Interobserver Agreement and Assay Reproducibility of Folate Receptor a Expression in Lung Adenocarcinoma

Transcription

1 Interobserver Agreement and Assay Reproducibility of Folate Receptor a Expression in Lung Adenocarcinoma A Prognostic Marker and Potential Therapeutic Target Ryan E. Bremer, PhD; Tatiana S. Scoggin, MS; Elizabeth B. Somers, BS; Daniel J. O Shannessy, PhD; David E. Tacha, PhD Context. Lung cancer is the leading cause of cancer deaths in the United States and globally. Folate-targeted drugs are among the promising new targeted therapies for lung cancer, provided predictive biomarkers can be identified for optimal patient selection. Objective. To evaluate the interobserver agreement and reproducibility of an immunohistochemistry assay for folate receptor a as a potential predictive marker for folate-targeted therapies. Design. Immunohistochemistry using anti folate receptor a antibody 26B3 was performed on formalin-fixed, paraffin-embedded tissues. The M-score, a semiquantitative measure of staining intensity and proportion of tumor cells staining, was determined for each specimen. Interobserver agreement was assessed using lung adenocarcinoma specimens stained at a single site and evaluated by 3 independent pathologists. Interinstrument reproducibility assessed 20 specimens stained by 3 different automated stainers. Interlaboratory agreement was determined on 5 specimens, repeatedly stained on each of 5 days, at 3 different study sites. Results. Folate receptor a expression was identified in 39 of 54 cases of lung adenocarcinoma (72%) and 4 of 37 cases of lung squamous cell carcinoma (11%). Agreement among 3 pathologists was found in 24 of 26 cases (92%). Interinstrument reproducibility was observed in 19 of 20 cases (95%). Agreement among 3 laboratories was found for 49 of 50 specimens (98%). Conclusions. Immunostaining of folate receptor a in lung adenocarcinomas is reproducible across staining platforms and among laboratories. Agreement among pathologists is achieved using a semiquantitative scoring method. An accurate and convenient method for determining folate receptor a expression offers a potentially invaluable tool for selecting patients for folate-targeted therapies. (Arch Pathol Lab Med. doi: /arpa OA) (Arch Pathol Lab Med. 2013;137: ; doi: /arpa OA) Lung cancer remains the leading cause of cancer deaths in Historically, the 5-year survival rates of lung cancer the world, with estimated deaths due to lung patients have been among the lowest, with reported survival cancer worldwide in In the United States, lung cancer rates of 6% to 18%. 3 However, as molecular targeted represents 28% of cancer-related deaths among men and 26% of cancer-related deaths of women, with an estimated total of therapies for lung cancer have reached the clinic, a new approach to treatment is emerging, offering the potential for new cases and deaths in 2011, exceeding the improvement in patient care and overall survival. 4 In an combined totals of colon, breast, and prostate cancers. 2 environment that already offers multiple options for targeted therapy, with more on the horizon, it will be critical to match patients to the most promising therapeutic option. In particular, diagnostic assessment of biomarkers Accepted for publication February 19, that are predictive of patient outcomes with a particular Published as an Early Online Release April 9, From the Department of Research & Development, Biocare targeted therapy will be of the utmost importance. Accurate Medical LLC, Concord, California (Drs Bremer and Tacha and Ms and specific measurement of predictive biomarkers, preferably using a familiar and robust methodology, will be a Scoggin); and the Department of Diagnostic Development, Morphotek Inc, Exton, Pennsylvania (Ms Somers and Dr O Shannessy). critical component of the treatment paradigm. Dr Tacha is a founder, chief scientific officer, and stockholder of Folate (folic acid) is an essential water-soluble vitamin, Biocare Medical, Concord, California. Dr Bremer and Ms Scoggin are employees of Biocare Medical. Ms Somers and Dr O Shannessy necessary for the biosynthesis of the purine nucleobases are employees of Morphotek, Exton, Pennsylvania. The authors have thymine and uracil and as a cofactor in certain metabolic no other relevant financial interest in the products or companies reactions. Because of its essential role in DNA synthesis and described in this article. repair mechanisms, folate may be required for the continued Portions of these data were presented as a poster at the annual growth and maintenance of malignant cells. As a result, the College of American Pathologists meeting; September 10, 2012; San Diego, California. folate pathway is an attractive objective for targeted Reprints: Ryan Bremer, PhD, Reagent Chemistry, Biocare Medical, therapies, including folate-linked chemotherapeutics and 4040 Pike Ln, Concord, CA ( rbremer@biocare.net). Arch Pathol Lab Med Vol 137, December 2013 humanized monoclonal antibodies. 5,6 Folate Receptor a in Lung Adenocarcinoma Bremer et al 1747

2 Folate receptor a (FRA) is a cell-surface glycoprotein responsible for the transport of folate across cell membranes. FRA has demonstrated restricted expression in polarized epithelial cells in a subset of normal tissues, as well as certain cancers of epithelial origin. 7 FRA expression has been identified in the majority of lung adenocarcinomas (LADCs) and serous (nonmucinous) ovarian and endometrial carcinomas, as well as a limited subset of breast cancers Analysis of messenger RNA levels in LADC has shown that a higher expression level of FRA is a positive prognostic factor for patient survival and disease-free survival, following surgical resection. 14 Conversely, increased FRA expression has been found to be a predictor of negative outcomes in ovarian, endometrial, and breast cancers. 15,16 Folate receptor a has been the target of therapeutic approaches that exploit expression in malignant lesions, including folate-conjugated chemotherapeutics and humanized anti-fra antibodies. 5,6 Currently, therapeutics of both types are in late-stage clinical development. In particular, an anti-fra monoclonal antibody, farletuzumab, is currently being evaluated as a therapeutic treatment in both non small cell LADC and ovarian cancer. 6 Immunohistochemistry of FRA In order to routinely evaluate FRA expression for clinical and research purposes, a new, highly specific mouse monoclonal anti-fra antibody, 26B3.F2, was developed and thoroughly characterized using immunohistochemistry (IHC) on formalin-fixed, paraffin-embedded tissues. 17 The 26B3 antibody identified expression of FRA in a variety of normal tissues, including pancreas, thyroid, hypophysis, breast, lung, salivary gland, kidney, and cervix. Notable malignant tissues that exhibited FRA expression included serous ovarian carcinoma, endometrial carcinoma, and the majority of cases of LADC. Importantly, IHC with 26B3 has demonstrated that FRA is a prognostic marker for patient outcomes in LADC, consistent with previous reports based on messenger RNA of FRA. 8,14,18 A semiquantitative determination of FRA expression based on staining intensity and the proportion of tumor stained (ie, M-score) was used to classify cases. 8 Receiver operating characteristic analysis identified an M-score of 10 (out of 50) as the optimal cutoff for discrimination of overall survival, with high FRA expression (M-score 10) associated with increased overall survival, and low FRA expression (M-score, 10) predictive of decreased overall survival. These results validate both the significance of FRA expression in LADC and the utility of the M-score, thus encouraging ongoing efforts to expand the use of FRA in characterizing cases of LADC. FRA as a Predictive Marker Optimal patient stratification, based on the determination of FRA expression, may be critical for the success of anti- FRA therapies currently in clinical development. Therefore, a sensitive, specific, and robust assay for determining FRA expression is required. Immunohistochemistry of formalinfixed, paraffin-embedded tissues would be the preferred method for FRA determination, because of its low cost, convenience, and routine use in clinical laboratories worldwide, provided that reproducibility across laboratories and instrument platforms can be established. Interlaboratory and interobserver agreement is imperative if FRA expression will be used as a predictive marker of efficacy for an anti-fra therapy or in assessing patient prognosis. Towards this goal, a protocol and reagents for IHC of formalin-fixed, paraffin-embedded tissues using anti-fra mouse monoclonal antibody 26B3 were developed and evaluated at 3 clinical sites and across different automated staining platforms. Initially, 30 slides stained at a single site were assessed by one pathologist from each of the 3 study sites. Each pathologist reported an M-score for each specimen, enabling direct evaluation of interobserver agreement on M-score when viewing the same stained slide. Subsequently, interlaboratory reproducibility of FRA staining was investigated by staining the same specimens at each of the 3 study sites, followed by determination of an M-score by the site pathologist. Finally, 20 cases were stained using 3 different automated staining instruments: Biocare s intellipath (Biocare Medical, Concord, California), Dako s Autostainer Plus (Dako, Carpinteria, California) and Dako s Autostainer Link, in addition to a manual staining method, and the M-scores compared for the same specimen stained by each method. MATERIALS AND METHODS Immunohistochemistry Formalin-fixed, paraffin-embedded whole tissues and tissue microarrays were analyzed by IHC. Tissue blocks were all obtained from archival sources, with no associated patient information, treatment, or follow-up data. Tissues were deparaffinized and rehydrated in the usual manner. Immunohistochemistry was performed with reagents provided by Biocare Medical (Folate Receptor Alpha IHC Assay Kit, BRI4006K), using the protocol provided in the kit instructions. Antigen retrieval was performed in a pressure cooker (958C/40 minutes; Decloaking Chamber; Biocare Medical), using a citrate-based retrieval solution (Diva Decloaker; Biocare Medical). Protein block (5 minutes) was followed by incubation with primary anti-fra 26B3 antibody (1 lg/ml), or negative control reagent (mouse immunoglobulin G1 j, 1lg/mL), for 30 minutes. Detection was performed using a secondary probe (10 minutes), followed by horseradish peroxidase conjugated polymer (10 minutes). Visualization with 3,3 0 -diaminobenzidine (5 minutes) identified both intracellular and membranous expression of FRA. Only membrane staining for FRA was considered to be positive FRA staining in the pathologist s interpretation of each specimen. Semiquantitative Scoring Membrane staining in the tumor of each stained specimen was evaluated by a site pathologist using the M-score, as previously reported. 8 Briefly, the M-score is a semiquantitative algorithm for scoring FRA expression based on the proportion of tumor cells staining at each intensity level (0, 1þ, 2þ, 3þ). The M-score is calculated as follows: M-score ¼ð3xþ2yþzÞ=6 where x is the percentage of tumor stained with intensity 3þ, y is the percentage of tumor stained with intensity 2þ, and z is the percentage of tumor stained with intensity 1þ. Observed M-scores may range from 0 (no tumor staining) to 50 (100% tumor staining at an intensity of 3þ). Sensitivity and Specificity of 26B3 in LADC and Squamous Cell Carcinoma Cases of LADC (n¼54) and lung squamous cell carcinoma (SqCC) (n ¼ 37) were stained with 26B3 using the standard protocol and reagents described above, using an intellipath stainer. Membrane staining of tumor cells was evaluated for each specimen 1748 Arch Pathol Lab Med Vol 137, December 2013 Folate Receptor a in Lung Adenocarcinoma Bremer et al

3 and determined to be positive for FRA if at least 10% of the tumor cells were stained at any intensity. Interobserver Agreement One specimen from each of 30 cases of LADC was stained with 26B3 and reviewed by 3 pathologists. Each pathologist determined an M-score for each specimen. Twenty-six of the cases were determined to be evaluable; 3 cases were found to lack sufficient tumor in the specimen, and 1 case contained a folded sample, which could not be reviewed. Agreement among pathologists was determined as described below. Interinstrument Reproducibility Twenty cases of LADC were stained using the standard protocol, by 4 different staining methods, including automated staining on Dako Autostainer, Dako PT Link 48 Autostainer, and Biocare intellipath, and a manual staining method. (Dako Target Retrieval Solution, ph 6, was used for retrieval on the Dako PT Link 18 Autostainer.) A single pathologist determined an M-score for each of the 80 slides. Interlaboratory and Interrun Reproducibility Five cases of LADC were stained on each of 5 days, at 3 independent study sites, in order to evaluate both interrun and interlaboratory reproducibility. All specimens were stained on both the Biocare intellipath and the Dako Autostainer on each of the 5 days. An independent pathologist at each study site determined an M-score for the 50 slides stained at that site (5 cases 3 5 days 3 2 instruments). Determination of Agreement For each study, agreement among observers, instruments, or repeated staining runs was determined by 2 different methods. First, using the previously established M-score 10 as the cutoff for establishing a favorable prognosis, each specimen was determined to be FRA positive if the M-score was 10 or greater, and FRA negative if the M-score was less than In this manner, the reliability of the IHC method at determining FRA expression for the purpose of predicting prognosis could be established. Additionally, in anticipation that future applications of IHC of FRA using 26B3 may stratify patients using different M-score criteria, agreement among the M-scores for each specimen was determined. A difference in M-scores of 15 was selected as an acceptable range for agreement among replicates of the same specimen in each study. M-scores that differed by more than 15 were recorded as not in agreement. In this manner, the reproducibility of M-score determination could be evaluated across the full range of possible M-scores, which may be useful for application in future studies. Overall percentage agreement was calculated in each study by dividing the number of specimens in agreement by the total number of specimens evaluated. The 95% confidence interval (CI) for the percentage agreement was calculated using the exact method of Clopper and Pearson. 19 RESULTS Sensitivity and Specificity in LADC and Lung SqCC Sensitivity and specificity of 26B3, using the established protocol and reagents, were determined using 54 cases of LADC and 37 cases of lung SqCC. The 26B3 was found to stain 39 of 54 cases of LADC (72%; 95% CI, 58% 84%) (Table 1). Significantly fewer cases of lung SqCC were positive for FRA expression with 26B3 (4 of 37; 11%; 95% CI, 3% 25%). The incidence of staining in LADC was consistent with previous reports. 8,9 However, 26B3 has been found to be more specific for LADC and to stain fewer cases of lung SqCC in this study (11%) and previously (14%) 8 than an alternative anti-fra antibody. 9 Given these differing results, studies are currently underway to further Table 1. Sensitivity of Anti-FRA Antibody 26B3 for Lung Adenocarcinoma (LADC) and Squamous Cell Carcinoma (SqCC), by Immunohistochemistry Anti-FRA 26B3 LADC Lung SqCC Positive cases/total cases 39/54 4/37 Sensitivity, % investigate staining in lung SqCC, in order to most accurately determine FRA expression levels. Immunohistochemistry with 26B3 results in membranous and intracellular staining, which varies in frequency and intensity from case to case, offering a convenient measure of FRA expression for each case (Figure 1, A through D). As expected, 26B3 also stains normal lung tissue (Figure 2). Interobserver Agreement Twenty-six cases of LADC were stained with 26B3 at a single study site and the slides were reviewed by 3 independent pathologists at 3 different sites. The M-scores determined by each pathologist for each case were evaluated for agreement. Using the FRA-positive/FRAnegative method of evaluating agreement, (ie, a specimen was determined FRA positive if the M-score is 10 or greater, and FRA negative if the M-score is less than 10; see Materials and Methods ), all 3 pathologists agreed on FRA expression in 23 of 26 (88%; 95% CI, 70% 97%) cases (Table 2). Two of the 3 instances of disagreement occurred in cases near the M-score ¼ 10 cutoff, with recorded M-scores of 9, 12, and 13 for one specimen and 5, 5, and 16 for the other specimen. In both cases, the interpretation among pathologists was consistent, with the expected differences in subjective interpretation accounting for the differing positive/negative results when FRA expression was near the cutoff value of M-score ¼ 10. The third case of disagreement arose from 1 pathologist s interpretation that 95% of the tumor in the sample exhibited 2þ staining, whereas the other 2 pathologists found 80% to 85% of the tumor sample had no staining for FRA. Direct comparison of explicit M-scores from each pathologist found agreement in 24 of 26 (92%; 95% CI, 75% 99%) cases, when an M-score range of 15 among pathologists was used as the acceptable limit for agreement (Table 2). One case of disagreement was the same case as that described above, with differing pathologist interpretation. The remaining case of disagreement exhibited more similar M-scores among pathologists (23, 28, 42); however, 1 pathologist similarly found 80% of the tumor to exhibit 3þ staining, whereas the other 2 pathologists allocated 40% to 50% of the tumor as 1þ or no staining. Interinstrument Reproducibility Folate receptor a staining of 20 cases of LADC was compared among specimens stained on 3 automated stainers (Dako Autostainer, Dako PT Link 48 Autostainer, and Biocare intellipath Autostainer), as well as a manual IHC staining method. Staining was performed at a single site and an M-score determined by a single pathologist. Seven of these cases were negative for FRA, whereas the remaining 13 cases exhibited a full range of FRA expression. Using the FRA-positive/FRA-negative method of evaluating agreement, 18 of 20 specimens (90%; 95% CI, 69% 98%) gave identical positive/negative results across all 4 staining methods (Table 2). One of the specimens lacking Arch Pathol Lab Med Vol 137, December 2013 Folate Receptor a in Lung Adenocarcinoma Bremer et al 1749

4 Figure 1. Anti folate receptor a (FRA) 26B3 staining of lung adenocarcinoma (LADC) with intensity 3þ (A), 2þ (B), 1þ (C), and LADC negative for FRA (D) (original magnifications 310 [A] and 320 [B through D]). complete agreement had M-scores near the M-score ¼ 10 cutoff (M-scores ¼ 8, 8, 8, 11). Although consistent staining was clearly observed across methods for this specimen, its FRA expression level near the M-score ¼ 10 cutoff places it in an intermediate range where a classification cannot be made with certainty. The other specimen lacking agreement exhibited a single staining failure on the Dako PT 48 Link Autostainer. In this case, M-scores by the other 3 staining methods were 35, 30, and 33, whereas the Dako PT 48 Link Autostainer specimen had an M-score of 2. Using the method of comparing all M-scores, 19 of 20 cases (95%; 95% CI ¼ 75% 99%) produced equivalent staining across all 4 staining methods when using a range of 15 as agreement in the M-scores for a given specimen among staining methods (Table 2). The only specimen lacking agreement was the same specimen as noted above, with an M-score ¼ 2 on the Dako PT 48 Link Autostainer. Figure 2. Anti folate receptor a 26B3 staining of normal lung (original Interrun and Interlaboratory Reproducibility Five cases of FRA-positive LADC were stained on each of 5 days, at 3 different study sites, and using 2 different instruments (Biocare intellipath and Dako Autostainer), in order to evaluate the reproducibility of FRA staining among runs on separate days, and, most importantly, among laboratories. Agreement among runs and among laboratories was performed using the method of direct comparisons of M-score, with a difference of 15 as the acceptable limit for agreement across M-scores for specimens stained on different days, or at different laboratories. In each case, staining the same specimen on repeated days produced highly consistent staining results. At each of magnification 310). the 3 study sites, there was 100% agreement (25 of 25; 1750 Arch Pathol Lab Med Vol 137, December 2013 Folate Receptor a in Lung Adenocarcinoma Bremer et al

5 Table 2. Interobserver Agreement and Interinstrument Reproducibility for Folate Receptor a (FRA) Staining With 26B3 Assay Kit Method for Evaluating Agreement FRA Positive, M-Score 10; Study FRA Negative, M-Score,10 Difference in M-Scores 15 Interobserver agreement, No./total (%) 23/26 (88) 24/26 (92) Interinstrument reproducibility, No./total (%) 18/20 (90) 19/20 (95) 95% CI ¼ 86% 100%) for all 5 cases among staining runs performed on the 5 different days at the site (Table 3). The same agreement was observed for specimens stained on both the Dako Autostainer and the Biocare intellipath. In order to evaluate interlaboratory reproducibility, the M- scores for each specimen were compared for each of 5 days. On all 5 days, 100% agreement was observed for all 5 specimens stained on the Biocare intellipath, among all 3 sites (25 of 25; 95% CI ¼ 86% 100%) (Table 3). Evaluation of specimens stained with the Dako Autostainer in each laboratory resulted in 100% agreement on 4 of 5 days (24 of 25; 95% CI ¼ 80% 99%). On a single day (day 2), one specimen from the Dako Autostainer differed in M-score among the 3 sites by 17 (M-score ¼ 30, 38, 47). Although this specimen was classified as not in agreement based on the study criteria, the observed M-score difference of 17 was only slightly outside the acceptable range of 15. In total, 50 comparisons were made among the 3 sites (5 specimens for 5 days, with 2 instruments). Among these specimens, 49 of 50 (98%; 95% CI ¼ 89% 99%) demonstrated agreement in M-score among laboratories. COMMENT Immunohistochemistry using anti-fra antibody 26B3 is a sensitive, specific, and reproducible method for the determination of FRA expression in LADC and lung SqCC. Use of the semiquantitative M-score as a measure of FRA levels demonstrated agreement in interpretation among pathologists and among specimens stained in different laboratories, as well as across various staining methods, during multiple days. Evaluation of FRA with 26B3 uses a convenient and routine IHC method that is familiar to laboratory personnel, and requires no further training or expertise than what is already available in anatomic pathology laboratories. Consistent staining of FRA with 26B3 among laboratories depends upon use of the same ancillary reagents (protein block, detection, chromogen) and a standard protocol; however, this is no more burdensome than what is currently performed for other predictive biomarkers. Interpretation of FRA expression depends upon the determination of percentage of the tumor staining at various intensities; although this practice is not routinely used for every specimen, pathologists are well prepared for such interpretation, without additional training. Importantly, the M-score in particular mitigates the minor differences in subjective interpretation that arise from assigning different values to each staining level among pathologists, resulting in a consistent scoring of FRA expression among observers. The reproducibility and agreement observed for FRA staining, with these protocols and reagents, supports its use as a method to routinely and conveniently evaluate FRA expression for the purposes of selecting patients for anti- FRA targeted treatments as well as for evaluating patient prognosis. Although ongoing clinical trials will evaluate the utility of FRA expression as a selection criteria for treatment, using FRA expression to assess the prognosis of LADC patients is a potentially valuable diagnostic option, currently available today, and supported by studies in the literature. 8,14,18 Anti-FRA 26B3 was found here, and previously, 8 to stain fewer cases of lung SqCC (11% and 14%, respectively) than previously reported for anti-fra antibody mab343 (51%). 9 Studies are currently underway to further investigate the staining of each of these antibodies in lung SqCC. Finally, FRA expression was found in 72% of LADC cases, using IHC with 26B3, consistent with previous reports. Importantly, this may identify a large subset of patients for whom an anti-fra targeted therapy may be valuable. The possibility of addressing the significant unmet medical need in lung cancer therapy encourages the continued development of anti-fra therapies and the diagnostic assays to identify the patients most likely to benefit from such treatments. The authors would like to thank Yao-Shi Fu, MD; Thomas Haas, DO; and Haodong Xu, MD, PhD, for determination of M-scores. References 1. Jemal A, Bray F, Center MM, Ferlay J, Ward E, Forman D. Global cancer statistics. CA Cancer J Clin. 2011;61(2): Siegel R, Ward E, Brawley O, Jemal A. Cancer statistics, 2011: the impact of eliminating socioeconomic and racial disparities on premature cancer deaths. CA Cancer J Clin. 2011;61(4): Siegel R, DeSantis C, Virgo K, et al. Cancer treatment and survivorship statistics, CA Cancer J Clin. 2012;62(4): Cagle PT, Chirieac LR. Advances in treatment of lung cancer with targeted therapy. Arch Pathol Lab Med. 2012;136(5): Low PS, Kularatne SA. Folate-targeted therapeutic and imaging agents for cancer. Curr Opin Chem Biol. 2009;13(3): Konner JA, Bell-McGuinn KM, Sabbatini P, et al. Farletuzumab, a humanized monoclonal antibody against folate receptor alpha, in epithelial ovarian cancer: a phase I study. Clin Cancer Res. 2010;16(21): Elnakat H, Ratnam M. Distribution, functionality and gene regulation of folate receptor isoforms: implications in targeted therapy. Adv Drug Deliv Rev. 2004;56(8): Table 3. Interrun Reproducibility and Interlaboratory Agreement for Folate Receptor a Staining With 26B3 Assay Kit Staining Instrument Study Biocare Medical intellipath a Dako Autostainer b Interrun reproducibility, No./total (%) 25/25 (100) 25/25 (100) Interlaboratory agreement, No./total (%) 25/25 (100) 24/25 (96) a Biocare Medical, Concord, California. b Dako, Carpinteria, California. Arch Pathol Lab Med Vol 137, December 2013 Folate Receptor a in Lung Adenocarcinoma Bremer et al 1751

6 8. O Shannessy DJ, Yu G, Smale R, et al. Folate receptor alpha expression in lung cancer: diagnostic and prognostic significance. Oncotarget. 2012;3(4): Cagle PT, Zhai QJ, Murphy L, Low PS. Folate receptor in adenocarcinoma and squamous cell carcinoma of the lung: potential target for folate-linked therapeutic agents [published online ahead of print September 14, 2012]. Arch Pathol Lab Med. 2013;137(2): Nunez MI, Behrens C, Woods D et al. High expression of folate receptor alpha in lung cancer correlates with adenocarcinoma histology and EGFR mutation. J Thorac Oncol. 2012;7(5): Toffoli G, Cernigoi C, Russo A, Gallo A, Bagnoli M, Boiocchi M. Overexpression of folate binding protein in ovarian cancers. Int J Cancer. 1997; 74(2): O Shannessy DJ, Somers EB, Maltzman J, Smale R, Fu YS. Folate receptor alpha (FRA) expression in breast cancer: identification of a new molecular subtype and association with triple negative disease. SpringerPlus. 2012;1: O Shannessy DJ, Somers EB, Smale R, Fu YS. Expression of folate receptora (FRA) in gynecologic malignancies and its relationship to the tumor type. Int J Gynecol Pathol. In press. 14. Iwakiri S, Sonobe M, Nagai S, Hirata T, Wada H, Miyahara R. Expression status of folate receptor alpha is significantly correlated with prognosis in nonsmall-cell lung cancers. Ann Surg Oncol. 2008;15(3): Toffoli G, Russo A, Gallo A, et al. Expression of folate binding protein as a prognostic factor for response to platinum-containing chemotherapy and survival in human ovarian cancer. Int J Cancer. 1998;79(2): Hartmann LC, Keeney GL, Lingle WL, et al. Folate receptor overexpression is associated with poor outcome in breast cancer. Int J Cancer. 2007;121(5): O Shannessy, DJ, Somers ES, Albone E, et al. Characterization of the human folate receptor alpha via novel antibody-based probes. Oncotarget. 2011; 2(12): Christoph DC, Reyna-Asuncion B, Hassan B, et al. Significance of folate receptor alpha and thymidylate synthase protein expression in patients with nonsmall-cell lung cancer treated with pemetrexed. J Thorac Oncol. 2013;8(1): Clopper CJ, Pearson ES. The use of confidence or fiducial limits illustrated in the case of the binomial. Biometrika. 1934;26: Arch Pathol Lab Med Vol 137, December 2013 Folate Receptor a in Lung Adenocarcinoma Bremer et al