Breast Epithelium Procurement from Stereotactic Core Biopsy Washings: Flow Cytometry-Sorted Cell Count Analysis1

428 Vol. 8, 428–432, February 2002 Clinical Cancer Research

Breast Epithelium Procurement from Stereotactic Core Biopsy Washings: Flow Cytometry-sorted Cell Count Analysis1

Daniel L. Stoler,2 Carleton C. Stewart, and INTRODUCTION Paul C. Stomper Identification and characterization of breast tumor markers Departments of Experimental Pathology [D. L. S.], Flow Cytometry has important implications for breast cancer screening and di- [C. C. S.], and Diagnostic Imaging [P. C. S.], Roswell Park Cancer agnosis, for assessing prognosis, and for predicting response to Institute, Buffalo, New York 14263 individual therapies. In addition, determination of the molecular and/or genetic changes that precede or accompany the cancer can have significant effects on patient management and out- ABSTRACT come. The greatest impact may be through defining these Purpose: Molecular studies of breast lesions have been changes in early lesions that have been implicated in breast constrained by difficulties in procuring adequate tissues for tumorigenesis, such as atypia, DCIS,3 and minimally invasive analyses. Standard procedures are restricted to larger, pal- disease. However, a majority of these lesions are clinically pable masses or the use of paraffin-embedded materials, occult and detected during screening mammography. Standard precluding facile procurement of fresh specimens of early procedures for fresh tissue procurement are limited to larger, lesions. We describe a study to determine the yield and palpable masses, and the use of paraffin-embedded materials characteristics of sorted cell populations retrieved in core precludes the use of fresh tissues. Procurement of fresh epithe- needle biopsy specimen rinses from a spectrum of breast lial cell specimens from earlier lesions poses unique challenges. lesions. There will be more than 1 million breast biopsies in the Experimental Design: Cells from 114 consecutive ster- United States alone this year, including an increasing proportion of percutaneous image-guided core biopsies of screening mam- eotactic core biopsies of mammographic lesions released into mography-detected lesions (1). Stereotactic core needle biopsy saline washes were submitted for flow cytometric analysis. of mammographically detected clinically occult lesions with a For each specimen, epithelial cells were separated from single-pass, vacuum-assisted biopsy needle provides a new stromal and blood tissue based on the presence of cytokera- source of raw material for research and clinical analyses. Cells tin 8 and 18 markers. Epithelial cell yields based on patho- can be recovered from washes of these specimens without logical diagnoses of the biopsy specimen, patient age, and sacrificing any histological material or requiring any added mammographic appearance of the lesion were determined. patient intervention (2, 3). Applying flow cytometry to these cell Results: Biopsies containing malignant lesions yielded washes enables the separation of breast epithelial, stromal, and significantly higher numbers of cells than were obtained blood elements. Our initial experiences with the analysis of flow from benign lesion biopsies. Significantly greater cell counts cytometry-sorted stereotactic breast core needle biopsies have were observed from lesions from women age 50 or above been described (2, 3). We have shown that samples obtained by compared with those of younger women. Mammographic this procedure from malignant and benign tissues provide density surrounding the biopsy site, the mammographic sufficient number of cells for analysis of aneuploidy by appearance of the lesion, and the number of cores taken at flow cytometry and for measurements of allelic imbalance in the time of biopsy appeared to have little effect on the yield extracted DNA (3) of sorted cell populations. This report of epithelial cells. describes a prospective study of this technique to determine the Conclusions: We demonstrate the use of flow cytomet- yield and characteristics of sorted cell populations retrieved in ric sorting of stereotactic core needle biopsy washes from core biopsy specimen rinses from a spectrum of lesions ranging lesions spanning the spectrum of breast pathology to obtain from benign nonproliferative and proliferative lesions to early epithelial cells in sufficient numbers to meet the require- breast carcinoma. ments of a variety of molecular and genetic analyses. MATERIALS AND METHODS The study population consisted of 114 of 115 consecutive patients with clinically occult, mammographically suspicious Received 10/10/01; accepted 11/15/01. lesions who underwent stereotactic core biopsy by one mam- The costs of publication of this article were defrayed in part by the mographer (P. S.) at the Roswell Park Cancer Institute mam- payment of page charges. This article must therefore be hereby marked mography center. Following an explanation of the procedure advertisement in accordance with 18 U.S.C. Section 1734 solely to and the nature of the research, all patients but one signed the indicate this fact. 1 This work was supported by Grant CNE-89517 from the American Cancer Society. 2 To whom requests for reprints should be addressed, at Department of Experimental Pathology, 501 Cell and Virus Building, Roswell Park Cancer Institute, Buffalo, NY 14263. Phone: (716) 845-8610; Fax: 3 DCIS, ductal carcinoma in situ; FNA, fine-needle aspiration; CGH, (716) 845-8126. comparative genomic hybridization.

informed consent prior to their biopsy. Patients’ ages ranged Table 1 Pathological diagnoses of stereotactic core biopsy from 35 to 86 years (median, 56 years). specimens Stereotactic core biopsy was performed using a stereotactic Pathological diagnoses n % table with digital imaging (LoRad, Danbury, CT) and a direc- Cytokeratin-positive specimens (n ϭ 108; 95%) tional vacuum-assisted biopsy device (Biopsys Medical, Irvine, Carcinoma (n ϭ 41) CA) equipped with 14- or 11-gauge needles. This device allows Invasive 21 19 extraction of several tissue cores with a single needle pass into DCIS 20 19 ϭ the mammographic lesion, thus minimizing possible contami- Benign lesions (n 67) Atypia/LCISa 44 nation from regions of the breast outside the area of the lesion. Hyperplasia 16 15 Five to Ͼ15 cores (median, 8) per biopsy were procured for Other proliferative 16 15 each patient. Core specimens were initially placed in normal Nonproliferative 31 29 ϭ saline during the procedure and subsequently fixed in formalin Cytokeratin-negative specimens (n 6; 5%) Carcinoma and submitted for histological evaluation by Institute Depart- Invasive 1 17 ment of Pathology staff. For those lesions associated with mi- DCIS 1 17 crocalcifications, core biopsy specimens were placed on films Benign lesions 4 66 and radiographed to ensure adequate sampling (i.e., that the a LCIS, lobular carcinoma in situ. microcalcification was contained within the sample). Cells released into normal saline were sent to the Flow Cytometry Core Facility for cell sorting based on the expression of cytokeratins 8 and 18 and were sorted into cytokeratin-expressing (epithelial) according to criteria described by Page et al. (5, 6) and used in and -nonexpressing cell populations as described below. Cyto- similar analyses (7). Categories of pathological diagnoses of the logical evaluation of the core washings was not performed. benign lesions consisted of (a) atypical hyperplasia (benign Prior to sorting, cells submitted for flow cytometric anal- tissue containing a focus or foci of atypical ductal or lobular ysis were fixed in 2 ml of 70% ethanol and stored at Ϫ20°C hyperplasia); (b) hyperplasia (benign tissue containing a focus until needed. Cells were collected by centrifugation at 1500 ϫ or foci of hyperplasia); (c) other proliferative changes (adeno- g for 3 min, washed with 1 ml of PBS, and recentrifuged. sis); and (d) nonproliferative changes, including fibroadenomas. FITC-conjugated antibodies specific for cytokeratins 8 and 18 Statistical comparisons between groups were performed ␹2 Ͻ were used as recommended by the manufacturer (Becton Dick- using the test with P 0.05 being the limit of statistical inson Biosciences, San Jose, CA). The pellet was resuspended in significance (8). 20 ␮l of antibody and incubated at 20°C for 30 min, diluted with 1 ml of PBS, and incubated an additional 10 min to allow RESULTS unbound antibody to diffuse out of the cells. Cells were col- Flow cytometry-determined cell counts were obtained for lected by centrifugation at 1500 ϫ g for 3 min. The supernatant 114 consecutive core needle biopsy washings. The pathological was removed, and the pellet was resuspended in 300 ␮lofPBS diagnoses for these lesions are shown in Table 1. Pathological for sorting. diagnoses of the mammographically suspect lesions were ap- All measurements and cell sorting were performed using a proximately equally distributed among carcinomas, benign pro- FACSvantage flow cytometer (Becton Dickinson Biosciences). liferative lesions, and benign nonproliferative lesions. Nearly An air-cooled argon laser, operating at 488 nm, was used to half (49%) of the breast carcinomas were DCIS. Two-thirds (14 excite the FITC, and instrument compensation was adjusted of 21) invasive carcinomas were minimal (Յ10 mm) on path- using normal blood gated on lymphocytes stained with FITC- ological exam. conjugated CD45 or CD4PE antibodies in separate tubes. Cytokeratin-positive (epithelial) cells were detected in 108 Logarithmic amplification was used to acquire all data. To of 114 (95%) patient specimens. The six cases (5%) where core ensure day-to-day consistency of measurements, a template biopsy washings contained only cytokeratin-negative cells con- defining the position of each cell type was created. Immunoflu- sisted of two carcinomas and four benign lesions (Table 1). Only orescence of control cells was required to be Ϯ40 channels of specimens that contained cytokeratin-positive cells were in- their mean at a resolution of 1024 each time a sort was per- cluded in the subsequent analysis. The cell counts described formed. Instrument quality-assurance protocols and procedures represent the cytokeratin-positive (epithelial) cell population. have been previously reported (4). The distribution of cytokeratin-positive cell counts as well All mammograms were prospectively interpreted by one as their association with each lesion’s pathological diagnosis, mammographer (P. S.). For the purposes of this analysis, the patient age, and mammographic lesion appearance are shown in mammographic appearance of the lesions was categorized as Table 2. The median cytokeratin-positive (epithelial) cell count soft tissue masses or lesions manifested by microcalcifications for all specimens was 9,117 cells; however, cell counts of only. The mammographic density of the breast parenchyma 200,000 or more could be obtained by this technique. Ninety- surrounding the lesion was categorized as dense (Ͼ90% dense five percent (103 of 108) of the samples yielded Ͼ1,000 cyto- tissue), fatty (Ͼ90% fatty tissue), or heterogeneous (11–89% keratin-positive cells, and approximately one-third of the sam- dense tissue). ples yielded Ն20,000 cells. For the purposes of this study, malignant lesion patholog- There was a significant difference in cytokeratin-positive ical diagnoses were categorized as invasive carcinoma and cell yield between the core biopsy washes of malignant and DCIS. Benign lesion pathological diagnoses were categorized benign biopsy lesions. All carcinomas and 93% of benign le-

Table 2 Stereotactic core biopsy specimen cytokeratin-positive cell counts Pathological diagnosis Age (yrs) Mammographic appearance CKa positive cell Total Malignant Benign Յ49 Ն50 Ca2ϩ w/o Mass count greater than (n ϭ 108) (n ϭ 41) (n ϭ 67) (n ϭ 38) (n ϭ 70) mass (n ϭ 62) (n ϭ 46) 1,000 103 (95)b 41 (100) 62 (93) 34 (90) 69 (99) 59 (95) 44 (96) 2,000 86 (80) 36 (88) 50 (75) 24 (63) 62 (89) 46 (74) 40 (87) 5,000 74 (69) 32 (78) 38 (57) 15 (40) 59 (84) 38 (61) 32 (70) 10,000 53 (49) 26 (63) 23 (34) 8 (21) 45 (64) 26 (42) 23 (50) 20,000 35 (32) 20 (49) 14 (21) 6 (16) 29 (41) 18 (29) 16 (35) 50,000 18 (17) 12 (29) 5 (7) 3 (8) 15 (21) 8 (13) 9 (20) 100,000 6 (6) 4 (10) 2 (3) 1 (3) 5 (7) 3 (5) 3 (7) 150,000 4 (4) 2 (5) 2 (3) 1 (3) 3 (4) 2 (3) 2 (4) 200,000 1 (1) 1 (2) 0 (0) 0 (0) 1 (1) 0 (0) 1 (2) Median cell count 9,117 20,096 6,641 4,061 14,546 8,466 8,973 a CK, cytokeratin; Ca2ϩ, calcification; w/o, without. b Percentages in parentheses.

sions yielded at least 1,000 cells. Nearly half (49%) of all core Table 3 Association between number of cores per biopsy and core rinses of malignant lesions produced Ͼ20,000 epithelial cells wash cell counts Յ compared with only 21% of benign lesion core rinses (P Range of cytokeratin- 0.01). The median value of this latter group was 6,641 cells, Number of cores positive cell count ϳ3-fold less than the median value (20,096) of the malignant per biopsy n (%) (median) group. 5–7 26 (24) 987–195,761 (12,674) A second significant difference detected was in the cytokeratin- 8–10 51 (47) 670–217,277 (7,989) positive cell counts obtained from younger versus older patients 11–15 28 (26) 744–173,029 (7,478) Ͼ (Table 2). In contrast to previous findings using specimen mam- 15 3 (3) 1,553–9,849 (2,183) mography-guided FNA biopsies, cytokeratin-positive cell counts were significantly greater in samples from women age 50 or above. The cell counts of specimens from patients below age 50 were on Ն average nearly 4-fold less than the counts from patients 50 years rinsed was lower than when fewer cores were used (Table 3), Ͼ of age. Cytokeratin-positive cell counts 5000 were obtained from only 3 such biopsies were performed, and this result probably Ն 84% (59 of 70) of samples from women 50 years of age com- reflects normal variation when small sample numbers are Ͻ pared with 40% (15 of 38) samples from patients 50 years of age analyzed. (P Յ 0.001). Microcalcifications frequently represent the only mammographic evidence of many breast carcinomas and benign lesions. DISCUSSION We compared the yield of cells from biopsies where a mass was Standard fresh tissue procurement procedures for breast not present but microcalcifications were detected with biopsies specimens are limited to those lesions that are clinically where a mammographically detectable mass was found (Table palpable, narrowing the spectrum of acquired samples to 2). No significant difference in cytokeratin-positive cell counts relatively large invasive carcinomas and few DCIS. Non- was found between these two groups. palpable minimally invasive and in situ breast carcinoma and Two potential influences on the yield of cells rinsed from benign lesions, such as atypias, hyperplasias, and other pro- the tissue biopsies are the mammographic density of the area of liferative lesions, are not often available for research pur- the breast surrounding the core biopsy and the number of cores poses. Molecular and biochemical analyses of these early that were taken. Increased mammographic density of the breast malignancies and nonmalignant lesions hold great promise or an increased proportion of radiodense stromal tissue and for identification of biomarkers for individuals with cancer or glandular epithelium relative to radiolucent fatty tissue sur- at high risk for cancer. In this report, we have demonstrated rounding the lesion had a negligible impact on the number of that the combination of stereotactic core biopsy washing cells obtained in the rinses. The median cell counts were 9138 procurement and flow cytometric sorting provides sufficient for rinses obtained from biopsies of lesions within dense tissue, numbers of epithelial cells from lesions spanning the entire 8143 for rinses from biopsies of lesions within fatty tissue, and gamut of breast pathology to meet the requirements of a 7939 for rinses from biopsies of lesions within heterogeneous variety of experimental analyses (Table 4). Such techniques tissue. would include PCR-based methods, cytogenetic techniques, Intuitively, the core rinse cell yield would be expected to be immunohistochemistry, and Western, Southern, and Northern increased with increased number of cores taken within the area blot analyses. Furthermore, this method requires no further of the lesion. No increase was observed when 8–10 or 11–15 patient intervention and does not hinder the histological cores were extracted compared with 5–7. Although the number diagnosis made from the stereotactic core biopsy procedure. of cytokeratin-positive cells obtained when Ͼ15 cores were Research findings made from the cells, which otherwise are

Table 4 Minimum cell requirements for common The use of stereotactic core needle biopsies and flow molecular/biochemical techniques cytometry for obtaining prognostic data has been reported (13). Minimum cell However, in this case, entire cores were used to analyze DNA Techniques requirement ploidy; thus the sample could not be evaluated histologically. PCR, RT-PCR,a immunohistochemistry, 1–1,000 Studies of breast ductal lavage has demonstrated that this tech- CGH with whole genome amplification nique can retrieve cytologically and molecularly aberrant epi- Western blot 2,000–5,000 thelial cells in women at high risk for breast cancer or women 2D protein gel analysis 50,000–100,000 with diagnosed cancer or DCIS (14, 15). Lavage may comple- Southern blot, Northern blot, CGH 100,000–200,000 without whole genome amplification ment mammography screening for the detection of cancer when combined with mammography. However, ductal lavage requires a RT-PCR, reverse-transcription-PCR; 2D, two-dimensional. additional patient intervention and cost, and cells retrieved by lavage cannot be localized to a specific site in the breast, limiting association of research findings with particular lesions. routinely discarded in saline washes, can be correlated with Because multiple needle cores are obtained from the mam- the lesion type. mographic and pathological lesions, the surface or margin of The cell count data presented here for 108 breast stereo- whole or fragmented samples includes the lesion pathology as tactic core biopsy washings show that a median of Ͼ9,000 categorized as well as adjacent and intermixed epithelium and cytokeratin-positive cells can be obtained during the stereotactic stroma. It is not known whether an underlying molecular or core biopsy procedure. Approximately one-third of all biopsy genetic change would be localized to a single epithelial patho- washes yielded Ͼ20,000 cytokeratin-positive cells, and yields logical finding, the adjacent epithelium, or a field of breast of 200,000 or more cells could be achieved. Accordingly, as tissue as well. These concepts are being investigated (16, 17). indicated in Table 4, virtually all samples would be amenable to Increased cell marker specificity would require microdissection PCR techniques as well as fluorescent in situ hybridization and techniques or further refinement of flow cytometric sorting CGH with prior whole genome amplification of the DNAs to be methods. hybridized. Western blot analysis could be performed on 80% of Using randomly selected PCR markers, we have previously specimens, whereas ϳ20% of samples would provide adequate demonstrated that the DNA extracted from sorted epithelial cells material for two-dimensional gel analysis of cellular proteins, obtained from stereotactic core needle biopsy washes is suitable but only 5–10% of samples could be used in DNA and RNA for the identification of loss of heterozygosity in benign breast analyses by Southern and Northern blot, respectively, or in CGH lesions and early breast carcinomas (3). We are now exploring without prior genome amplification. the use of DNA obtained from cells by this technique for the The yield of cells from biopsies containing malignant le- analysis of allelic loss in genes that have been implicated in sions was significantly higher than that from benign lesion breast tumor progression and genomic instability. We conclude biopsies, although even for this latter group the median yield of that stereotactic core needle biopsy washes, which are otherwise cells was Ͼ6,000. Cell counts were significantly greater in routinely discarded by-products, provide adequate fresh breast samples from women Ն50 of age than in samples from younger epithelial cell samples for a considerable number of investiga- women; median counts for each group were 14,546 and 4,061, tive molecular analyses. Flow cytometry can segregate epithelial respectively. Other factors that appear to have little effect on the cells, also serving as a screening test for adequate cytokeratin- yield of cells into biopsy washes are the mammographic density positive cell counts for more efficient utilization in further surrounding the biopsy site, the mammographic appearance of molecular or genetic studies. The capability of analyses of the the lesion (calcifications or mass), and the number of cores full spectrum of benign and early malignant lesions could taken at the time of biopsy. lead to improved understanding of the nature of breast tumor The observations regarding cell count and age and mam- progression. mographic density of surrounding parenchyma differ from our reported experience with specimen mammography-guided ACKNOWLEDGMENTS FNA, which showed an increased yield of epithelial cells in We thank Janice L. Hoffman and David L. Sheedy for outstanding women of younger age and women with mammographic dense technical assistance, Karin A. Brady for excellent secretarial skills, and breast parenchyma (9). Mammographic dense tissue is com- Garth R. Anderson for valuable scientific discussions and input. posed of fibrous and glandular tissue rather than fat (10) and, in general, decreases with age (11). A prior histological study of REFERENCES normal breast stereotactic core biopsy specimens showed that 1. Liberman, L. Cost-effectiveness of percutaneous image-guided core dense or heterogeneous tissue is more likely to contain terminal breast biopsy. Semin. Breast Dis., 4: 68–77, 2001. ductal lobular units, which are thought to be a major cellular 2. Stomper, P. C., Budnick, R. M., and Stewart, C. C. Breast stereotac- compartment for the development of premalignant or malignant tic core biopsy washings. Abundant cell samples from clinically occult breast lesions (12). A fundamental difference in the FNA sam- lesions for flow cytometric analysis. Investig. Radiol., 33: 51–55, 1998. pling and core wash samples may explain the contrasting cell 3. Stomper, P. C., Stewart, C. C., and Stoler, D. L. Stereotactic core biopsy breast and blood cell by-products: a source of material for yields. Core biopsy specimens containing fatty tissue, which is molecular genetics research—initial experience. Radiology, 218: 881– more common in older women, are more likely to fragment after 885, 2001. extraction and may shed more cells into the saline than cores 4. Stewart, C. C., and Stewart, S. J. Immunophenotyping. In: J. P. containing dense tissue. Robinson, Z. Darzynkiewicz, P. N. Dean, L. G. Dressler, P. S. Rabino-

vitch, C. C. Stewart, H. J. Tanke, and L. L. Wheeless. (eds.). Current women 25–79 years old. AJR Am. J. Roentgenol., 167: 1261–1265, Protocols in Cytometry. New York: Wiley, 1997. 1996. 5. Page, D. L., and Anderson, T. J. Diagnostic Histopathology of the 12. Mansoor, S., Ip, C., and Stomper, P. C. Yield of terminal duct Breast, pp, 137–145. New York: Churchill Livingstone, 1987. lobule units in normal breast stereotactic core biopsy specimens: impli- 6. Page, D. L., Vander Zwaag, R., Rogers, L. W., Williams, L. T., cations for biomarker studies. Breast J., 6: 220–224, 2000. Walker, W. E., and Hartmann, W. H. Relation between component parts 13. Lovin, J. D., Stinton, E. B., Burke, B. J., and Reddy, V. V. of fibrocystic disease complex and breast cancer. J Natl. Cancer Inst. Sterotaxic core breast biopsy: value in providing tissue for flow cyto- (Bethesda), 61: 1055–1063, 1978. metric analysis. AJR Am. J. Roentgenol., 162: 609–612, 1994. 7. Stomper, P. C., Budnick, R. M., and Stewart, C. C. Use of specimen 14. Evron, E., Dooley, W. C., Umbricht, C. B., Rosenthal, D., Sacchi, mammography-guided FNA (fine-needle aspirates) for flow cytometric N., Gabrielson, E., Soito, A. B., Hung, D. T., Ljung, B., Davidson, N. E., multiple marker analysis and immunophenotyping in breast cancer. Cytometry, 42: 165–173, 2000. and Sukumar, S. Detection of breast cancer cells in ductal lavage fluid by methylation-specific PCR. Lancet, 357: 1335–1336, 2001. 8. Jekel, J. F., Elmore, J. G., and Katz, D. L. Bivariate analysis. In: Epidemiology, Biostatistics and Preventive Medicine, pp. 145–154. 15. Dooley, W. C., Veronesi, U., Elledge, R., O’Shaughnessy, J., Philadelphia: W. B. Saunders Company, 1996. Ljung, B., and Arias, R. Detection of premalignant breast cells by ductal 9. Stomper, P. C., Nava, M. E. R., Budnick, R. M., and Stewart, C. C. lavage. Obstet. Gynecol., 97 (Suppl 4): S2, 2001. Specimen mammography-guided fine-needle aspirates of mammo- 16. Susnik, B., Worth, A., LeRiche, J., and Palcic, B. Malignancy- graphically normal fatty and dense benign tissue: analysis of cell num- associated changes in the breast. Changes in chromatin distribution in ber and type. Breast Cancer Res. Treat., 45: 135–139, 1997. epithelial cells in normal-appearing tissue adjacent to carcinoma. Anal. 10. Page, D. L., and Winfield, A. C. The dense mammogram. AJR Quant. Cytol. Histol., 17: 62–68, 1995. Am. J. Roentgenol., 147: 487–489, 1986. 17. Deng, G., Lu, Y., Zlotnikov, G., Thor, A. D., and Smith, H. S. Loss 11. Stomper, P. C., D’Souza, D. J., DiNitto, P. A., and Arredondo, of heterozygosity in normal tissue adjacent to breast carcinomas. Sci- M. A. Analysis of parenchymal density on mammograms in 1353 ence (Wash. DC), 274: 2057–2059, 1996.

Downloaded from clincancerres.aacrjournals.org on September 23, 2021. © 2002 American Association for Cancer Research. Breast Epithelium Procurement from Stereotactic Core Biopsy Washings: Flow Cytometry-sorted Cell Count Analysis

Daniel L. Stoler, Carleton C. Stewart and Paul C. Stomper

Clin Cancer Res 2002;8:428-432.

Updated version Access the most recent version of this article at: http://clincancerres.aacrjournals.org/content/8/2/428

Cited articles This article cites 12 articles, 1 of which you can access for free at: http://clincancerres.aacrjournals.org/content/8/2/428.full#ref-list-1

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://clincancerres.aacrjournals.org/content/8/2/428. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.