. Initially, in February 2018 the tinibmaleate (2017–06–27). European Committee for Medicinal Products for Hu- 2. National Cancer Institute Drug Dictionary. https://www.cancer. gov/publications/dictionaries/cancer-drug/def/neratinib-maleate man Use (CHMP) did not recommended registration (2018–12–01). of neratinib due to uncertainties whether the clinical 3. Burstein HJ, Sun Y, Dirix LY, et al. Neratinib, an irreversible ErbB receptor tyrosine kinase inhibitor, in patients with advanced ErbB2- benefit outweighs the increased toxicity [14]. However, -positive breast cancer. J Clin Oncol. 2010; 28(8): 1301–1307, doi: four months later, on 28 June 2018 the CHMP revised its 10.1200/JCO.2009.25.8707, indexed in Pubmed: 20142587. position and recommended registration of neratinib, but 4. Martin M, Bonneterre J, Geyer CE, et al. A phase two randomised trial of neratinib monotherapy versus lapatinib plus capecitabine combi- only in the population of patients with HER2-positive nation therapy in patients with HER2+ advanced breast cancer. Eur and hormone receptor-positive tumours. Since then, J Cancer. 2013; 49(18): 3763–3772, doi: 10.1016/j.ejca.2013.07.142, indexed in Pubmed: 23953056. neratinib can be considered as an option for extended 5. Awada A, Dirix L, Manso Sanchez L, et al. Safety and efficacy of adjuvant treatment of HER2-positive and hormonal neratinib (HKI-272) plus vinorelbine in the treatment of patients with ErbB2-positive metastatic breast cancer pretreated with anti-HER2 the- receptor-positive early-stage breast cancer [13]. rapy. Ann Oncol. 2013; 24(1): 109–116, doi: 10.1093/annonc/mds284, The monthly cost of neratinib therapy in the USA indexed in Pubmed: 22967996. is estimated at about 10,000 dollars, and, considering 6. Saura C, Garcia-Saenz JA, Xu B, et al. Safety and efficacy of neratinib in combination with capecitabine in patients with metastatic human epi- a standard 12-month regimen, a single neratinib treatment dermal growth factor receptor 2-positive breast cancer. J Clin Oncol. requires expenditure of 120,000 dollars. Unfortunately, 2014; 32(32): 3626–3633, doi: 10.1200/JCO.2014.56.3809, indexed in Pubmed: 25287822. toxicities related to neratinib occur commonly, add a sub- 7. Chow LWC, Xu B, Gupta S, et al. Combination neratinib (HKI-272) and stantial burden, require intensive prophylaxis, and can paclitaxel therapy in patients with HER2-positive metastatic breast cancer. Br J Cancer. 2013; 108(10): 1985–1993, doi: 10.1038/bjc.2013.178, impair quality of life (at least temporarily during the first indexed in Pubmed: 23632474. months of treatment). Valuable insights might come with 8. Slamon D, Eiermann W, Robert N, et al. Breast Cancer International Re- the data regarding survival parameters, but they will not be search Group. Adjuvant trastuzumab in HER2-positive breast cancer. N Engl J Med. 2011; 365(14): 1273–1283, doi: 10.1056/NEJ- available until late 2019. Additionally, quality of life analy- Moa0910383, indexed in Pubmed: 21991949. sis of neratinib has been published only as a conference 9. Chan A, Delaloge S, Holmes F, et al. Neratinib after trastuzumab- -based adjuvant therapy in patients with HER2-positive breast cancer paper, which limits the ability to draw proper conclusions. (ExteNET): a multicentre, randomised, double-blind, placebo-controlled, phase 3 trial. The Lancet Oncology. 2016; 17(3): 367–377, doi: 10.1016/s1470-2045(15)00551-3. 10. Martin M, Holmes FA, Ejlertsen B, et al. ExteNET Study Group. Neratinib Conclusions after trastuzumab-based adjuvant therapy in HER2-positive breast cancer (ExteNET): 5-year analysis of a randomised, double-blind, placebo-controlled, phase 3 trial. Lancet Oncol. 2017; 18(12): 1688–1700, Recent advances in systemic therapy of solid tu- doi: 10.1016/S1470-2045(17)30717-9, indexed in Pubmed: 29146401. 11. Delaloge S, Ye Y, Cella D, et al. Effects of neratinib on health-related qual- mours are limited mostly to targeted therapies and ity of life (HRQoL) in early-stage HER2+ breast cancer: longitudinal anal- drugs aimed at the immune system (mostly immune yses from the phase III ExteNET trial. ESMO 2017 poster 06 Sep 2017. 12. Piccart-Gebhart M, Holmes E, Baselga J, et al. Adjuvant Lapatinib checkpoint inhibitors). The last cytotoxic drug for breast and Trastuzumab for Early Human Epidermal Growth Factor Receptor cancer was registered in the USA in 2012. Since then, 2-Positive Breast Cancer: Results From the Randomized Phase III Adjuvant Lapatinib and/or Trastuzumab Treatment Optimization Trial. all drugs registered were molecularly targeted thera- J Clin Oncol. 2016; 34(10): 1034–1042, doi: 10.1200/JCO.2015.62.1797, pies. Trials evaluating immune checkpoint inhibitors in indexed in Pubmed: 26598744. 13. FDA. Highlights of prescribing information Nerlynx (neratinib) tablets, triple-negative breast cancer are underway. Strong pres- for oral use. https://www.accessdata.fda.gov/drugsatfda_docs/labe- sion from millions of breast cancer patients around the l/2017/208051s000lbl.pdf (2018–10–02). 14. European Medicines Agency. Positive opinion on the marketing au- globe, as well as technological development, has led to thorisation for Nerlynx (neratinib). Outcome of re-examination. EMA. the introduction of novel and promising therapies. The 2018; 433352: 1–2. exponentially growing body of evidence regarding can- 15. Neratinib Is Approved: Should We Reject It Anyway? The ASCO Post. http://www.ascopost.com/issues/december-25-2017/neratinib-is-ap- cer genetics, supported by more precise technological proved-should-we-reject-it-anyway/ (2017–12–25).

114 REVIEW ARTICLE

Katarzyna Kozak, Piotr Rutkowski Department of Soft Tissue/Bone Sarcoma and Melanoma, Maria Sklodowska-Curie Institute — Oncology Centre, Warsaw, Poland

Why do we need a new BRAF-MEK inhibitor combination in melanoma?

Address for correspondence: ABSTRACT Lek. Katarzyna Kozak Despite the increasing role of immunotherapy, BRAF/MEK inhibitor combinations have still a central role in the Klinika Nowotworów Tkanek Miękkich, treatment of BRAF V600-mutant melanoma. Encorafenib-binimetinib is the third BRAF-MEK inhibitor combination ap- Kości i Czerniaków proved for the metastatic melanoma with BRAF V600 mutation. Data from phase III trial demonstrated high antitumor Centrum Onkologii — Instytut efficacy and good tolerability of encorafenib-binimetinib. Compared to other combinations (dabrafenib-trametinib, im. Marii Skłodowskiej-Curie vemurafenib-cobimetinib) the new combination showed favourable results in terms of the low rates of pyrexia ul. Roentgena 5, 02–781 Warszawa and photosensitivity. Trials with triplet regimens that combine encorafenib-binimetinib with immunotherapy or Phone: 22 546 21 84 a third targeted agent in an effort to overcome mechanisms of resistance to BRAF/MEK inhibition are ongoing. Fax: 22 643 93 75 Key words: advanced melanoma, BRAF mutation, encorafenib, binimetinib e-mail: [email protected]

Oncol Clin Pract 2019; 15, 2: 115–119 Oncology in Clinical Practice 2019, Vol. 15, No. 2, 115–119 DOI: 10.5603/OCP.2018.0054 Translation: dr n. med. Dariusz Stencel Copyright © 2019 Via Medica ISSN 2450–1654

In the last few years, treatment of patients with is a selective inhibitor of MEK1 and MEK2 kinases, BRAF-mutant advanced melanoma has changed radi- which are components of MAPK signalling pathway. cally, not only in terms of new therapeutic options, but Its effectiveness was also evaluated in patients with also in terms of the number of available drugs. Mo- melanoma with a rare NRAS mutation (phase III lecularly targeted therapies (dabrafenib with trametinib, NEMO study). However, the progression-free survival vemurafenib with cobimetinib) and immunotherapy (PFS) improvement compared to dacarbazine (median (nivolumab, pembrolizumab, nivolumab with ipili- 2.8 vs. 1.5 months) was too small to allow registration of mumab) have significantly improved overall survival in a drug in this indication [12]. this group of patients [1–9]. Currently, a registered The activity of the combination of encorafenib with combination of BRAF/MEK inhibitors (encorafenib binimetinib in patients with metastatic BRAF-mutant with binimetinib) is being added to this group. melanoma was evaluated for the first time in a phase Similarly to dabrafenib and vemurafenib, en- Ib/II study. The doses selected for phase II were 400, corafenib is an ATP-competitive BRAF V600 kinase 450, or 600 mg daily for encorafenib and 90 mg daily inhibitor. It differs from other drugs in this group for binimetinib. Response was observed in 72–78% of by a more than 10 times longer dissociation half-life patients, and median PFS was 11.3 months [13]. These (> 30 h), which results in extended inhibition of mito- encouraging results led to a phase III trial (COLUM- gen-activated protein kinase (MAPK) signalling path- BUS) comparing the efficacy of encorafenib + bini- way [10]. It probably results in more potent anti-cancer metinib combination with vemurafenib and encorafenib activity, with a smaller paradoxical upregulation of in monotherapy. In the first part of this study, the MAPK pathway in healthy tissues responsible for the patients were randomly assigned (1:1:1) to one of three development of side effects [10, 11]. In turn, binimetinib arms, receiving: encorafenib at a dose of 300 mg/day,

115 Oncology in clinical practice 2019, Vol. 15, No. 2

Table 1. Summary of treatment outcomes according to COLUMBUS protocol

COMBO450 COMBO300 ENCO300 (part 1+ 2) WEMURAFENIB n = 192 n = 258 n = 280 n = 191 Centrally Locally Centrally Locally Centrally Locally Centrally Locally assessed assessed assessed assessed assessed assessed assessed assessed Median PFS 14.9 14.8 12.9 12.9 9.2 9.2 7.3 7.3 (months; 95% CI) (11.0–18.5) (10.4–18.4) (10.1–14.0) (10.9–14.8) (7.4–11.0) (7.4–11.1) (5.6–8.2) (5.7–8.5) ORR (%; 95% CI) 63 75 66 73 50 56 40 49 (56–70) (68–81) (60–72) (67–78) (44–56) (50–62) (33–48) (42–57) CR (%) 8 16 8 11 5 8 6 7 PR (%) 55 59 58 62 45 49 35 42 Median DOR 16,6 16.2 12.7 13.1 12.9 13.0 12.3 8.4 (mo.; 95% CI) (12.2–20.4) (11.1–20.4) (9.3–15.1) (10.8–16.6) (8.9–15.5) (9.5–15.0) (6.9–16.9) (5.8–11.0)

Cl — confidence interval; CR — complete response; ORR — overall response rate; PR — partial response; DOR — duration of response vemurafenib at a dose 1920 mg/day or, encorafenib at to encorafenib monotherapy (HR 0.77, p = 0.029) but a dose of 450 mg/day in combination with binimetinib at shorter compared to the combination COMBO450 [16]. a dose of 90 mg/day. In total 577 patients were enrolled This confirms the relationship between encorafenib dose with unresectable/metastatic BRAF-mutant melanoma and the effectiveness of combined therapy. Table 1 sum- without prior systemic treatment or after one prior im- marises the treatment outcomes in the COLUMBUS munotherapy line. At median follow-up of 16.6 months, study, and Table 2 summarises the results of clinical independently assessed median PFS was 14.9 months trials with encorafenib and binimetinib. in the encorafenib + binimetinib arm (95% confidence The higher effectiveness of combination treatment interval [CI]: 11.0–18.50), 7.3 months in the vemurafenib is accompanied by better tolerance. Grade 3/4 adverse monotherapy arm (95% CI: 5.6–8.2), and 9.6 months events (AEs) were observed less frequently in patients (95% CI: 7.5–14.8) in the encorafenib monotherapy arm. receiving encorafenib with binimetinib (combination Locally assessed median PFS values were similar. Hazard therapy — 58%, vemurafenib — 63%, encorafenib ratio (HR) was 0.54 for combination therapy vs. vemu- — 66%), similarly to AEs requiring treatment inter- rafenib (p = 0.001) and 0.75 for combination therapy ruptions or dose modification. The maximum dose of vs. encorafenib (p = 0.051) in independent assessment encorafenib used as monotherapy, determined based on [14]. Of note, it is the first study showing a difference in previous research, is 300 mg/day [10]. The addition of efficacy of individual BRAF inhibitors in monotherapy binimetinib improved tolerance of encorafenib to the (encorafenib vs. vemurafenib), which confirms the high extent that the dose of encorafenib used in the combina- specificity of BRAF kinase inhibition by encorafenib. tion was increased to 450 mg/day, which contributed to In October 2018 the overall survival (OS) data of higher treatment effectiveness. However, it should be patients treated in the first part of the COLUMBUS remembered when modifying treatment to reduce the study were published [15]. Treatment with encorafenib dose of encorafenib to 300 mg/day in case of an inter- at a dose of 450 mg/day in combination with binimetinib ruption or withdrawal of binimetinib. at a dose of 90 mg/day (COMBO450) reduced the risk of The most common AEs observed in patients receiv- death compared to vemurafenib at a dose of 1920 mg/day ing combination therapy are gastrointestinal tract disor- (HR 0.61 [95% CI: 0.47–0.79], p < 0.001). Median OS ders (app. 30–40%), increased creatine kinase activity was 33.6 months (95% CI: 24.4–39.2) for the patients (23%), and fatigue (29%). Whilst gastrointestinal tract treated with COMBO 450 vs. 16.9 months (95% CI: disorders occurred more often than in patients receiving 14.0–24.5) for patients receiving vemurafenib. The monotherapy, muscle and joint pains, skin complications three-year OS rate for the combination of encorafenib (such as rash, hyperkeratosis, hand-foot syndrome, and with binimetinib was 47%. hypersensitivity to light), as well as hair loss were less In the second part of the COLUMBUS study mono- frequent. AEs specifically related to MEK inhibition, therapy with encorafenib at a dose of 300 mg/day was such as exudative serous chorioretinopathy (20–23%) compared with a combination of encorafenib at a dose and left ventricle functional disorders (2%), occurred of 300 mg/day with binimetinib at a dose of 90 mg/day more frequently during combination treatment [14, 15]. (COMBO300). Median PFS for combinations with The nature of AEs is similar in all BRAF/MEK encorafenib at a dose of 300 mg was 12.9 months (95% inhibitor combinations; only their prevalence is dif- CI: 10.1–14.0) and was significantly longer compared ferent. Fever, which is a typical AE of dabrafenib with

116 Katarzyna Kozak, Piotr Rutkowski, Why do we need a new BRAF-MEK inhibitor combination in melanoma?

Table 2. Summary of clinical trials of encorafenib and binimetinib

Study (year) Study design Efficacy outcomes Safety outcomes Ascierto et al. Multicentre, open Investigator-assessed RR: 20% in patients (2013) [9] phase II study, BINI with NRAS and BRAF mutations (6/30 and Common AE (NRAS+ and BRAF+; n = 71): 45 mg twice daily in 8/41 patients) acne-like dermatitis (46%), peripheral oede- melanoma patients PR confirmed in only 3 and 2 patients, ma (34%), diarrhoea (32%), elevated CPK with NRAS (n = 30) with no CR activity (28%), ocular toxicity (18%) and BRAF (n = 41) SD in 13 (42%) NRAS+ patients and Grade 3/4: 4 (5.6%) patients mutation 13 (32%) BRAF+ patients Treatment discontinuation due to AE: Survival: 15 (21%) patients — median PFS for NRAS+: 3.7 months Dose reduction due to AE: 33 (46%) patients (95% CI: 2.5–5.4) — median PFS for BRAF+: 3.6 months (95% CI: 2.0–3.8) Dummer et al. Multicentre, open Confirmed RR: 15.2% for BINI (95% CI: Common AE (BINI): increased CPK activity (2017) phase III study, ran- 11.2–20.1) vs. 6.8% for DTIC (95% CI: (42%), diarrhoea (40%), peripheral oedema NEMO [12] domisation 2:1: 3,1–12,5); p = 0.015 (36%), rash (36%), acne-like dermatitis BINI 45 mg twice SD: 40.5% (BINI) vs. 17.3% (DTIC) (35%), ocular toxicity (17%) daily (n = 269) vs. DTIC Survival: Severe AE: 91 (33.8%) patients, treatment 1000 mg/m2 IV every — median PFS: 2.8 months (95% CI: discontinuation due to AE: 66 (24.5%) 3 weeks (n = 133) in 2.8–3.6) for BINI vs. 1.5 months (95% CI: patients melanoma patients 1.5–1.7) for DTIC (HR: 0.62; p < 0.001) Dose reduction due to AE: 163 (60.6%) with NRAS mutations patients Dummer et al. Multicentre, open COMBO vs. VEM vs. ENCO Common AE (COMBO only): nausea (41%), (2018) phase III study, Confirmed RR: 63% (56–70) diarrhoea (36%), vomiting (30%), fatigue COLUMBUS [14] randomisation vs. 40% (33–48) vs. 51% (43–58) (29%), arthralgia (26%), elevated CPK activity 1:1:1 (n = 577): Survival: (23%), headache (22%), fever (18%), ocular ENCO 450 mg once toxicity (13%) — median PFS: 14.9 months (11.0–18.5) daily + BINI 45 mg vs. 7.3 months (5.6–8.2) vs. 9.6 months Grade 3/4 AE: 58% of patients twice daily (COMBO) (7.5–14.8); HR: 0.54 for COMBO Treatment discontinuation due to AE: vs. VEM 960 mg twice vs. VEM (p = 0.001) and 0.75 for COM- 16 (8%) patients daily vs. ENCO 300 mg BO vs. ENCO (p = 0.051) Dose reduction due to AE: 21 (11%) patients once daily (part 1) — median OS for COMBO: 33.6 months Dose interruption due to AE: 88 (46%) patients in melanoma patients with BRAF mutations

BINI — binimetinib; RR — response rate; PR — partial response; CR — complete response; SD — stable disease; PFS — progression-free survival; AE — adverse event; CPK — creatine phosphokinase; DTIC — dacarbazine; IV — intravenously; HR — hazard ratio; ENCO — encorafenib; VEM — vemurafenib trametinib (> 50% of patients), occurs less frequently ing. There are a few clinical trials ongoing at the present in patients receiving combinations of encorafenib with time: IMMU-TARGET (NCT02902042), assessing binimetinib (18%) and is not recurrent. Phototoxicity, the effectiveness of combination of encorafenib and also called photoirritation, which in turn occurs in half of binimetinib with anti-PD1 antibody, pembrolizumab; the patients treated with vemurafenib with cobimetinib, SECOMBIT (NCT02631447), assessing the optimal affects only 5% of patients treated with encorafenib and treatment sequence — encorafenib + binimetinib in binimetinib. Table 3 presents detailed data regarding the first line, nivolumab + ipilimumab in the second therapy tolerance in the COLUMBUS study. line — in comparison with the reverse sequence; EBIN The results of the COLUMBUS study led to the (NCT03235245), assessing the effectiveness of im- registration of a combination of encorafenib with munotherapy (nivolumab + ipilimumab) preceded by binimetinib by the US Food and Drug Administration a 12-week induction phase with the use of encorafenib (FDA) and European Medicines Agency (EMA) for and binimetinib; and LOGIC2, in which patients after the treatment of patients with unresectable/metastatic failure of treatment with encorafenib and binimetinib melanoma with BRAF mutation. receive further combinations of drugs based on the As combinations of BRAF/MEK inhibitors have assessment of molecular disorders in cancer tissue col- been used in daily clinical practice for several years, lected after disease progression. Activity of encorafenib attempts to modify the treatment in order to extend and binimetinib is also evaluated in patients with meta- the response duration or breaking the resistance to static colorectal cancer with BRAF mutation (phase III molecularly targeted drugs has become more interest- BEACON CRC study, NCT02928224).

117 Oncology in clinical practice 2019, Vol. 15, No. 2

Table 3. The most frequent adverse events in the arms containing encorafenib in the phase III COLUMBUS study

COMBO300 ENCO300 (part 1 + 2) COMBO 450

n = 257 n = 276 n = 192 Median duration of treatment 52.1 31.5 51 exposure (weeks) Adverse events (%) Any grade Grade 3/4 Any grade Grade 3/4 Any grade Grade 3/4 Diarrhoea 28 2 12 1 36 3 Nausea 27 2 36 3 41 2 Joint pain 22 1 43 8 26 1 Fatigue 22 1 26 1 29 2 Elevated creatine kinase activity 20 5 1 0 23 7 Vomiting 15 < 1 25 4 30 2 Elevated GGTP activity 14 5 11 4 15 9 Muscle pain 14 < 1 27 8 14 0 Alopecia 13 0 49 < 1 14 0 Headaches 12 < 1 26 3 22 2 Elevated ALT activity 11 5 4 1 13 6 Skin hyperkeratosis 10 0 39 3 14 1 Dry skin 8 0 28 0 14 0 Rash 15 1 43 5 23 1 Palmoplantar keratoderma 7 < 1 24 1 9 0 Palmar-plantar erythrodysesthesia 4 < 1 47 11 7 0 syndrome Fever 17 0 16 0 18 4 Left ventricle malfunctions 6 1 3 1 8 2

GGTP — gamma-glutamyl transpeptidase; ALT — alanine aminotransferase

Table 4. Phase III studies with BRAF or MEK inhibitors alone or in combination in the treatment of advanced melanoma

Authors Long et al. 2014 [8] Robert et al. Larkin et al. 2014 [9] Dummer et al. 2018 Long et al. 2017 [1] 2015 [5] Ascierto et al. 2016 [6] [14, 15] Drug Dabrafenib Dabrafenib Vemura- Dabrafenib Vemura- Vemurafenib Encorafenib Encorafenib + trametinib fenib + trametinib fenib + cobimetinib + binimetinib + binimetinib COMBO 450 COMBO 300 ORR (%) 53 69 51 64 50 70 63 66 Median PFS 8.8 11 7.3 12.6 7.2 12.3 14.9 12.9 (months) Median OS 18.7 25.1 18.0 25.6 17 22.3 33.6 (months) 2-/3-year OS 43/32% 52/44% 39/31% 53/45% 2-year OS rate: rate 57.6%

ORR — overall response rate; PFS — progression-free survival; OS — overall survival

Conclusions combinations of targeted therapies in terms of both PFS and OS. Table 4 summarises the results of clinical trials Encorafenib with binimetinib is already the third with various BRAF/MEK inhibitors. Undoubtedly, it registered combination of BRAF/MEK inhibitors. Re- is difficult to directly compare the survival of partici- sults of a phase III study showed very good tolerance of pants in these clinical studies, so a randomised clinical this treatment and the best survival among all available trial is needed. The better results of treatment with

118 Katarzyna Kozak, Piotr Rutkowski, Why do we need a new BRAF-MEK inhibitor combination in melanoma? encorafenib and binimetinib can be explained by slightly 3 trial. Lancet Oncol. 2016; 17(9): 1248–1260, doi: 10.1016/S1470- -2045(16)30122-X, indexed in Pubmed: 27480103. different patient populations (e.g. a lower percentage 7. Robert C, Schachter J, Long GV, et al. KEYNOTE-006 investiga- of patients with elevated lactate dehydrogenase activity) tors. Pembrolizumab versus ipilimumab in advanced melanoma. N Engl J Med. 2015; 372(26): 2521–2532, doi: 10.1056/NEJ- or better access to immunotherapy in subsequent treat- Moa1503093, indexed in Pubmed: 25891173. ment lines. On the other hand, median PFS and OS in 8. Long GV, Stroyakovskiy D, Gogas H, et al. Combined BRAF and MEK patients treated with vemurafenib in the COLUMBUS inhibition versus BRAF inhibition alone in melanoma. N Engl J Med. 2014; 371(20): 1877–1888, doi: 10.1056/NEJMoa1406037, indexed study are very close to those observed in the coBRIM in Pubmed: 25265492. or COMBI-v studies. Higher efficacy of therapy can 9. Larkin J, Ascierto P, Dréno B, et al. Combined vemurafenib and cobimetinib in BRAF-mutated melanoma. N Engl J Med. 2014; 371(20): therefore result simply from better pharmacological 1867–1876, doi: 10.1056/nejmoa1408868. properties of encorafenib. In conclusion, a combination 10. Delord JP, Robert C, Nyakas M, et al. Phase i dose-escalation and -expansion study of the BRAF inhibitor encorafenib (LGX818) in metastatic of encorafenib with binimetinib is a valuable distinctive BRAF-mutant melanoma. Clin Cancer Res. 2017; 23(18): 5339–5348, alternative to other drug combinations. doi: 10.1158/1078-0432.CCR-16-2923, indexed in Pubmed: 28611198. 11. Adelmann CH, Ching G, Du L, et al. Comparative profiles of BRAF inhibitors: the paradox index as a predictor of clinical toxicity. Onco- target. 2016; 7(21): 30453–30460, doi: 10.18632/oncotarget.8351, indexed in Pubmed: 27028853. References 12. Dummer R, Schadendorf D, Ascierto PA, et al. Binimetinib versus dacarbazine in patients with advanced NRAS-mutant melanoma (NEMO): 1. Long GV, Flaherty KT, Stroyakovskiy D, et al. Dabrafenib plus trametinib a multicentre, open-label, randomised, phase 3 trial. Lancet Oncol. versus dabrafenib monotherapy in patients with metastatic BRAF 2017; 18(4): 435–445, doi: 10.1016/S1470-2045(17)30180-8, indexed V600E/K-mutant melanoma: long-term survival and safety analysis of in Pubmed: 28284557. a phase 3 study. Ann Oncol. 2017; 28(7): 1631–1639, doi: 10.1093/an- 13. Sullivan RJ, Weber JS, Patel SP, et al. A phase Ib/II study of BRAF inhibi- nonc/mdx176, indexed in Pubmed: 28475671. tor (BRAFi) encorafenib (ENCO) plus MEK inhibitor (MEKi) binimetinib 2. Long GV, Stroyakovskiy D, Gogas H, et al. Dabrafenib and trametinib (BINI) in cutaneous melanoma patients naive to BRAFi treatment. versus dabrafenib and placebo for Val600 BRAF-mutant melanoma: J Clin Oncol. 2015; 33(15 (Suppl)): 9007, doi: 10.1200/jco.2015.33.15_ a multicentre, double-blind, phase 3 randomised controlled trial. Lan- suppl.9007. cet. 2015; 386(9992): 444–451, doi: 10.1016/S0140-6736(15)60898-4, 14. Dummer R, Ascierto PA, Gogas HJ, et al. Encorafenib plus binimetinib indexed in Pubmed: 26037941. versus vemurafenib or encorafenib in patients with BRAF-mutant 3. Larkin J, Chiarion-Sileni V, Gonzalez R, et al. Combined nivolumab and melanoma (COLUMBUS): a multicentre, open-label, randomised ipilimumab or monotherapy in untreated melanoma. N Engl J Med. phase 3 trial. Lancet Oncol. 2018; 19(5): 603–615, doi: 10.1016/S1470- 2015; 373: 23–34, doi: 10.1056/NEJMoa1504030. 2045(18)30142-6, indexed in Pubmed: 29573941. 4. Wolchok JD, Rollin L, Larkin J. Nivolumab and ipilimumab in ad- 15. Dummer R, Ascierto PA, Gogas HJ, et al. Overall survival in patients with vanced melanoma. N Engl J Med. 2017; 377(25): 2503–2504, doi: BRAF-mutant melanoma receiving encorafenib plus binimetinib versus 10.1056/NEJMc1714339. vemurafenib or encorafenib (COLUMBUS): a multicentre, open-label, 5. Robert C, Karaszewska B, Schachter J, et al. Improved overall sur- randomised, phase 3 trial. Lancet Oncol. 2018; 19(10): 1315–1327, vival in melanoma with combined dabrafenib and trametinib. N Engl doi: 10.1016/S1470-2045(18)30497-2, indexed in Pubmed: 30219628. J Med. 2015; 372(1): 30–39, doi: 10.1056/NEJMoa1412690, indexed 16. Dummer R, Ascierto PA, Gogas HJ, et al. Results of COLUMBUS Part in Pubmed: 25399551. 2: A phase 3 trial of encorafenib plus binimetinib versus encorafenib 6. Ascierto PA, McArthur GA, Dréno B, et al. Cobimetinib combined with in BRAF-mutant melanoma. ESMO Congress 2017 September 2017; vemurafenib in advanced BRAF(V600)-mutant melanoma (coBRIM): Madrid, Spain; 2017.First report on part 2 of the COLUMBUS study updated efficacy results from a randomised, double-blind, phase investigating the combination of encorafenib and binimetinib.

119 CASE REPORT

Małgorzata Flis1, Paweł Krawczyk2, Izabella Drogoń1, Katarzyna Kurek1, Robert Kieszko2, Janusz Milanowski2 1Independent Public Teaching Hospital No 4 , Lublin, Poland 2Department of Pneumonology, Oncology and Allergology, Medical University of Lublin, Poland

The effectiveness of chemotherapy in small cell lung cancer patients with BRCA2 gene mutation and Schwartz-Bartter syndrome

Address for correspondence: Lek. Małgorzata Flis ABSTRACT Oddział Chorób Płuc i Gruźlicy Small cell lung cancer (SCLC) currently comprises 15–20% of all lung cancers. It is characterised by rapid growth Samodzielny Publiczny Szpital Kliniczny and early appearance of distant metastases. It is closely related to smoking. A characteristic feature of this type of cancer is the frequent coexistence of paraneoplastic syndromes (about 50% of patients). Paraneoplastic nr 4 w Lublinie syndromes are clinically important because they can be the first sign of cancer. Early diagnosis of disturbing ul. Jaczewskiego 8, 20–954 Lublin symptoms is an important factor in increasing the effectiveness of treatment and the patient’s chance for longer Phone: 507 824 113 survival. The most frequent and important paraneoplastic syndromes in the course of SCLC are primarily the e-mail: [email protected] syndrome of inappropriate antidiuretic hormone hypersecretion (SIADH), paraneoplastic cerebellar degeneration,

Oncology in Clinical Practice and Lambert-Eaton syndrome. This paper presents the case of a patient who came to the hospital with symptoms 2019, Vol. 15, No. 2, 120–123 of hyponatraemia. Looking for the causes of hyponatraemia, the syndrome of abnormal secretion of antidiuretic DOI: 10.5603/OCP.2019.0008 hormone (Schwartz-Bartter syndrome) in the course of small cell lung cancer was confirmed. The fact that the Copyright © 2019 Via Medica patient was genetically burdened with the family history of breast cancer and was a carrier of the BRCA2 gene ISSN 2450–1654 mutation was also significant. According to the latest research, mutation in the BRCA2 gene significantly affects the chemosensitivity of cancer cells, and thus increases the body’s response to treatment. The patient received chemotherapy with carboplatin and etoposide, resulting in partial remission of cancer after two treatment cycles. Key words: small cell lung cancer, BRCA2 mutation, Schwartz-Bartter syndrome

Oncol Clin Pract 2019; 15, 2: 120–123

Introduction disease. Therefore, in the majority of patients at the time of qualification for oncological treatment, there are Lung cancer is currently the most frequently di- present features of significant advanced local or distant agnosed malignant tumour in the world. There are metastases. Another factor conditioning high cancer around 1.8 million new cases of disease per year, and mortality is a relatively poor response to chemotherapy. this number is constantly increasing. In Poland, the Small cell lung cancer (SCLC) accounts for 15–20% incidence of lung cancer is estimated at approximately of lung cancer cases, but is characterised by an extremely 23,000 new cases per year. Lung cancer is the leading aggressive course, the presence of metastases at the cause of cancer deaths among men and women around time of diagnosis, and the lack of surgical treatment the world. High mortality is caused by, among others, options. The basic method of treatment of patients with late diagnosis of the cancer process. Lung cancer does small cell lung cancer is chemoradiotherapy or chemo- not show characteristic early-stage symptoms, and no therapy. Significant antitumour activity is demonstrated effective screening methods have been developed so by alkylating drugs (cisplatin, carboplatin), anthra- far that would reduce the number of deaths due to this cyclines (doxorubicin, epirubicin), topoisomerase in-

120 Małgorzata Flis et al., The effectiveness of chemotherapy in small cell lung cancer patients with BRCA2 gene mutation and Schwartz-Bartter syndrome hibitors (etoposide, topotecan), followed by antimitotic drugs (vincristine, paclitaxel) and some antimetabolites (methotrexate). Higher efficacy of therapy was noted after the application of multidrug regimens. Response to treatment and its duration is still unsatisfactory. Median survival in patients with limited stage of small cell lung cancer (stages I–IIIA) is 14–20 months, and in patients in the generalised stage (stages IIIB–IV) it is 9–11 months. The search for new factors that influence the effectiveness of chemotherapy is ongoing [1–3]. Due to the development of genetic tests, it was observed that the presence of mutations in the BRCA2 gene, in addition to an increased risk of breast and ovarian cancer, significantly contributes to the increase of chemosensitivity [4–6]. The subject of the study is the case of a 67-year-old patient with small cell lung cancer with BRCA2 mutation and Schwartz-Bartter Figure 1. Chest review picture taken in March 2018. syndrome, in which a significant regression of advanced No irregularities were found SCLC was obtained as a result of carboplatin chemotherapy with etoposide.

Case report

The 67-year-old patient had been under the care of the oncological surgery clinic for many years due to the presence of breast cancer in the family (her mother was ill). At one of the follow-up visits, the need for genetic testing for the possible presence of mutations in the BRCA1 or BRCA2 genes was suggested. In December 2016, confirmation of the presence of the N372H mutation in the BRCA2 gene was obtained. The patient was informed about the increased risk of cancer, followed the doctor’s instructions, and set the time limits for periodic examinations. In March 2018, the patient came to the hospital because of severe weakness, nausea, and dizziness, which had been intensifying for several days. After the basic tests, hyponatraemia was found. The serum sodium level was 116 mmol/l. No abnormalities were observed Figure 2. Infiltrative changes of the right lung (April 2018) in the X-ray of the chest (Fig. 1). After symptomatic treatment, the patient was referred for further ambulatory diagnosis. of Schwartz-Bartter syndrome, or syndrome of inad- Two weeks later, the diagnostics of electrolyte equate secretion of vasopressin, without any under- disturbances began. During hospitalisation, moderate lying cause, should always arouse oncological alert- hyponatraemia was observed despite the treatment ness, especially in smokers. This patient had smoked being used. Adrenal insufficiency and inadequate cigarettes for over 30 years. Paraneoplastic syndromes substitution of thyroid hormones was excluded (the sometimes precede the symptoms of the cancer that patient suffered from post-operative hypothyroidism). they accompany. That was also the case with this pa- The study revealed reduced osmolality of the plasma tient. The diagnosis of computed tomography of the with normal osmolality of urine, and the urinary ex- neck, chest, and abdomen was extended, which re- cretion of sodium remained > 30 mmol/l. In addition, vealed the presence of infiltrative changes in the upper normal daily diuresis, and decreased urea and uric lobe of the right lung (Fig. 2) as well as the presence acid levels were found. The overall clinical picture of a mediastinal lymph node with 20–30-mm disinte- indicated a syndrome of inappropriate antidiuretic gration with right bronchus pressure, causing a sig- hormone hypersecretion (SIADH). The occurrence nificant obstruction of the high-lobe bronchi (Fig. 3).

121 Oncology in clinical practice 2019, Vol. 15, No 2

Figure 3. Lymph node pack with disintegration (April 2018) Figure 4. Regression of infiltrative changes of the right lung (June 2018)

In addition, the presence of the node of supra and subclavian nodes was revealed. Oncological diagnostics of the disclosed changes has begun. Bronchofiberoscopy and endobronchial ultra- sound — thin needle aspiration (EBUS-TNA) — were performed, collecting a fragment of bronchi and lymph nodes for pathomorphological examination. From the collected material a diagnosis of small cell carcinoma with expression of Ki-67 in 90% of tumour cells (high mitotic capacity) was obtained. Due to the advanced degree of SCLC, the patient was qualified for chemotherapy. The need to reduce the amount of fluids taken (SIADH syndrome) led to the selection of chemotherapy with carboplatin (instead of cisplatin) and etoposide. The patient tolerated the treatment well (no side effects were reported). During oncological treatment, the patient required periodic sodium supplementation. After two treatment cycles, the response to treatment was assessed. In computed tomography, a significant regression of pathological lymph nodes of the Figure 5. Regression of mediastinal node changes (June 2018) mediastinum was observed (maximum size up to 13 mm within the lymph node bundle of the right cavity, the size of the remaining nodes did not exceed the limit values, however, that bone metastases are often unnoticed Fig. 4). In addition, there was almost complete regression in radiological descriptions because of their initially of pathological foci in the upper right lobe (only residual osteolytic nature. Only the action of chemotherapy or lesions with fibrotic features were visualised, Fig. 5). There molecular-targeted therapies leads to osteosclerotic were no metastases to the central nervous system. outbreaks visible in computed tomography. Anxiety was, however, caused by two osteosclerotic Oncological treatment continued. The patient lesions visible in the thoracic vertebrae 3 and 11, not received another two cycles of chemotherapy with car- described in the previous study. Bone scintigraphy was boplatin and etoposide. Due to CTC (common toxicity performed, which revealed increased bone metabolism criteria) and CTC grade III thrombocytopaenia after within the thoracic vertebra of the third, which may the fourth cycle of chemotherapy, filgrastim injections, raise suspicion of a metastatic focus. It should be noted, fludrocortisone, and antibiotic cover were used.

122 Małgorzata Flis et al., The effectiveness of chemotherapy in small cell lung cancer patients with BRCA2 gene mutation and Schwartz-Bartter syndrome

An imaging assessment was made after the fourth cy- chemotherapy with platinum compounds. The occur- cle of treatment. In computed tomography, the changes rence of BRCA2 mutation was associated with high were stationary and were comparable with the previous sensitivity to chemotherapy. Response to treatment study. Stability of the disease was obtained. was reported in 100% of patients with BRCA2 mutation, 82% of patients with BRCA1 mutation, and 80% of patients with mutations in these genes. Median Summary progression-free survival was 18 months in carriers of the BRCA2 mutation, 11.7 months in patients with BRCA2 is a suppressor gene that acts as a negative BRCA1 mutation, and 12.5 months in patients without regulator of the cell cycle and has an effect on maintain- BRCA gene mutations (statistically significant differ- ing the stability of genetic material. The BRCA2 gene ences) [4]. encodes the BRCA2 protein, whose main function is to The above observations can be justified by the participate in DNA repair. BRCA proteins (the most increased efficacy of cell chemotherapy with higher important of them is the BRCA1 protein) together with DNA instability and impaired remedial function. Most the protein kinase associated with ataxia-telangiectasia cytostatics work by inhibiting the effect of mitosis (cell and Rad family proteins are involved in the repair of division) mainly through DNA damage and structures single-strand DNA damage (caused by the formation of responsible for cell division. DNA damage ultimately platinum compounds adducts to DNA) and damage to drives cells to the path of apoptosis. In the present case, both DNA strands (caused inter alia by radiotherapy). the presence of the BRCA2 gene mutation, in addition to If immediate DNA repair is not possible due to the the obvious effect on tumour development, appears to extent of damage, BRCA proteins participate in the have a simultaneous effect on the efficacy of cytostatic activation of cell cycle checkpoint kinase 1/2, which drugs through a synergistic effect on cell cycle disorder results in cell cycle arrest and prolongation of the time and DNA repair leading to cancer cell damage. necessary for DNA repair. This effect is enhanced by the involvement of BRCA proteins in the process of All authors have read and approved the final one activating transcription of other genes related to DNA working version to be submitted. Małgorzata Flis was the repair and cell cycle control and cell direction on the main co-creator in writing the manuscript. Malgorzata pathway of apoptosis [4–6]. Flis, Izabella Drogoń and Katarzyna Kurek participated Accordingly, cells with a mutation causing loss of in collecting and interpreting data. Paweł Krawczyk, Ro- BRCA1 or BRCA2 protein activity have increased sen- bert Kieszko and Janusz Milanowski contributed to data sitivity to DNA damaging agents, e.g. ionising radiation analysis and interpretation. or cytostatic drugs that cause DNA strand breaks. The basic DNA repair process is stopped, and the cell enters the alternative DNA repair pathways [4–6]. References Previous observations and studies reveal the dual nature of the BRCA2 gene mutation. On the one hand, 1. Tyczynski JE, Bray F, Parkin DM. Lung cancer in Europe in 2000: epidemiology, prevention, and early detection. Lancet Oncol. 2003; mutations in the BRCA2 gene significantly increase the 4(1): 45–55, doi: 10.1016/s1470-2045(03)00960-4, indexed in Pub- risk of breast cancer (up to 56%) and ovarian cancer (up med: 12517539. 2. DeVita VT, Chu E. A history of cancer chemotherapy. Cancer Res. 2008; to 27%). At the same time, other studies based on the 68(21): 8643–8653, doi: 10.1158/0008-5472.CAN-07-6611, indexed in evaluation of the effectiveness of oncological treatment Pubmed: 18974103. reveal a significant effect of the presence of mutations in 3. Krzakowski M, Orłowski T, Roszkowski K, et al. Polska Grupa Raka Płuca. Drobnokomórkowy rak płuca — zalecenia diagnostyczno-tera- the BRCA2 gene on positive response to chemotherapy, peutyczne Polskiej Grupy Raka Płuca. Onkol Prakt Klin. 2007; 3: 1–7. especially with the participation of platinum compounds 4. Yang Da, Khan S, Sun Y, et al. Association of BRCA1 and BRCA2 mutations with survival, chemotherapy sensitivity, and gene mutator and other alkylating cytostatics. Yang et al. [4] analysed phenotype in patients with ovarian cancer. JAMA. 2011; 306(14): 1557– data from The Cancer Genome Atlas (TCGA) obtained –1565, doi: 10.1001/jama.2011.1456, indexed in Pubmed: 21990299. from 316 women with low-differential serous ovarian 5. Xia F, Taghian DG, DeFrank JS, et al. Deficiency of human BRCA2 leads to impaired homologous recombination but maintains normal nonhomol- cancer subjected to chemotherapy. In 35 women the ogous end joining. Proc Natl Acad Sci U S A. 2001; 98(15): 8644–8649, BRCA1 mutation was found, and in 27 women the doi: 10.1073/pnas.151253498, indexed in Pubmed: 11447276. 6. Petrucelli N, Daly MB, Feldman GL. Hereditary breast and ovarian cancer BRCA2 mutation was found, while in 219 none of the due to mutations in BRCA1 and BRCA2. Genet Med. 2010; 12(5): 245–259, two mutations was noted. The patients underwent doi: 10.1097/GIM.0b013e3181d38f2f, indexed in Pubmed: 20216074.

123 CASE REPORT

Piotr Tomczak1, 2, Zuzanna Synowiec2 1Department of Oncology, Poznan University of Medical Sciences, Poland 2Chemotherapy Department, University Hospital of Lord’s Transfiguration in Poznan, Poland

Nivolumab in the treatment of advanced renal cell carcinoma

Address for correspondence: ABSTRACT Dr n. med. Piotr Tomczak Nivolumab is a programmed death receptor-1 (PD-1) blocking antibody approved for the treatment of advanced Szpital Kliniczny Przemienienia Pańskiego and metastatic renal cell carcinoma. Treatment with nivolumab is characterised by favourable toxicity profile. The Oddział Chemioterapii occurrence of grade 3 and 4 toxicity during the therapy is low. This article describes a medical history of a patient ul. Szamarzewskiego 82/84, 60–568 Poznań with metastatic renal cell carcinoma treated with nivolumab. Phone: 61 854 90 38 Key words: renal cell carcinoma, immunotherapy, nivolumab Fax: 61 854 90 72 e-mail: [email protected] Oncol Clin Pract 2019; 15, 2: 124–126

Introduction pean Commission registered nivolumab for treatment of advanced clear cell renal cell carcinoma (ccRCC) after The occurrence rate of kidney cancer is about 2–3% failure of an earlier antiangiogenic therapy [5]. of all human malignancies. In 90% of cases renal cell carcinoma (RCC) occurs [1]. The drugs used so far in the therapy of advanced renal cell carcinoma included Case report tyrosine kinase inhibitors (TKI) and mTOR inhibitors (mammalian target of rapamycin inhibitors) [2]. A 55-year-old man was admitted to the Depart- Nivolumab is a monoclonal antibody that binds to ment of Oncology in July 2015 with the diagnosis of the programmed cell death-1 receptor on the T lym- stage IV kidney cancer. In June 2015 he underwent phocyte and blocks its connection to the PD-L1 ligand radical left-sided nephrectomy, and histopathological present on cancer cells or other cell types present in examination confirmed the diagnosis of ccRCC, G3, the tumour microenvironment [3]. As a result of this pT3b, pNx. At the time of admission to the hospital, inhibition, the activity of effector lymphocytes is not the patient was in good general condition; his perfor- inhibited, which leads to the intensification of their cy- mance status (PS) was 1 in the Eastern Cooperative totoxic effect on cancer cells. The CheckMate 025 study Oncology Group (ECOG) scale. He was treated for showed a 27% reduction in the hazard ratio (HR) of hypertension, type 2 diabetes mellitus, and Parkinson’s death (HR 0.73; p = 0.002) in the group of patients disease. Complete blood count showed mild anaemia receiving nivolumab compared to patients treated with (Hb 7.8 mmol/l). Prognosis of the patient in Memo- everolimus. Median overall survival (mOS) was 25 and rial Sloan Kettering Cancer Centre scale (MSKCC) 16. 9 months, respectively, in the studied groups [4]. In [6] was moderate. Computed tomography (CT) of the 2015 with accordance to the results of this study, the abdominal cavity revealed presence of a 48 × 33 mm Food and Drug Administration (FDA) and the Euro- lymphadenopathy in the left paraaortic region, 22 mm

124 Piotr Tomczak, Zuzanna Synowiec, Nivolumab in the treatment of advanced renal cell carcinoma foci in the right kidney, and thrombosis in the left renal first radiological assessment carried out in September vein stump. On the basis of CT imaging, an embolism 2016 progression of the disease was observed. There was in the LS10 artery was found. The patient received an increase in metastases in the liver; the largest at that anticoagulant therapy and was qualified for first-line time had a diameter of 23 mm. In addition, progression treatment with pazopanib. of the metastatic focus in the L1 vertebra with spinal The patient started the therapy on July 17, 2015. Dur- muscle infiltration and tumour penetration into the ing the first 14 days of therapy, stage II inflammation of spinal canal was observed. The second line of treatment the oral mucosa was diagnosed according to common was carried out for three months. Due to the pain, the toxicity criteria for adverse events (CTCAE) v. 4.0. In ad- patient was referred to lumbar radiotherapy and later dition, the patient reported reduced appetite, headaches, received 4 mg of zoledronic acid every four weeks. Since muscle aches, and increased blood pressure. During the September 2016 he has not been treated systemically due next 14 days of treatment, blood pressure increased even to the lack of available therapeutic options in the third more up to 220/110 mm Hg, and the patient required line of treatment. medical assistance in the hospital emergency department, In December 2016 the patient was qualified for treat- where the treatment of hypertension was modified. Pazo- ment with nivolumab under the Early Access Programme. panib therapy was stopped to normalise blood pressure, A CT examination performed before starting the patient and it was decided to return to treatment at a reduced on the third line of treatment showed further progression dose of 600 mg once a day. However, due to persistently of metastatic lesions in the liver, with the largest diameter high blood pressure values, despite the intensification of 42 mm. In addition, the scan showed reduction of spinal hypotensive treatment, it was necessary to further reduce muscle infiltration at L1 level, probably due to radio- the dose of pazopanib to 400 mg once daily. During fur- therapy (8 Gy/T). At the beginning of immunotherapy, ther therapy, the patient remained in good condition and the patient was in good general condition with well con- the blood pressure values remained normal. trolled blood pressure and sugar levels. Treatment with According to Response Evaluation Criteria in Solid nivolumab was carried out at a dose of 3 mg/kg starting Tumours (RECIST) v. 1.1, in imaging studies performed from January 4, 2017 and then from May 2018 at a con- after three months of treatment, the presence of partial stant dose of 240 mg every two weeks [7]. response (PR) was noted, which lasted for the next six During the treatment, first-degree diarrhoea epi- months. In May 2016 the progression of the disease in sodes occurred and a first-degree creatinine level in- the form of numerous new metastases in the liver (the crease was observed but did not exceed that level. These largest with a diameter of 17 mm) and a metastatic focal did not require additional interventions [8]. In the CT point in the vertebral body L1 was found. The first line examination performed after three months of treatment, of treatment lasted for 11 months. metastases in the liver were reduced to 24 mm in the At the beginning of June 2016, according to the pro- largest diameter, and in the next evaluation — to 10 mm. gram of the National Health Fund (NFZ), the patient Partial response to treatment (according to RECIST was qualified for a second line of treatment with axitinib. 1.1 criteria) has been maintained from June 2017 until During the therapy, hoarseness and second-degree diar- now (October 2018). The patient has been continuing rhoea occurred. Due to coexisting arterial hypertension, the treatment for 22 months, all the time with good increasing the drug dose was not possible, the patient tolerance and quality of life. continued therapy at a dose of 2 × 5 mg/day. At the Table 1 presents the results of all lines of treatment.

Table 1. Course of treatment of the patient

Line of treatment I II III Drug Pazopanib Axitinib Nivolumab Best response PR PD PR Treatment time, PFS (months) 11 3 > 22 The cause of ending of the treatment Progression Progression Treatment is continued Treatment interruption Yes No No Dose reduction Yes (twice) No, and without dose Does not apply increasing Adverse effects Yes Yes Yes Adverse effects 3/4 Yes No No Quality of life Quite good Good Very good

PD — progression of the disease; PR — partial response; PFS — progression-free survival

125 Oncology in clinical practice 2019, Vol. 15, No. 2

Discussion ference was statistically significant (p < 0.001). A large proportion of the objective responses obtained in the Treating the vascular endothelial growth factor CheckMate 025 study during treatment with nivolumab receptors (VEGFR) with tyrosine kinase inhibitors is as- were permanent [4]. In the presented case, partial re- sociated with the occurrence of class-specific side effects sponse with the third line of treatment was observed in such as arterial hypertension, hypothyroidism, inflam- a patient with moderate prognosis, with metastases to mation of the oral mucosa, or chemotherapy-induced the bones and liver, which lasted for 16 months. The re- acral erythema [9]. In the course of immunocompetent sults of treatment with immunotherapy are better in the treatment with PD-1 receptor inhibitors, a different pro- presented case in comparison to treatment with previous file of side effects is observed. Typical adverse reactions therapies in which tyrosine-kinase inhibitors were used. associated with anti-PD-1 treatment include pneumonia, nephritis, diarrhoea, and hypothyroidism, which require appropriate management depending on the severity [8]. Conclusions In the CheckMate025 studies, 79% of patients treated with nivolumab had adverse reactions of all de- Nivolumab is an anti-PD-1 drug with proven ef- grees, compared to 88% treated with everolimus, while fectiveness in the treatment of patients with RCC and third- and fourth-degree adverse effects were observed a favourable toxicity profile. In Poland it is currently in 19% vs. 37% of these patients, respectively. The qual- available for use in the second line of treatment of ity of life in the group of patients receiving nivolumab advanced ccRCC and is a valuable therapeutic option was also better compared to the group of patients treated in this indication. with everolimus [4]. In this case, it is worth noting that the patient remains in good general condition during immunotherapy, with minor side effects. Reference The benefit of nivolumab treatment in the Check- Mate 025 study was reported in patients in all prognostic 1. Wysocki P, Borkowski T. Nowotwory układu moczowo-płciowego. In: Krzakowski M. ed. Onkologia kliniczna, t. 2. Via Medica, Gdańsk categories. Prognosis of patients was evaluated on the 2015: 751–760. basis of the scale consisting of three factors (presence of 2. Escudier B, Porta C, Schmidinger M, et al. ESMO Guidelines Com- mittee. Renal cell carcinoma: ESMO Clinical Practice Guidelines for anaemia, hypercalcaemia, and reduced efficiency). Pa- diagnosis, treatment and follow-up. Ann Oncol. 2016; 27(Suppl 5): tients without the above-mentioned factors were a group 58–68, doi: 10.1093/annonc/mdw328, indexed in Pubmed: 27664262. 3. Krawczyk P, Wojas-Krawczyk K. Immunoterapia ukierunkowana na with favourable prognosis, with one unfavourable factor immunologiczne punkty kontroli. In: Wysocki P. ed. Immunoonkologia. — a group with moderate prognosis, and finally with ViaMedica , Gdańsk 2016: 47–57. two or three factors — a group with poor prognosis. In 4. Motzer R, Escudier B, McDermott D, et al. Nivolumab versus everolimus in advanced renal-cell carcinoma. N Engl J Med. 2015; 373(19): this study, considering the group with favourable prog- 1803–1813, doi: 10.1056/nejmoa1510665. nosis, the median overall survival (OS) for patients 5. Xu JX, Maher VE, Zhang L, et al. FDA approval summary: nivolumab in advanced renal cell carcinoma after anti-angiogenic therapy and treated with nivolumab was not achieved compared to exploratory predictive biomarker analysis. Oncologist. 2017; 22(3): 19.6 months in patients treated with everolimus (HR 311–317, doi: 10.1634/theoncologist.2016-0476, indexed in Pubmed: 28232599. 0.80). In the group with moderate prognosis the median 6. Motzer RJ, Mazumdar M, Bacik J, et al. Survival and prognostic stratifi- OS was 21.8 vs. 18.4 months (HR 0.81), and in the group cation of 670 patients with advanced renal cell carcinoma. J Clin Oncol. 1999; 17(8): 2530–2540, doi: 10.1200/JCO.1999.17.8.2530, indexed with poor prognosis 15.3 vs. 7.9 (HR 0.48) [4, 10]. in Pubmed: 10561319. Benefit of treatment was also noted among patients 7. Zhao X, Suryawanshi S, Hruska M, et al. Assessment of nivolumab bene- with metastases to the bones and liver, i.e. those belong- fit-risk profile of a 240-mg flat dose relative to a 3-mg/kg dosing regimen in patients with advanced tumors. Ann Oncol. 2017; 28(8): 2002–2008, ing to the group with worse prognosis. In patients with doi: 10.1093/annonc/mdx235, indexed in Pubmed: 28520840. liver metastases the median OS was 18.3 vs. 16.0 months 8. Brahmer J, Lacchetti C, Schneider B, et al. American Society of Clinical Oncology. Management of immune-related adverse events in patients (HR 0.81), whereas in the group with bone metastases treated with immune checkpoint inhibitor therapy: American Society it was 18.5 vs. 13.8 months (HR 0.72). In the group of of Clinical Oncology clinical practice guideline. J Clin Oncol. 2018; 36(17): 1714–1768, doi: 10.1200/jco.2017.77.6385. patients with pulmonary metastases better results were 9. Eisen T, Sternberg CN, Robert C, et al. Targeted therapies for renal cell obtained — the median OS was 25 vs. 18.7 months (HR carcinoma: review of adverse event management strategies. J Natl Cancer Inst. 2012; 104(2): 93–113, doi: 10.1093/jnci/djr511, indexed 0.72) [10]. in Pubmed: 22235142. The percentage of objective response rate (ORR) in 10. Escudier B, Sharma P, McDermott D, et al. CheckMate 025 randomized the group of patients treated with nivolumab was 25% phase 3 study: outcomes by key baseline factors and prior therapy for nivolumab versus everolimus in advanced renal cell carcinoma. compared to 5% in patients receiving everolimus; the dif- Eur Urol. 2017; 72(6): 962–971, doi: 10.1016/j.eururo.2017.02.010.

126 CASE REPORT

Aneta Lebiedzińska1, 2, Dawid Sigorski1, Maciej Michalak3, Zygmunt Kozielec4, Anna Doboszyńska2, Dariusz Zadrożny5, Paweł Różanowski1 1Clinical Department of Oncology and Immuno-Oncology, The Ministry of Interior and Administration’s Hospital with Warmia and Mazury Cancer Centre in Olsztyn, Poland 2Department of Pulmonology, Faculty of Health Sciences, University of Warmia and Mazury in Olsztyn, Poland 3Department of Radiology, Faculty of Medicine, University of Warmia and Mazury in Olsztyn, Poland 4Department of Pathomorphology, Faculty of Medicine, University of Warmia and Mazury in Olsztyn, Poland 5Department of General and Oncological Surgery, Faculty of Medicine, University of Warmia and Mazury in Olsztyn, Poland

Complete pathological remission after palliative therapy with sorafenib in hepatocellular carcinoma — case report

Address for correspondence: ABSTRACT Lek. Aneta Lebiedzińska Hepatocellular carcinoma (HCC) is the most frequent primary malignant liver cancer. The five-year overall survival Oddział Onkologii i Immunoonkologii (OS) in men diagnosed with HCC does not exceed 9%. Patients (pts) with advanced disease are treated with Szpital MSWiA w Olsztynie sorafenib (multikinase inhibitor). In randomised trials the OS advantage was within the range of three months e-mail: [email protected] for sorafenib. Stabilisation of disease was achieved in 71% of patients, and no case of CR was reported. We

Oncology in Clinical Practice present a case of 60-year-old patient with locally advanced cT3aN0M0 (stage IIIA according to seventh TNM) 2019, Vol. 15, No. 2, 127–131 bifocal HCC (12 × 10 cm and 10 × 8 cm). The diagnosis was confirmed by pathologic examination. Due to the DOI: 10.5603/OCP.2019.0010 clinical stage, palliative treatment with sorafenib was administered from January 2016 to February 2017. A clini- Translation: dr n. med. Dariusz Stencel cal partial response (cPR) enabled surgery. In May 2017, left-sided liver bisegmentomy and resection of residual Copyright © 2019 Via Medica lesion in segment 6 were performed. The pathological report revealed ypCR. Subsequently, pathology verifica- ISSN 2450–1654 tion changed the primary diagnosis to PR. In September 2017 thermoablation of lesion in segment 5 of the liver was performed. The increased AFP (alpha-fetoprotein) level at baseline was normalised during treatment. The sorafenib therapy was completed after one year. The patient remains in follow-up with no evidence of relapse. Treatment with sorafenib in the presented case enabled radical therapy, so the palliative treatment turned out to be an induction treatment. Clinical CR (especially pCR) in advanced non-operable solid tumours after systemic treatment is quite rare (3–15%), and even less common in HCC. So far, only a few cases of achievement of CR during sorafenib therapy in HCC have been described. Key words: hepatocellular carcinoma, sorafenib, complete response

Oncol Clin Pract 2019; 15, 2: 127–131

Introduction leading to increased incidence include alcoholic liver cirrhosis or caused by hepatitis C or B virus (HCV or Hepatocellular carcinoma (HCC) is the most com- HBV) infection, diabetes, obesity, exposure to aflatox- mon primary liver malignant tumour. There are more ins, genetic factors associated with congenital metabolic than 600,000 new cases per year in the world (in Poland disorders such as haemochromatosis, tyrosinaemia about 1–2 thousand cases are diagnosed annually). (type I), galactosaemia, porphyria, or alpha 1-anti Hepatocellular carcinoma is the third cause of can- trypsin deficiency. Hepatocellular carcinoma is several cer-related death [1–4]. The five-year overall survival times more frequent in male than in female patients, (OS) in male patients diagnosed with HCC does not and the median of morbidity occurs around the age of exceed 9% [5]. The most common aetiological factors 50–60 years [6, 7].

127 Oncology in clinical practice 2019, Vol. 15, No. 2

Standard chemotherapy is not used in the treatment Infection with hepatitis B and C viruses (HBV and of patients with advanced and inoperable HCC. The HCV) was excluded. The hepatic efficiency was assessed most scientific evidence concerns the use of doxoru- as Child-Pugh A (5 points), with no cirrhosis, and base- bicin, which produce remissions in approximately 10% line alpha-fetoprotein (AFP) level before treatment was of patients. Multidrug regimens do not increase the 83.88 IU/mL (N 0–5.8 IU/mL). response rate but are associated with higher toxicity [8]. The patient was qualified for first-line palliative A breakthrough in systemic treatment of HCC has been treatment with sorafenib in a daily dose of 800 mg orally. the use of sorafenib, a multi-tyrosine kinase inhibitor. From January 12, 2016 to February 6, 2017 the patient received 15 courses of treatment with good initial tolerance. Fatigue syndrome (Common Toxicity Case report Criteria [CTC] 1) and hand–foot syndrome (CTC 1) were observed. The patient did not required reduction In January 2016, a 60-year-old male patient diag- of initial sorafenib dose. nosed with HCC was admitted to the oncology depart- In subsequent imaging studies, a gradual decrease ment. The diagnosis was established on the basis of in the size of liver lesions was noted. In the follow-up core needle biopsy (CNB) of hepatic lesions described CT study in February 2017 (Fig. 2, 3), two lesions were in imaging studies performed in the course of routine visible: in segment 6 (55 × 43 mm) and in segment 3 internal diagnostics due to increased exercise intoler- (55 × 37 mm). Clinical partial remission (cPR) was ance in the last few months. described, according to the RECIST (Response Evalu- At admission, the patient was in a good general ation Criteria in Solid Tumours) classification 1.1. Nor- condition, the ECOG (Eastern Co-operative Oncology malisation of AFP level was obtained after 12 weeks of Group) performance status (PS) was 1, and the patient treatment. did not report clinically significant complaints. Concomi- Due to a very good treatment response, the patient tant diseases included well-controlled hypertension, was qualified for radical surgical treatment. In May 2017, type 2 diabetes, obesity, and mixed hyperlipidaemia. left-sided liver bisegmentomy with resection of residual Computed tomography (CT) imaging of the segments 2 and 3 and resection of liver segment 6 were chest, abdomen, and pelvis was performed in January performed. Complete remission (ypCR) was found — no 2016 (Fig. 1) — it revealed hypervascular tumours with cancer cells were detected in postoperative histological a wash-out effect and central disintegration. There were evaluation. Due to the fact that CR is an extremely rare two lesions identified: in segment 6 with transverse phenomenon in this type of cancer, and because of the dimensions 120 × 100 mm and in segments 2/3 with preoperative radiological response at the level of partial size 100 × 82 mm. No metastases outside the liver were remission, re-evaluation of histological preparations found. Clinical stage was defined as cT3aN0M0 (IIIA was performed. It was found that HCC islands were according to TNM, seventh edition). visible in the postoperative material, surrounded by fibrosis and lymphocytic infiltrates, which corresponds to partial remission (ypPR) in response to preoperative treatment rather than complete remission as described in the primary material. The follow-up CT imaging performed in June 2017 (Fig. 4) showed the state after resection of left-sided lobe segments — the presence of a tumour-resection bed in the right lobe, without features of local recurrence. In contrast, in the right lobe a new hypervascular nodule with wash-out effect was described (Fig. 5). In September 2017, thermoablation of the focal area, visible in imaging studies in segment 5 was performed. Due to the lack of lesions that can be monitored (measurable) in imaging studies and a very good effect of local treatment, the patient was considered cured and excluded from treatment within the drug program. The patient has not received systemic therapy with sorafenib since February 2017. He has been observed since then. The last CT scan was performed in September 2018 — the Figure 1. Baseline status before treatment — 04/01/2016. Tumours picture remained stable and there were no signs of in the right and left liver lobe, examination in the venous phase recurrence or dissemination of the disease. AFP level after intravenous administration of iodine contrast medium

128 Aneta Lebiedzińska et al., Complete pathological remission after palliative therapy with sorafenib in hepatocellular carcinoma

Figure 2. Status after sorafenib treatment — 02/02/2017. Shrinkage Figure 4. Status after resection of the lateral segments (2 and 3) of lesion in the left lobe, venous phase of the left lobe together with the tumour and segment 6 — 23/06/2017

Figure 3. Status after sorafenib treatment — 02/02/2017. Shrinkage Figure 5. Status after surgery — 23/06/2017. The control shows of lesion in the right lobe, venous phase the metastasis in the right lobe remained within normal limits. The patient has reached randomised SHARP study and extended the median over 12 months of disease-free survival (DFS). The use OS in patients with HCC by three months; 71% of of sorafenib in the described case allowed us to conduct patients achieved stabilisation of the disease, no CR a radical procedure, and the treatment, assumed to was reported. Only 2% of patients achieved a clinical be palliative, became a remission-inducing and ena- response at the PR level [9]. The results obtained were bling resection. confirmed in the ASIA-PACIFIC study; however, the small sample size (226 compared to 602 patients in the SHARP study) requires caution in the interpretation of Discussion its conclusions [10]. To date, only a dozen CR cases have been described Patients with inoperable HCC are qualified for in the literature in patients with HCC during sorafenib therapy with sorafenib, an inhibitor of tyrosine kinase treatment, and pCR has been found only in a few cases of membrane receptor and serine/threonine kinases in the world [11]. of the intracellular RAS/MAPK signalling pathway. In the recent years, several clinical trials have been Sorafenib is the first drug to be used in the prospective completed with drugs that can be used in HCC systemic

129 Oncology in clinical practice 2019, Vol. 15, No. 2 treatment (e.g. sunitinib, erlotinib, everolimus, bri- anti-PD-L1 (atezolizumab) in combination with beva- vanib), but none of them has yet proved its effectiveness cizumab [17]. These drugs are tested both in palliative in HCC patients [12–15]. and in adjuvant treatment. It may also be possible to In 2016, a study with regorafenib (RESORCE) was identify the predictors of treatment response in patients completed, which showed a positive effect of the drug with HCC undergoing immunotherapy. on progression-free survival (PFS) prolongation by ap- There are first reports on the impact of PD-L1 ex- proximately three months in comparison with placebo pression level, baseline AFP concentration, and CD8+ in second-line treatment [16]. T cell infiltration, which may corelate with response to The results of other studies have been published immune checkpoint‑targeted cancer immunotherapies — in reference to the first-line treatment REFLECT [20, 21]. The correlations of immune response with mi- study with lenvatinib compared to sorafenib, in further crobiota or monocyte level in peripheral blood that are treatment lines in the CELESTIAL study with cabozan- the subject of many ongoing studies are noteworthy [22]. tinib, and in the REACH-2 study with ramucirumab, Based on the aforementioned case report, the im- which proved the effect of these drugs on extended portance of the quality of histological evaluation results survival versus placebo [17]. should also be underlined. In the presented case, the In 2018, very interesting results of a retrospective pCR described in another centre was questioned and, analysis were published on the basis of sorafenib treat- after consultation, finally described as pPR. Complete ment of over 800 patients diagnosed with HCC, among pathological remissions in HCC after using sorafenib which a subgroup of 81 patients (10%) was identified should now be treated as anecdotal, and each of them receiving the drug for more than 12 months. In the group should undergo independent histological evaluation in with long-term treatment 11 patients (13.7%) achieved the reference centre. radiological PR, and another five (6.3%) patients clinical CR. Two patients underwent liver transplantation, and three others underwent resection of a primary tumour. References The authors of the study suggest that patients receiving sorafenib for over one year can have significant benefit 1. Altekruse SF, McGlynn KA, Reichman ME. Hepatocellular carcinoma incidence, mortality, and survival trends in the United States in terms of long-term survival [18]. from 1975 to 2005. J Clin Oncol. 2009; 27(9): 1485–1491, doi: To date, predictive factors of response to treatment 10.1200/JCO.2008.20.7753, indexed in Pubmed: 19224838. 2. Małkowski P, Pacholczyk M, Łagiewska B, et al. Rak wątrobowokomór- have not been identified in HCC patients receiving kowy — epidemiologia i leczenie. Przegl Epidemiol. 2006; 60: 731–740. multi-kinase inhibitors. In the majority of patients, the 3. Krawczyk M. Rak wątrobowo-komórkowy. Medical Science Review — Hepatologia. 2008; 8: 92–99. results of treatment are not satisfactory. Currently, only 4. Bosch F, Ribes J, Díaz M, et al. Primary liver cancer: Worldwide sorafenib treatment in the first line is generally available incidence and trends. Gastroenterology. 2004; 127(5): S5–S16, doi: in Poland (drug program). Treatment may be provided 10.1053/j.gastro.2004.09.011. 5. Sant M, Allemani C, Santaquilani M, et al. EUROCARE Working Group. to patients without extrahepatic metastases, in good EUROCARE-4. Survival of cancer patients diagnosed in 1995-1999. general condition, and without signs of liver failure. Due Results and commentary. Eur J Cancer. 2009; 45(6): 931–991, doi: 10.1016/j.ejca.2008.11.018, indexed in Pubmed: 19171476. to the lack of availability of subsequent treatment lines, 6. Te Sligte K, Bourass I, Sels JP, et al. Non-alcoholic steatohepatitis: it was extremely difficult to decide about discontinuation review of a growing medical problem. Eur J Intern Med. 2004; 15(1): 10–21, doi: 10.1016/j.ejim.2003.12.008, indexed in Pubmed: 15066643. of sorafenib treatment in the present case. Maintenance 7. Sorrentino P, D’Angelo S, Ferbo U, et al. Liver iron excess in patients therapy with sorafenib after surgical intervention and with hepatocellular carcinoma developed on non-alcoholic steato-hep- thermoablation is a therapeutic option; nevertheless, its atitis. J Hepatol. 2009; 50(2): 351–357, doi: 10.1016/j.jhep.2008.09.011, indexed in Pubmed: 19070395. course is so non-standard that there are no guidelines 8. Yeo W, Mok TS, Zee B, et al. A randomized phase III study of doxorubi- for further treatment or evidence of the efficacy of cin versus cisplatin/interferon alpha-2b/doxorubicin/fluorouracil (PIAF) combination chemotherapy for unresectable hepatocellular carcinoma. continuing systemic treatment. J Natl Cancer Inst. 2005; 97(20): 1532–1538, doi: 10.1093/jnci/dji315, It is noteworthy that, based on previous experience, indexed in Pubmed: 16234567. 9. Llovet JM, Ricci S, Mazzaferro V, et al. SHARP Investigators Study the results of adjuvant treatment with sorafenib in pa- Group. Sorafenib in advanced hepatocellular carcinoma. N Engl tients who underwent radical surgical treatment did not J Med. 2008; 359(4): 378–390, doi: 10.1056/NEJMoa0708857, indexed improve DFS or OS [19]. At present, the patient has in Pubmed: 18650514. 10. Cheng AL, Kang YK, Chen Z, et al. Efficacy and safety of sorafenib in been under observation for almost 1.5 years without patients in the Asia-Pacific region with advanced hepatocellular carci- the need to receive treatment — he lives in a sense of noma: a phase III randomised, double-blind, placebo-controlled trial. Lancet Oncol. 2009; 10(1): 25–34, doi: 10.1016/S1470-2045(08)70285- health and has returned to normal life and social roles. 7, indexed in Pubmed: 19095497. New molecular targets that might be useful in the 11. Pinter M, Sieghart W. Long-term remission in advanced stage hepatocellular carcinoma? A chance for cure? Memo. 2018; 11(3): 185–192, treatment of HCC patients are still being sought. The doi: 10.1007/s12254-018-0431-z, indexed in Pubmed: 30220924. results were published of studies with anti-PD-1 (nivolu 12. Cheng AL, Kang YK, Lin DY, et al. Sunitinib versus sorafenib in ad- mab, pembrolizumab), anti-PD-L1 (durvalumab), vanced hepatocellular cancer: results of a randomized phase III trial. J Clin Oncol. 2013; 31(32): 4067–4075, doi: 10.1200/JCO.2012.45.8372, and anti-CTLA-4 (tremelimumab) drugs as well as indexed in Pubmed: 24081937.

130 Aneta Lebiedzińska et al., Complete pathological remission after palliative therapy with sorafenib in hepatocellular carcinoma

13. Zhu AX, Rosmorduc O, Evans TR, et al. SEARCH: a phase III, randomized, 18. Sacco R, Granito A, Bargellini I, et al. Clinical outcomes with long-term double-blind, placebo-controlled trial of sorafenib plus erlotinib in patients sorafenib treatment of patients with hepatocellular carcinoma: a mul- with advanced hepatocellular carcinoma. J Clin Oncol. 2015; 33(6): ticenter real-life study. Future Oncol. 2018 [Epub ahead of print], doi: 559–566, doi: 10.1200/JCO.2013.53.7746, indexed in Pubmed: 25547503. 10.2217/fon-2018-0281, indexed in Pubmed: 30091371. 14. Zhu AX, Kudo M, Assenat E, et al. Effect of everolimus on survival 19. Bruix J, Takayama T, Mazzaferro V, et al. STORM investigators. Adju- in advanced hepatocellular carcinoma after failure of sorafenib: the vant sorafenib for hepatocellular carcinoma after resection or ablation EVOLVE-1 randomized clinical trial. JAMA. 2014; 312(1): 57–67, doi: (STORM): a phase 3, randomised, double-blind, placebo-controlled 10.1001/jama.2014.7189, indexed in Pubmed: 25058218. trial. Lancet Oncol. 2015; 16(13): 1344–1354, doi: 10.1016/S1470- 15. Johnson PJ, Qin S, Park JW, et al. Brivanib versus sorafenib as first- 2045(15)00198-9, indexed in Pubmed: 26361969. line therapy in patients with unresectable, advanced hepatocellular 20. El-Khoueiry AB, Sangro B, Yau T, et al. Nivolumab in patients with carcinoma: results from the randomized phase III BRISK-FL study. advanced hepatocellular carcinoma (CheckMate 040): an open-la- J Clin Oncol. 2013; 31(28): 3517–3524, doi: 10.1200/JCO.2012.48.4410, bel, non-comparative, phase 1/2 dose escalation and expansion indexed in Pubmed: 23980084. trial. Lancet. 2017; 389(10088): 2492–2502, doi: 10.1016/S0140- 16. Bruix J, Qin S, Merle P, et al. RESORCE Investigators. Regorafenib 6736(17)31046-2, indexed in Pubmed: 28434648. for patients with hepatocellular carcinoma who progressed on 21. Duffy AG, Ulahannan SV, Makorova-Rusher O, et al. Tremelimumab sorafenib treatment (RESORCE): a randomised, double-blind, pla- in combination with ablation in patients with advanced hepatocel- cebo-controlled, phase 3 trial. Lancet. 2017; 389(10064): 56–66, doi: lular carcinoma. J Hepatol. 2017; 66(3): 545–551, doi: 10.1016/j. 10.1016/S0140-6736(16)32453-9, indexed in Pubmed: 27932229. jhep.2016.10.029, indexed in Pubmed: 27816492. 17. Pinter M, Peck-Radosavljevic M. Review article: systemic treatment 22. Hampton T. Gut microbes may shape response to cancer immuno- of hepatocellular carcinoma. Aliment Pharmacol Ther. 2018; 48(6): therapy. JAMA. 2018; 319(5): 430–431, doi: 10.1001/jama.2017.12857, 598–609, doi: 10.1111/apt.14913, indexed in Pubmed: 30039640. indexed in Pubmed: 29411013.

131 CASE REPORT

Kamila Kaźmierczak1, Joanna Kufel-Grabowska1, 2, Tomasz Kozłowski3, Błażej Nowakowski1, 4 1Department of Surgical, Oncological and Endoscopic Gynaecology, Greater Poland Cancer Centre, Poznan, Poland 2Department of Electroradiology, Poznan University of Medical Sciences, Poznan, Poland 3Department of Anaesthesiology and Intensive Care, Greater Poland Cancer Centre, Poznan, Poland 4Department of Cancer Pathology and Prophylaxis, Poznan University of Medical Sciences, Poznan, Poland

When to say “no” to a patient with an ovarian tumour and in poor general condition?

Address for correspondence: ABSTRACT Lek. Kamila Kaźmierczak Neuroendocrine tumours (NET) originating from the ovary are very rare, constituting about 0.52–1.7% of all Oddział Ginekologii Operacyjnej, NETs. Primary carcinoids constitute about 0.1% of ovarian tumours and 0.3% of all carcinoids. They rarely show Onkologicznej i Endoskopowej hormonal activity. They are most often diagnosed post-operatively, based on pathomorphological examination Wielkopolskie Centrum Onkologii using immunohistochemical methods. Due to the small number of cases, most information on the management ul. Garbary 15, 61–866 Poznań of patients with this diagnosis comes from retrospective studies and case reports. e-mail: [email protected] This paper presents a case report of a 63-year-old woman who was admitted to the department of surgical gynaecology with the diagnosis of a 15-cm right ovary tumour. Her general condition was poor due to severe Oncology in Clinical Practice 2019, Vol. 15, No. 2, 132–134 respiratory failure and severe tricuspid valve insufficiency. The clinical picture and the performed echocardiographic DOI: 10.5603/OCP.2019.0018 examination aroused the suspicion of carcinoid heart disease (Hedinger syndrome — a cardiological syndrome Translation: dr Elżbieta Stelmaszczyk of carcinoids). Due to the determination and cooperation of a multidisciplinary medical team, despite a very bad Copyright © 2019 Via Medica prognosis, the patient underwent surgery. Immediately after the operation, the patient’s condition was critical, but it ISSN 2450–1654 gradually improved. In the postoperative pathomorphological examination, a highly differentiated neuroendocrine tumour (grade 1 – G1) was diagnosed at stage IA according to the FIGO classification. The patient was referred to the endocrinology department, where receptor scintigraphy was performed without revealing other tumour changes. The patient did not require adjuvant therapy. Making a decision about surgical treatment of a patient in poor physical condition with a possibly reversible cause of heart failure was the right thing to do, and it allowed her to return to normal physical activity. Key words: neuroendocrine tumour, carcinoid syndrome, carcinoid heart disease, Hedinger syndrome

Oncol Clin Pract 2019; 15, 2: 132–134

Introduction and the prognosis is very good at that time with five-year survival of over 90% [4]. At more advanced stages, Neuroendocrine tumours (NET) of the ovary occur survival is much worse, and only 33% of the patients very rarely, constituting 0.52–1.72% of all NETs [1]. survive for over five years [5]. Primary carcinoids constitute about 0.1% of ovarian Neuroendocrine tumours are usually located in the tumours and 0.3% of all carcinoids [2]. Among primary digestive tract (75–85%), and more rarely pertain to neuroendocrine ovarian tumours, there are carcinoids, the respiratory system (15–25%). Most of them show large-cell carcinomas, and small-cell carcinomas of the no clinical symptoms and are diagnosed incidentally. hypercalcaemic and lung type [3]. These diseases are In 20–30% of the patients, the first symptoms are re- divided based on the FIGO classification. In over half lated to the production of hormones, which enables the of these cases, carcinoids are diagnosed at an early stage diagnosis of carcinoid syndrome. The most common

132 Kamila Kaźmierczak et al., When to say “no” to a patient with an ovarian tumour and in poor general condition? symptoms of carcinoid syndrome include flushing and to function independently and spent most of her days diarrhoea, along with symptoms of bronchospasm and in a half-sitting position. An echocardiogram revealed heart damage. These symptoms are related to the hy- severe tricuspid regurgitation, a lack of flap coaptation, persecretion of serotonin, and other active substances, and normal ventricular systolic activity. Peripheral and occur in the case of significantly advanced disease blood saturation was 80%. The general clinical picture in patients with liver metastases [6]. In the case of liver raised suspicion of carcinoid heart disease. A rise in failure due to the presence of metastases, an excess of markers was also observed, with the values being — Ca- serotonin flows into the right heart directly with venous -125 (133 u/ml) and HE4 (496 pmol/l). Bilateral renal flow. The precise pathomechanism of serotonin-induced retention was found, with ureteral widening, and the valve damage is not known — the visible effect is the presence of free fluid in the lesser pelvis. appearance of lesions mainly in the endocardium and The most likely cause of worsening right ventricular the endothelial layer of large vessels, which manifest insufficiency was pressure on the inferior vena cava as sharply delineated fibrous thickening. These lesions cause by a massive tumour of the right ovary; due to this, cause morphological, and subsequently mechanical, the patient was transferred to the oncological gynaeco damage to the valves, mainly those of the right heart, logy ward. The patient was qualified for a life-saving which leads, most often, to regurgitation (and less procedure, having been informed of the high risk of often to tricuspid stenosis). Cardiological damage is perioperative complications. Anaesthesia and perio- seen in 50% of patients with carcinoid syndrome and it perative care for this patient constituted a complex significantly worsens their prognosis due to progressive medical problem, because — as well as right ventricular right-heart failure. insufficiency — symptoms resulting from pressure on the Cardiological carcinoid syndrome (Hedinger syn- inferior vena cava were present. Before the procedure, drome) without other clinical symptoms exists unusu- a detailed risk assessment according to the NSQIP ally seldom. In the case of the presented patient with (National Surgical Quality Improvement Program) a tumour of the right ovary, venous blood flows directly was done — the risk of serious complications, includ- from the inferior vena cava, omitting the portal circula- ing death, was estimated at 22.7%. The results were tory system, due to which an excess of serotonin is not discussed during a meeting of the team of gynaecologists metabolised by the hepatocytes, causing “right heart” and anaesthesiologists, the patient and her family were damage. also informed on the above results. When making the decision to operate, the fact that the planned surgical procedure, despite a high risk of complications, was Case report the only possibility of treating the cause, and ultimately made a final diagnosis possible. The removal of the In November 2017, a 63-year-old patient with diag- pressure on the vena cava caused by the tumour was nosis of a tumour of the right ovary, with a diameter of a condition for any possible cardiosurgical treatment. 15 cm, was referred to the Gynaecology Clinic of the After the procedure, the patient’s overall condition Greater Poland Oncology Centre to be qualified for was severe — she spent 16 days in the intensive care surgical treatment. The patient was in intermediate unit, including nine days of sedation, she was intubated, overall condition, and peripheral cyanosis along with ab- ventilated mechanically — with an oxygen concentration dominal swelling and swelling of the limbs was notable. up to 65% — and also required pharmacological support The patient complained of dyspnoea on exertion, and of the cardiovascular system, as well as diuresis stimula- increased concentration of CA-125 (cancer antigen-125; tion. A gradual improvement of overall condition was 88.66 u/ml) as well as HE4 (human epididymis protein; observed. At the time of transfer to the gynaecological 159.50 pmol/l) was found. On gynaecological and vaginal ward, the patient was cardiovascularly stable, breathed sonographic examination, the presence of a right ovar- without dyspnoea even in supine position, was supported ian tumour was confirmed, with a diameter of 15 cm, with passive oxygen therapy only at times, and had which, due to its size, constituted an absolute indication a blood oxygen saturation of 95%. In the final pathologi- for surgical treatment. Sonographic examination of the cal testing, a highly-differentiated (G1) NET of the right uterus and left adnexa was insignificant. The patient was ovary was found. On the 20th day after surgery, the pa- referred to a cardiology ward for a full evaluation of the tient was transferred to the endocrinology department, cardiovascular system. Within three weeks of the gynae- where imaging was performed. An octreoscan did not cological consultation, the patient’s overall condition reveal the presence of NET focal lesions, the patient did deteriorated significantly — respiratory insufficiency, not require adjuvant treatment. The patient’s general lower body oedema, and peripheral cyanosis worsened condition improved significantly, and the symptoms of and central cyanosis appeared. The patient was unable inferior vena cava syndrome withdrew.

133 Oncology in clinical practice 2019, Vol. 15, No. 2

Discussion a chance for remission, or a prognosis improvement, while further symptomatic treatment would have ended The average lifespan of patients with cardiac car- up in a rapid death. Although symptomatic treatment cinoid syndrome, with symptoms of moderate and with diuretics, digoxin, and limiting the intake of fluids severe cardiac insufficiency (NYHA III/IV), is about and sodium may alleviate the symptoms of right-heart 11 months. Most patients die due to progressive heart failure at first, they do not improve the final progno- failure due to tricuspid valve regurgitation. The aver- sis. Removal of the primary tumour may be associated age delay in cardiosurgical treatment due to the lack with the patient’s complete recovery but will not undo of clear guidelines for invasive treatment of Hedinger the valve damage [11, 12]. The only effective treatment syndrome amounts to 24 months, which significantly of valve defects due to carcinoid syndrome is cardiosur- worsens the prognosis [7]. Currently, based on a 20-year gical treatment consisting of valve replacement surgery, observation of over 200 patients with cardiac carcinoid which decreases the symptoms and increases quality of syndrome at the Mayo Clinic in the USA, it is thought life. In the case of the presented patient, an active ap- that early surgical intervention in this group of patients proach caused significant improvement in her overall may increase their chances of survival [8]. Three-year condition, and thanks to the pathological diagnosis survival is noted amongst 31% of patients with cardiac further cardiosurgical treatment is possible. The patient carcinoid syndrome, compared to 60% for those without is currently awaiting cardiac surgery. cardiological involvement [9]. The main clinical symptoms in the patient discussed here were those of severe right-heart failure and inferior References vena cava syndrome, which made surgical intervention a high-risk path. The echocardiographic imaging could 1. Soga J, Osaka M, Yakuwa Y. Carcinoids of the ovary: an analysis of 329 reported cases. J Exp Clin Cancer Res. 2000; 19(3): 271–280, have suggested carcinoid heart disease; however, a lack indexed in Pubmed: 11144518. of general symptoms caused by an excess of serotonin 2. Talerman A. Germ cell tumor of the ovary. W: Blaunstein’s pathology of the female genital tract Kurman R (red). Springer-Verlag, New York suggested a different cause for “right heart” damage. 1993: 886–893. According to the latest guidelines by the Polish Neu- 3. Kuc-Rajca M, Dańska-Bidzińska A. Współczesne zasady leczenia nowotworów neuroendokrynnych kobiecych narządów płciowych. roendocrine Tumour Network (Polska Sieć Guzów Neu- Ginekol Pol. 2011; 82: 685–689. roendokrynnych) from 2017 regarding the diagnostics 4. Soga J. Carcinoids and their variant endocrinomas. An analysis of 11842 reported cases. J Exp Clin Cancer Res. 2003; 22(4): 517–530, and treatment of patients with neuroendocrine tumours, indexed in Pubmed: 15053292. chromogranin concentration is no longer recommended 5. Modlin IM, Lye KD, Kidd M. A 5-decade analysis of 13,715 carcinoid tumors. Cancer. 2003; 97(4): 934–959, doi: 10.1002/cncr.11105, as routine testing, and the serial assessment of chro- indexed in Pubmed: 12569593. mogranin may be useful for monitoring the course of the 6. Młodzianowski A. Zespół Hedingera — sercowe objawy zespołu rakowiaka. Choroby Serca i Naczyń. 2014; 11: 34–38. disease. In carcinoid syndrome, 5-hyrdoxyindoloacetic 7. Connolly HM, Nishimura RA, Smith HC, et al. Outcome of cardiac acid concentration is used [10]. surgery for carcinoid heart disease. J Am Coll Cardiol. 1995; 25(2): 410–416, indexed in Pubmed: 7829795. The symptoms of carcinoid syndrome were not 8. Møller JE, Pellikka PA, Bernheim AM, et al. Prognosis of carcinoid heart present in our patient, and the rise of CA-125 and disease: analysis of 200 cases over two decades. Circulation. 2005; 112(21): 3320–3327, doi: 10.1161/CIRCULATIONAHA.105.553750, HE4 markers, along with the presence of free fluid in indexed in Pubmed: 16286584. the pelvic cavity, suggested a malignant tumour typical 9. Pellikka PA, Tajik AJ, Khandheria BK, et al. Carcinoid heart disease. Clinical and echocardiographic spectrum in 74 patients. Circulation. of the ovary. Imaging tests showed no traits of dissemi- 1993; 87(4): 1188–1196, indexed in Pubmed: 7681733. 10. Zalecenia ogólne dotyczące postępowania diagnostyczno-terapeu- nation, which was an argument for taking on the risk of tycznego w nowotworach neuroendokrynnych Polskiej Sieci Guzów surgery. Considering the possibility of cardiosurgical Neuroendokrynnych 2017. 11. Caowalit N, Conolly HM, Schaff HV, et al. Carcinoid heart disease treatment after the symptoms of inferior vena cava associated with primary carcinoid heart disease. Am J Med. 1985; syndrome have subsided, it was decided that an excision 79: 339–354. 12. Robboy SJ, Norris HJ, Scully RE. Insular carcinoid primary in the ovary. of the tumour would be performed. This decision was A clinicopathologic analysis of 48 cases. Cancer. 1975; 36(2): 404–418, extremely risky, but an attempt at the procedure gave indexed in Pubmed: 1157010.

134 current literature review

Maciej Kawecki Department of Oncology and Radiotherapy, Maria Sklodowska-Curie Institute of Oncology, Warsaw

Safety and effectiveness of direct oral anticoagulants in the primary prophylaxis of venous thromboembolic disease among cancer patients initiating chemotherapy

Modern oncology relies not only on the advances in initiated chemotherapy and had intermediate or high cancer treatment, but also on the optimization of sup- risk of thromboembolic events (2 or more point in the portive care. Proper pain management, intensive nutri- Khorana scale). No screening for asymptomatic venous tional care and appropriate identification and treatment thrombotic disease was performed before treatment of adverse events results improve outcomes in the both initiation. The intervention was planned for 180 days palliative and radical setting. Venous thromboembolic in both trial arms. The trial’s primary endpoint was ob- disease, including its’ most dangerous form — pulmo- jective occurrence of venous thromboembolic event in nary embolism, has direct effect on patient’s survival 180-day observation. The primary safety endpoint was and quality of life. Thromboembolic disease affect not occurrence of major bleeding episode. Additionally, only cancer patients in the in-patient setting, who nearly safety analysis included also outcomes regarding rate always require primary prophylaxis for thromboembolic of clinically relevant non-major bleeding (CRNMB) disease, but also patients treated in the out-patient set- episodes and overall survival. The trial included 574 pa- ting, who present various risk of thromboembolic com- tients, randomized in 1:1 ratio to both arms, from all plications depending on a cancer type and the specific 1809 patients screened for eligibility. The primary risk factors (Khorana scale [1]) Depending on the num- analysis included 563 patients who received at least one ber of risk factors, Khorana scale stratify patients into dose of allocated treatment. Median treatment time low, intermediate or high risk group. Achieved result and rate of treatment discontinuation before planned may provide physician with guidance on whether specific 180 days was similar between arms. Primary endpoint patient may benefit from primary venous thromboem- occurred in 12 patients (4.2%) in the apixaban arm and bolic prophylaxis with low-molecular weight heparin 28 patients (10.2%) in the placebo arm (hazard ratio (LMWH) administered subcutaneously. Currently, no [HR] 0.41; 95% confidence interval [CI] 0.26–0.65; guidelines recommend routine venous thromboembolic p < 0.001). During active treatment period the primary prophylaxis during chemotherapy, leaving this decision endpoint occurred in 3 (1%) patients receiving apixa- to the leading physician. Unfortunately, mostly due to ban and in 20 (7.3%) patients receiving placebo. In the the way of administration, primary prophylaxis with safety analysis, major bleeding episode was detected in LMWH presents a major burden for cancer patients 10 patients (3.5%) patients in the apixaban arm and in and can be unacceptable, especially if used long-term. 5 patients (1.8%) in the placebo arm, which resulted in Considering this burden, application of direct oral anti- HR of 2.0 (95% CI 1.01–3.95; p = 0.046). Most bleeding coagulants (DOAs) instead of LMWH offers attractive episodes were mild, without critical organ bleedings or alternative. However, due to the increased risk of bleed- bleeding-related deaths. During the active treatment ing events and potential drug interactions, introduction period, major bleeding episodes were seen in 6 patients of DOAs into clinical practice had had to be preceded (2.1%) receiving apixaban and in 3 patients (1.1%) by clinical trials dedicated specifically to the cancer pa- receiving placebo. Rate of adverse events were similar tients. Currently, we can refer to the results of two trials between both arms. During observation, 35 patients assessing effectiveness of DOAs in primary prophylaxis (12.2%) in the apixaban arm and 27 patients (9.8%) of venous thromboembolic diseases. The results, as often in the placebo arm died, with 87% of deaths related to in medicine, are not fully convergent. cancer. As the primary endpoint was met, AVERT study The results of first trial were published by Carrier is clearly a positive one. It confirmed value of apixaban et al. in “The New England Journal of Medicine” on in the primary prophylaxis of venous thromboembolic the 21 of February 2019 [2]. The AVERT trial was diseases in cancer patients receiving chemotherapy. randomized, double-blinded, phase III clinical trial that The results of second trial were published by Khora- compared apixaban, administered orally at a dose of na et al. in the same issue of “The New England Journal 2.5 mg twice daily, with placebo in cancer patients who of Medicine” from the 21 of February 2019 [3]. CASSINI

Oncology in Clinical Practice 2019, Vol. 15, No. 2, 135–138. DOI: 10.5603/OCP.2019.0020, copyright © 2019 Via Medica, ISSN 2450–1654 135 Oncology in clinical practice 2018, Vol. 15, No. 2 was randomized, double-blinded, phase III trial that the rivaroxaban group and 2.0% in the placebo group; compared rivaroxaban 10 mg orally per day with placebo the difference did not reach statistical significance). in cancer patients who were initiating chemotherapy, Rates of all adverse events also did not differ between had intermediate or high risk of venous thromboem- both arms. One case of bleeding-associated death bolic disease according to Khorana scale and who had was observed in the rivaroxaban arm. Generally, the no asymptomatic venous thrombosis. The intervention CASSINI trial is a negative study, because the primary period was 180 days. Every 8 weeks participants under- endpoint was not met. Nevertheless, in contrast to the went ultrasonographic screening to exclude presence AVERT trial, no statistically significant increase of of thrombotic changes in lower extremities. Primary rivaroxaban-associated bleeding was observed, and the endpoint was composite and consisted from objective numerical outcomes achieved during active treatment occurrence of proximal deep-vein thrombosis in lower were clearly superior in the rivaroxaban arm, which extremities, symptomatic deep-vein thrombosis in up- argues in favour of rivaroxaban activity in the primary per extremities or distal deep-vein thrombosis in lower prophylaxis of venous thromboembolic disease. extremities, symptomatic or asymptomatic pulmonary The presented trials bring important data allow- embolism and death due to venous thromboembolism ing for better understanding of cancer-related venous as assessed up to 180 days from the treatment initiation. thromboembolic disease. However, it is difficult to pre- Additional analysis assessing primary endpoint dur- dict their impact on routine clinical practice. Despite the ing active treatment was pre-planned. The primary fact that according to the AVERT trial every cancer pa- safety endpoint was the occurrence of a major bleed- tient initiating chemotherapy with intermediate or high ing episode, with the rate of CRNMB as a secondary risk according to Khorana scale should receive primary safety endpoint. The trial included 1080 patients, among prophylaxis for venous thromboembolic disease with whom 49 (4.5%) were excluded due to the presence of apixaban, we must be aware of the details that hinder asymptomatic thrombosis, and 190 (17.5%) were not extrapolation of AVERT data to the general population. randomised due to other reasons. In the end, 841 pa- Firstly, the AVERT trial included only 574 patients tients who underwent randomisation (in a 1:1 ratio) from all 1809 screened for eligibility, which indicates represented the intention-to-treat population assessed significant patient selection. Secondly, despite the reduc- in efficacy analysis, and 809 patients who received treat- tion of risk of venous thromboembolic events, not even ment represented the safety-analysis population. About a numerical reduction of deaths was seen in the apixaban 43.7% of patients receiving rivaroxaban and 50.2% of arm. In contrast, a marginal trend for improved survival patients receiving placebo discontinued the intervention was seen in the placebo arm (HR 0.98–1.71). The deci- before reaching the planned 180 days (with similar rates sion regarding initiation of apixaban prophylaxis should of reasons for discontinuation and mean intervention include the fact that no impact on mortality should time of 4.3 months). The primary composite endpoint be expected. Nevertheless, we may currently reco- occurred within the 180-day observation period in 25 pa- gnise apixaban as an oral alternative to LWMH in the tients (6.0%) receiving rivaroxaban and in 37 patients primary prophylaxis of venous thromboembolic disease. (8.8%) receiving placebo (HR 0.66; 95% CI 0.40–1.09; Concurrently, independently of the negative results of p = 0.1), with nearly 39% of all events occurring after the CASSINI trial, data regarding rivaroxaban activity the treatment discontinuation. In a pre-planned analysis can be considered interesting. Rivaroxaban prophylaxis limited to the active treatment period, primary endpoint numerically reduced the risk of thromboembolic events, was noted in 11 patients (2.6%) in the rivaroxaban arm with low rates of bleeding complications. Moreover, and 27 patients (6.4%) in the placebo arm (HR 0.40; even though the CASSINI trial was too underpowered 95% CI 0.20–0.80). Lower rate of thromboembolic com- to detect differences in survival, a lower rate of deaths plications within the arterial system and visceral organs was seen among patients receiving rivaroxaban (with was also noted in the patients receiving rivaroxaban. number-needed-to-treat [NNT] of only 26). Both trials Additionally, a lower number of deaths was observed bring valuable data regarding the safety of DOAs in in the rivaroxaban arm compared to the placebo arm cancer patients, confirming their acceptable and man- (20% vs. 23.8%). This was confirmed by a pre-planned ageable toxicity profile. We currently dispose evidence composite analysis that included primary endpoint regarding DOA safety not only in the primary prophy- combined with death from all-causes, which occurred laxis of venous thromboembolic disease (AVERT and in 23.1% of patients receiving rivaroxaban compared to CASSINI trials) but also in treatment and secondary 29.5% of patients receiving placebo (HR 0.75; 95% CI prophylaxis (SELECT-D and Hokusai VTE Cancer 0.57–0.97). Major bleeding episodes were noted in eight trials). If further research provides better tools for patients (2.0%) receiving rivaroxaban and in four pa- patient selection in terms of safety, we may soon expect tients (1.0%) receiving placebo, with a HR of 1.96 (95% DOAs to fill in for LMWH as the basic anticoagulants CI 0.59–6.49). Rates of CRNMB were similar (2.7% in in oncology.

136 Maciej Kawecki, Current literature review

It is not a change; it is a revolution — novel options in the first-line treatment of clear cell renal cell carcinoma

Recent years have brought tremendous changes status. The objective response rate was also higher in in the treatment of patients with renal cell carcinoma, the pembrolizumab and axitinib arm — 59.3% (95% CI comparable only with the introduction of tyrosine kinase 54.5–63.9) as compared to 35.7% (95% CI 31.1–40.4) in inhibitors (TKIs) over a decade ago. Renal cell carci- the sunitinib arm (p < 0.001). The rates of all adverse noma, like melanoma or lung cancer, are an example events were similar in both arms — 98.4% in patients of cancers in which modern immunotherapy shows receiving combination vs. 99.5% in patients receiving greatest potential. Currently, nivolumab is an option in sunitinib. Rates of adverse events grade 3 and were, second-line treatment after TKI failure and a combina- respectively, 75.8% and 70.6%. Rates of patients who re- tion of nivolumab and ipilimumab can be considered quired treatment discontinuation reached 10.7% in the as the standard of care in the first-line treatment of combination arm and 13.9% in the sunitinib arm. Rates patients with intermediate and poor prognosis accord- of treatment-related adverse events that led to death ing to International Metastatic Renal Cell Carcinoma were 0.9% (four patients) in the pembrolizumab-axitinib Database Consortium (IMDC) criteria. Promising data group and 1.6% (seven patients) in the sunitinib group. coming from other trials assessing combinations of The toxicity profile of the pembrolizumab and axitinib immunotherapy with other molecularly driven agents combination was similar to previous studies except for suggest further changes in the field of renal cell carci- the increased incidence elevated liver enzymes of grade noma. With the recent results of two phase III trials it is 3 and higher. About 50% of patients who progressed becoming increasingly clear that we should forget about on pembrolizumab and axitinib received subsequent monotherapy in the first-line treatment of patients with treatment, as compared to 60.7% of patients who advanced clear cell renal cell carcinoma. progressed on sunitinib (including 37.6% who received The results of the first aforementioned trial, KEY- PD-1/PD-L1 inhibitors). Based on the results of KEY- NOTE-426, were published by Rini et al. in “The New NOTE-426 we can currently recognise the combination England Journal of Medicine” of 21 March 2019 [4]. of pembrolizumab with axitinib as a novel option in the KEYNOTE-426 was a randomised, non-blinded, phase first-line setting for patients with advanced clear cell III trial that compared standard first-line treatment renal cell carcinoma, irrespective of IMDC prognostic with sunitinib (50 mg orally per day for four weeks with group or PD-L1 expression. a two-week break) with an experimental combination The results of a second trial were published in the of pembrolizumab (200 mg intravenously every three same issue of “The New England Journal of Medicine” weeks) and axitinib (5 mg orally two times per day con- from 21 March 2019 by Motzer et al. [5]. JAVELIN Re- tinuously, with dose titration if applicable). The trial nal 101 was randomised, unblinded phase III trial that included patients with previously untreated advanced compared standard first-line treatment with sunitinib clear cell renal cell carcinoma and performance of at (50 mg orally per day for four weeks with a two-week least 70 according to the Karnofsky scale. The primary break) with an experimental combination of avelumab endpoint was overall survival (OS) and progression-free (10 mg/kg of bodyweight every two weeks) and axitinib survival (PFS). The presented data came from the first (5 mg orally two times per day continuously, with dose interim analysis. From 1062 screened patients, 861 were titration if applicable) in patients with advanced renal randomised in a 1:1 ratio to both trial arms. After a me- cell carcinoma with a clear cell component, who did not dian observation time of 12.8 months, the trial met its receive prior systemic treatment. The trial included pa- primary endpoint at statistical significance predicted tients with very good (ECOG 0) or good (ECOG 1) perfor first interim analysis. The rate of 12-month survival formance status irrespective of IMDC prognostic group. was 89.9% (95% CI 86.4–92.4) in the combined treat- The primary endpoint was PFS, but an amendment im- ment arm compared to 78.3% (95% CI 73.8–82.1) in plemented before data unblinding introduced two new, the sunitinib arm. Median OS was not reached in either independent primary endpoints: PFS and OS in patients of the arms, but the risk of death was 47% lower in with present expression of PD-L1. Secondary endpoints patients receiving pembrolizumab with axitinib (hazard included, among others, PFS and OS in the overall ratio for death 0.53; 95% CI 0.38–0.74; p < 0.0001). population and response rate. Altogether, 886 patients Median PFS reached 15.1 months (95% CI 12.6–17.7) were recruited and randomised in a 1:1 ratio to both in the combination group vs. 11.1 months (95% CI trial arms. A group of 560 patients with PD-L1-positive 8.7–12.5) in the sunitinib group, with HR reaching tumours constituted the primary endpoint population. 0.69 (95% CI 0.57–0.84; p < 0.001). Benefit in OS and After a median follow-up in PD-L1-positive population PFS was confirmed in all analysed subgroups, includ- of 9.9 months in combination arm and 8.4 months in su- ing all IMDC prognostic groups, and PD-L1 expression nitinib arm, statistically significant improvement in PFS

137 Oncology in clinical practice 2018, Vol. 15, No. 2 was seen in the combination arm: 13.8 months (95% CI arm. Deaths related to adverse events occurred in three 11.1 to not reached) vs. 7.2 months (95% CI 5.7–9.7) in patients receiving avelumab with axitinib and in one the sunitinib arm (stratified HR for progression or death patient receiving sunitinib. After disease progression, 0.61; 95% CI 0.47–0.79; p < 0.001). The difference in 20.8% of patients in the combination arm and 39.2% PFS was significant in all analysed subgroups. Due to the of patients in the sunitinib arm received subsequent low number of deaths in the PD-L1-positive population treatment. Most of the patients (66.7%) in the sunitinib (13.7% in the combination arm vs. 15.2% in the sunitinib arm received therapies aimed at PD-1 or PD-L1 after arm), evaluation of overall survival difference did not study discontinuation. Results of the JAVELIN Renal show statistically significant differences, but the wide 101 trial suggest that a combination of avelumab and range of confidence intervals should be noticed. Benefit axitinib provides benefit over standard treatment for in term of PFS was also seen in the general popula- patients with advanced clear cell renal cell carcinoma in tion: 13.8 months (95% CI 11.1 to not reached) in the the first line of treatment. Considering the increase in patients receiving avelumab and axitinib vs. 8.4 months overall survival seen in the KEYNOTE-426 trial, it cur- (95% CI 6.9–11.1) in the patients receiving sunitinib rently seems that pembrolizumab-based combinations (stratified HR for progression or death 0.69; 95% CI are more promising, at least until publication of further 0.56–0.84; p < 0.001). Similarly to the PD-L1-positive results of the JAVELIN Renal 101 trial. population, low death rates in the general population Both described trials are examples of revolution- (14.3% in the combination arm vs. 16.9% in the sunitinib ary changes in the first-line treatment of renal cell arm) hindered evaluation of overall survival and only carcinoma. Although standard treatment for a long showed a trend towards benefit from avelumab and time, monotherapy with TKIs is now outdated. Cur- axitinib (stratified HR for death 0.78; 95% CI 0.55–1.08; rently important clinical questions include the choice, p = 0.14). The response rate in the PD-L1-positive which patients should receive immunotherapy doublet population was 55.2% (95% CI 49.0–61.2) in the com- (nivolumab + ipilimumab) and which combination of im- bination arm vs. 25.5% (95% 20.6–30.9) in the sunitinib munotherapy with TKIs or another antiangiogenic agent. arm. Similar response rates were achieved in the general Moreover, including only advanced trials, the nearest population (51.4% vs. 25.7%, respectively). Rates of all future may provide us with at least four immunotherapy adverse events were 99.5% in patients receiving combi- combinations with proven benefit on overall survival. In nation vs. 99.3% in patients receiving sunitinib. Rates perfect conditions, the decision regarding treatment in of adverse events grade 3 and higher were, respectively, each individual is becoming more difficult. In Poland, we 71.2% and 71.5%. Adverse events that led to treatment are left with technology that should be called obsolete. discontinuation occurred in 7.6% of patients in the From a bitter, ironic perspective, one must admit that the combination arm and 13.4% of patients in the sunitinib decisions are a lot simpler if there is no choice.

References

1. Khorana AA, Kuderer NM, Culakova E, et al. Development and validation of a predictive model for chemotherapy-associated thrombosis. Blood. 2008; 111(10): 4902–4907, doi: 10.1182/blood-2007-10-116327, indexed in Pubmed: 18216292. 2. Carrier M, Abou-Nassar K, Mallick R, et al. AVERT Investigators. Apixaban to prevent venous thromboembolism in patients with cancer. N Engl J Med. 2019; 380(8): 711–719, doi: 10.1056/NEJMoa1814468, indexed in Pubmed: 30511879. 3. Khorana AA, Soff GA, Kakkar AK, et al. CASSINI Investigators. Rivaroxaban for thromboprophylaxis in high-risk ambulatory patients with cancer. N Engl J Med. 2019; 380(8): 720–728, doi: 10.1056/NEJMoa1814630, indexed in Pubmed: 30786186. 4. Rini BI, Plimack ER, Stus V, et al. KEYNOTE-426 Investigators. Pembrolizumab plus axitinib versus sunitinib for advanced renal-cell carcinoma. N Engl J Med. 2019; 380(12): 1116–1127, doi: 10.1056/NEJMoa1816714, indexed in Pubmed: 30779529. 5. Motzer RJ, Penkov K, Haanen J, et al. Avelumab plus axitinib versus sunitinib for advanced renal-cell carcinoma. N Engl J Med. 2019; 380(12): 1103–1115, doi: 10.1056/NEJMoa1816047, indexed in Pubmed: 30779531.

138