The practice of subtyping of non-small cell lung carcinoma

Transcription

1 ORIGINAL ARTICLE Subtyping of Non-small Cell Lung Carcinoma A Comparison of Small Biopsy and Cytology Specimens Carlie S. Sigel, MD,* Andre L. Moreira, MD, PhD,* William D. Travis, MD,* Maureen F. Zakowski, MD,* Raymond H. Thornton, MD, Gregory J. Riely, MD, PhD, and Natasha Rekhtman, MD, PhD* Background: There is growing evidence that lung adenocarcinoma and squamous cell carcinoma (SQCC) have distinct oncogenic mutations and divergent therapeutic responses, which is driving the heightened emphasis on accurate pathologic subtyping of non-small cell lung carcinoma (NSCLC). The relative feasibility and accuracy of NSCLC subtyping by small biopsy versus cytology is not well established, particularly in current practice where immunohistochemistry (IHC) is becoming routinely used to aid in this distinction. Methods: Concurrent biopsy and cytology specimens obtained during a single procedure and diagnosed as NSCLC during a 2-year period (n 101) were reviewed. Concordance of diagnoses in the two methods was assessed. Accuracy was determined based on subsequent resection or autopsy diagnosis (n 21) or IHC for thyroid transcription factor 1 and p63 on a subset of cases (n 43). Results: The distribution of definitive versus favored versus unclassified categories (reflecting the degree of diagnostic certainty) was similar for biopsy (71% versus 23% versus 6%, respectively) and cytology (69% versus 19% versus 12%, respectively), p When results from paired specimens were combined, the rate of definitive diagnoses by at least one method was increased to 84% and the unclassified rate was decreased to 4%. NSCLC subtype concordance between biopsy and cytology was 93%. Kappa coefficient (95% confidence interval) for agreement between methods was 0.88 ( ) for adenocarcinoma and 0.76 ( ) for SQCC. In pairs with discordant diagnoses (n 7) the correct tumor type was identified with a similar frequency by biopsy (n 4) and cytology (n 3). Factors contributing to mistyping were poor differentiation, necrosis, low cellularity, and lack of supporting IHC. All concordant diagnoses for which verification was available (n 57) were correct. IHC was used more frequently to subtype NSCLC in biopsy than cytology (32% versus 6%; p ). Conclusions: Small biopsy and cytology achieve comparable rates of definitive and accurate NSCLC subtyping, and the optimal results *Department of Pathology, Interventional Radiology Service, Department of Radiology, and Thoracic Oncology Service, Division of Solid Tumor, Department of Medicine, Memorial Sloan-Kettering Cancer Center, New York City, New York. Disclosure: The authors declare no conflicts of interest. Address for correspondence: Natasha Rekhtman, MD, PhD, Department of Pathology, Memorial Sloan-Kettering Cancer Center, 1275 York Avenue, New York, NY rekthman@mskcc.org Copyright 2011 by the International Association for the Study of Lung Cancer ISSN: /11/ are attained when the two modalities are considered jointly. The lower requirement for IHC in cytology highlights the strength of cytomorphology in NSCLC subtyping. Whenever clinically feasible, obtaining parallel biopsy and cytology specimens is encouraged. Key Words: Adenocarcinoma, Squamous cell carcinoma, Cytology, Biopsy. (J Thorac Oncol. 2011;6: ) The practice of subtyping of non-small cell lung carcinoma (NSCLC) as adenocarcinoma (ADC) versus squamous cell carcinoma (SQCC) in pathologic specimens has undergone a major paradigm shift in recent years. 1,2 Specifically, subtyping of NSCLC is no longer viewed as unimportant because of the discovery that several targetable molecular alterations, particularly EGFR mutations 3,4 and EML4-ALK rearrangements, 5 are largely restricted to ADC and because of a non-squamous indication for bevacizumab and pemetrexed. 6,7 Other important molecular differences between ADC and SQCC are increasingly identified, 8,9 suggesting that future targeted therapies will be increasingly histology specific. Selection of patients for appropriate molecular tests and histology-based therapies necessitates accurate pathologic distinction of ADC versus SQCC. As a result, a great emphasis is currently placed on accuracy of NSCLC subtyping in pathologic specimens. Because 70% of NSCLC are unresectable at presentation, 10 treatment is often based on diagnoses from small biopsy or cytology specimens. Until recently, NSCLC subtyping was based entirely on morphology with elective use of mucin stains. However, the classic morphologic criteria for differentiating ADC versus SQCC (glandular architecture and intracellular mucin versus keratinization and intercellular bridges, respectively) may be focal or subtle in small specimens, particularly in poorly differentiated tumors. This explains the historically high rate of unclassified NSCLC designated as NSCLC-not otherwise specified (NOS) in both small biopsy and cytology. 11,12 An important recent development in NSCLC subtyping is mounting evidence that immunohistochemistry (IHC) is a powerful ancillary tool for the distinction of ADC and SQCC, which substantially lowers the rate of unclassified and misclassified NSCLC and decreases the interobserver variability Journal of Thoracic Oncology Volume 6, Number 11, November

2 Sigel et al. Journal of Thoracic Oncology Volume 6, Number 11, November 2011 in small specimens In particular, several recent studies have shown that ADC and SQCC have virtually nonoverlapping coexpression profiles of thyroid transcription factor 1 (TTF-1) and p63 the master transcriptional regulators in glandular and squamous cell lineages, respectively. 13,17,20 22 These robust markers are now increasingly incorporated into routine clinical practice to aid in the diagnosis of morphologically unclassifiable NSCLC (NSCLC-NOS). Recently, our group performed a large cytology/resection correlation study that showed that in a practice with routine utilization of IHC, the rate of unclassified NSCLC is low (3%), and accuracy of ADC versus SQCC subtyping is high 97% overall and 100% for IHC-aided diagnoses, 23 suggesting that cytologic diagnoses in current practice are suitable for guiding the new NSCLC subtype-based therapeutic decisions. However, a direct comparison between the performance of cytology versus small biopsy for NSCLC subtyping in the current practice has not been performed. In this study, we directly compared the rate with which specific subtyping is achieved, the concordance and accuracy of NSCLC diagnoses in cytology versus small biopsy specimens obtained during a single procedure in the context of our routine practice. As a means of verification of diagnostic accuracy, we used not only resection/autopsy (as has been done traditionally) but also IHC, based on the recent evidence that IHC profiles are an extremely accurate means of determining tumor cell lineage in NSCLC. In addition, we compared the frequency with which IHC is used to aid the diagnosis in biopsy versus cytology in our routine practice. PATIENTS AND METHODS A review of the Memorial Sloan-Kettering Cancer Center Department of Pathology database was conducted to identify all concurrent small biopsy and cytology specimens with a diagnosis of NSCLC during a 2-year period (January 1, 2006 January 1, 2008). The decision to obtain parallel specimens (as opposed to a sole biopsy or cytology) was made at the discretion of the individual interventional radiologist or bronchoscopist. Selection was limited to cases in which both biopsy and cytology were either diagnostic or suspicious for malignancy, and NSCLC was reported by at least one of the methods. Because the goal of the study was to analyze NSCLC subtyping in actual clinical practice, we recorded original diagnoses rendered during the initial case review without modification. For determination of accuracy, the diagnoses rendered by biopsy/cytology were compared with subsequent resection (n 20) or autopsy (n 1) when available. In addition, IHC for TTF-1 (SPT24, NovoCastra, 1:50 dilution) and p63 (4A4, Dako, 1:700 dilution) was used as another means of diagnosis verification. IHC was performed either as part of initial work-up (n 38) or as part of this study to resolve discordant diagnoses (n 5). Interpretation of TTF- 1/p63 immunoprofiles was based on the algorithm established in recent studies. 13,17,20,22 The morphologic features for all discordant and unclassified cases were reviewed by two pathologists with subspecialty expertise in cytopathology (N.R. and C.S.S.) and thoracic pathology (N.R.). The types of cytologic preparations included the following (according to standard specimen processing protocol in our cytology laboratory): (1) air-dried smear stained with Diff-Quik (used for immediate adequacy assessment), (2) smear fixed in 95% ethanol and stained with hematoxylin and eosin (H&E) or Papanicolaou stain, (3) ThinPrep monolayer prepared from the needle rinse in CytoLyt fixative, and (4) H&E-stained section from a paraffin-embedded cell block prepared by centrifugation of CytoLyt fluid. No cell blocks were available for cases with a very scant needle rinse, which yielded no visible cell pellet after centrifugation. All fineneedle aspirates (FNAs) were obtained with on-site adequacy assessment by a cytotechnologist or pathologist. As part of routine clinical practice during the study period, biopsy and cytology specimens were reviewed independently by different pathologists. Biopsy specimens were reviewed by pathologists with subspecialty expertise in thoracic pathology, and the cytology specimens were reviewed by experienced staff cytopathologists. A standard practice at our institution is for cytology specimens to be reviewed and reports finalized a day earlier than biopsy specimens due to a shorter processing time of cytologic specimens, and therefore, in most cases, cytologic diagnoses were rendered without the knowledge of biopsy diagnosis. However, a correlative review may have been performed at the discretion of individual pathologists, and therefore, we cannot exclude that for a subset of cases, diagnoses were rendered with the knowledge of the diagnosis in the other specimen. During the study period, the use of the terms definitive versus favored versus unclassified reflected the degree of diagnostic certainty as judged on case-by-case basis by the individual pathologists. In general, definitive diagnoses (ADC or SQCC) were rendered when morphologic features were diagnostic of a tumor type (such as obvious glandular features or cytoplasmic mucin for ADC versus keratinization for SQCC) or when IHC was interpreted as diagnostic of a tumor type. The favored category was used when morphology and IHC were interpreted as suggestive of a tumor type but qualitatively or quantitatively insufficient for definitive diagnosis. An alternative use of the favored category has been recently proposed, 24 but this study reflects the use of this category in clinical practice during the study period. The unclassified (NSCLC-NOS) category was used for samples where tumor type could not be determined by morphology, but cellularity was insufficient for IHC, or IHC results were judged as equivocal. In addition, in some underclassified cytology samples, IHC was intentionally withheld because of the deferral of IHC to the concurrent biopsy to avoid duplication of the workup. Because bronchoscopic specimens usually include several cytologic specimens (bronchial brush, wash, lavage, and transbronchial FNA), the most specific diagnosis reached by at least one of the methods was recorded for the study purposes. For example, if a bronchial brush was diagnosed as ADC and bronchial wash as NSCLC-NOS, the overall diagnosis for cytology was recorded as ADC. The significance of associations was assessed by twotailed Fisher s exact test, with a p value of 0.05 considered 1850 Copyright 2011 by the International Association for the Study of Lung Cancer

3 Journal of Thoracic Oncology Volume 6, Number 11, November 2011 NSCLC Subtyping by Biopsy versus Cytology statistically significant. Agreement was analyzed by calculation of Cohen s kappa coefficient with the strength of association interpreted according to the following guidelines: very good ( ), good ( ), moderate ( ), fair ( ), and poor ( 0.20). 25 This study was performed with the approval of the Institutional Review Board of Memorial Sloan-Kettering Cancer Center (New York, NY). RESULTS Specimen Characteristics One hundred one patients had a concurrent cytology and small biopsy specimens from the same diagnostic procedures with diagnosis of NSCLC during the study period. Specimens included concurrent transthoracic specimens transthoracic core biopsies paired with transthoracic FNAs (n 60), and concurrent bronchoscopic specimens transbronchial biopsies paired with transbronchial FNAs (n 4) and/or bronchial wash/brush/lavage (n 37). Seven transthoracic core biopsies were paired with core touch preparations (core imprints); for the purposes of this study they were analyzed as a single group with transthoracic FNAs. Average follow-up was 14.6 months (range: 3 days 41 months). The patient characteristics were male:female ratio, 1:1.9; average age, 68 years; and age range, years. TABLE 1. Distribution of Tumor Subtyping Categories Reflecting the Degree of Diagnostic Certainty for Biopsy versus Cytology Biopsy, N (%) Cytology, N (%) Total ADC a SQCC a Total ADC a SQCC a Definitive 72 (71) 44 (77) 28 (74) 70 (69) 47 (81) 23 (74) Favored 23 (23) 13 (23) 10 (26) 19 (19) 11 (19) 8 (26) Unclassified 6 (6) 12 (12) Total a Original diagnosis rendered on biopsy or cytology. ADC, adenocarcinoma; SQCC, squamous cell carcinoma. TABLE 2. Correlation of Tumor Subtyping Categories for Biopsy versus Cytology, N (%) Biopsy Definitive Favored Unclassified Total Cytology Definitive 58 (57) 11 (11) 1 (1) 70 Favored 9 (9) 9 (9) 1 (1) 19 Unclassified 5 (5) 3 (3) 4 (4) 12 Total TABLE 3. Correlation of Tumor Subtypes Diagnosed by Biopsy versus Cytology, N (%) Biopsy ADC a SQCC a Unclassified Total Cytology ADC a 51 (50) 5 (5) 2 (2) 58 SQCC a 2 (2) 29 (28) 0 31 Unclassified 4 (4) 4 (4) 4 (4) 12 Total a Includes both definitive and favored diagnoses. ADC, adenocarcinoma; SQCC, squamous cell carcinoma. Comparison of Degree of Diagnostic Certainty in Biopsy versus Cytology The level of diagnostic certainty in NSCLC subtyping consisted of three categories: definitive, favored, and unspecified (NSCLC-NOS). As presented in Table 1, the distribution of definitive versus favored versus unclassified categories was similar for biopsy (71% versus 23% versus 6%, respectively) and cytology (69% versus 19% versus 12%, respectively), p This distribution was similar for ADC and SQCC in both methods. As presented in Table 2, a more definitive diagnosis was made by biopsy in 17% of cases and cytology in 13% of cases. By combining the results of the two methods, the number of patients with unclassified diagnoses was reduced to 4%, and the rate of definitive diagnosis by at least one modality was increased to 84%. We did not find a difference in the level of diagnostic certainty between various cytology methods (transthoracic FNA versus transbronchial FNA versus bronchial brush/wash/lavage) and between biopsy methods (transthoracic biopsy versus transbronchial biopsy) in this study. Unclassified biopsy (n 6) and cytology (n 12) specimens were distributed as follows: unclassified biopsy/ classified cytology (n 2), unclassified cytology/classified biopsy (n 8), and both methods unclassified (n 4) (Table 1). In particular, review of unclassified biopsy/classified cytology showed that in one instance, a tumor in the biopsy sample was highly necrotic, whereas the cells were better preserved in cytology, and in the second instance, a tumor had poorly differentiated histology and IHC was noncontributory, whereas morphologic features in concurrent cytology were diagnostic of a tumor type. In the unclassified cytology/ classified biopsy group, the majority (7/8) of biopsy diagnoses were aided by IHC. Four cases unclassified by both methods were uniformly high-grade, poorly differentiated, necrotic tumors. In two cases, the possibility of neuroendocrine carcinoma was considered, one case had basaloid morphology and one case had pleomorphic/sarcomatoid morphologic feature. In all cases, IHC results were equivocal. Resection follow-up for these cases was not available. Concordance and Accuracy of NSCLC Subtyping in Biopsy versus Cytology The concordance in the NSCLC subtypes between biopsy and cytology was 93% (Table 3). Kappa coefficients with 95% confidence intervals for the agreement of biopsy and cytology diagnoses were 0.88 ( ) for ADC, 0.76 ( ) for SQCC, and 0.40 ( ) for NSCLC- NOS. NSCLC subtype was discordant in seven cases (7%). For discordant diagnoses, subsequent resection (n 1), autopsy (n 1), or additional IHC (n 5) revealed that the Copyright 2011 by the International Association for the Study of Lung Cancer 1851

4 Sigel et al. Journal of Thoracic Oncology Volume 6, Number 11, November 2011 TABLE 4. Review of Cases with Discrepant Biopsy and Cytology Diagnoses Case ID Cyto Method Bx Method Cyto Dx Bx Dx Final Dx Method of Final Dx Modality with Correct Dx Factors Contributing to Misclassification a 1 TTNA TTB Favor SQCC Favor ADC ADC Resection Bx Poor differentiation, necrosis, low cellularity 2 a BBR TBB Favor SQCC Favor ADC ADC Autopsy Bx Poor differentiation 3 BBR TBB Favor ADC SQCC SQCC IHC Bx Poor differentiation 4 b BW TBB ADC SQCC SQCC IHC Bx Poor differentiation 5 c BW TBB ADC Favor SQCC ADC IHC Cyto Poor differentiation 6 BW TBB ADC Favor SQCC ADC IHC Cyto Poor differentiation, necrosis 7 TTNA TTB Favor ADC SQCC ADC IHC Ctyo Poor differentiation, necrosis a Figure 1A B. b Figure 1C D. c Figure 1E F. d All incorrect subtypes were assigned in the absence of confirmatory immunohistochemistry. ADC, adenocarcinoma; BBR, bronchial brushing; BW, bronchial washing; Bx, biopsy; Cyto, cytology; Dx, diagnosis; SQCC, squamous cell carcinoma; TBB, transbronchial biopsy; TTB, transthoracic biopsy; TTNA, transthoracic fine needle aspiration. correct diagnosis was rendered with a similar frequency by biopsy (n 4) and cytology (n 3) (Table 4). Misclassified cytologic specimens included an even number of SQCC (n 2) and ADC (n 2), whereas misclassified biopsies were all ADC morphologically mimicking SQCC due to solid growth pattern and abundant pink cytoplasm (n 3). The morphologic pitfalls are illustrated in Figure 1. Overall, all misclassified case were scant specimens of poorly differentiated and/or necrotic tumors classified in the absence of IHC. All biopsy and cytology specimens with concordant diagnoses for which resection or IHC verification was available (n 53) were correct. Utilization of IHC in Routine Practice As summarized in Table 5, utilization of IHC to aid in the diagnosis of ADC and SQCC was significantly higher in biopsy than cytology (32% versus 6%, p ). Diagnosis of ADC was more commonly aided by IHC in biopsy than cytology (21% versus 6%, p 0.003), as was the diagnosis of SQCC (11% versus 0, p ). The majority of IHC-aided diagnoses were classified as definitive in both biopsy (66%) and cytology (83%), and the rest were classified as favored. For paired specimens where the diagnosis of SQCC required IHC in biopsy but not cytology, definite evidence of keratinization was evident in Papanicolaou-stained cytological preparations but was inapparent on H&E-stained biopsies (Figures 2A C). Similarly, clear evidence of glandular differentiation (polarized cells with a flat luminal edge and/or acinar structures) was readily identifiable in cytology specimens in which concurrent biopsy was indefinite for the tumor type and required IHC for diagnosis (Figures 2D F). DISCUSSION The main finding of this study is that in our institutional experience paired small biopsy and cytology specimens achieve a comparable certainty and accuracy in NSCLC subtyping, and optimal results are obtained when the findings from both methods are combined. In particular, results from combined methods maximize the number of definitive diagnoses by at least one modality and decrease the rate of unclassified diagnoses (NSCLC-NOS). The number of prior studies which have directly compared the efficacy of NSCLC subtyping in paired small biopsy versus cytology specimens is limited, and none have been performed in a practice where IHC is routinely used to subtype morphologically challenging cases. Similar to the findings of this study, prior studies found that the feasibility and accuracy of subtyping of malignant tumors overall was comparable in paired biopsy and cytology specimens Of note, in prior studies, the ability to determine specific diagnoses for benign tumors and mass-forming lesions (such as organizing pneumonia) was superior for biopsy, 26,27 which remains the main advantage of biopsy over cytology. The reported rate of unclassified specimens and accuracy of NSCLC subtyping in unpaired small specimens varies significantly among prior studies, 11,12,29,30 with unclassified NSCLC rate of more than 30% reported in some studies. 30 This is in stark contrast with the low rate of unclassified NSCLC by combined (4%) and individual methods in this series and recent cytology 23 and small biopsy 31 studies from our institution. The low unclassified rate (and high concordance and accuracy) of biopsy and cytology in this study is attributable at least in part to routine utilization of IHC for subtyping of difficult cases. This approach is becoming incorporated in routine practice in recent years and is recommended in the recent IASLC/ATS/ERS ADC classification proposal. 24 In addition to the role of IHC, the high rates of NSCLC subtyping in this series may be partially reflective of practice in a subspecialized referral cancer center. A key observation in this study is that morphologic evidence of differentiation as ADC or SQCC is frequently more apparent in cytologic specimens than in small biopsies. This is demonstrated by the significantly lower need for IHC to identify a tumor type in cytology compared with biopsy (6% versus 31%, respectively) despite a similar degree of certainty and accuracy of NSCLC subtyping by these methods. The main advantage of cytologic preparations in NSCLC subtyping is that the Papanicolaou stain (a routine stain in cytology) specifically highlights keratinizing cells, whereas 1852 Copyright 2011 by the International Association for the Study of Lung Cancer

5 Journal of Thoracic Oncology Volume 6, Number 11, November 2011 NSCLC Subtyping by Biopsy versus Cytology TABLE 5. Utilization of Immunohistochemistry to Distinguish of Adenocarcinoma and Squamous Cell Carcinoma in Concurrent Biopsy and Cytology Specimens, N (%) Biopsy (n 101) Cytology (n 101) p ADC diagnosis aided by IHC 21 (21) 6 (6) SQCC diagnosis aided by IHC 11 (11) Total number of ADC and SQCC diagnoses aided by IHC 32 (32) 6 (6) IHC, immunohistochemistry; ADC, adenocarcinoma; SQCC, squamous cell carcinoma. FIGURE 1. Examples of rare cases with discrepant adenocarcinoma (ADC) versus squamous cell carcinoma (SQCC) diagnoses in biopsy and cytology. A and B, ADC identified as SQCC by cytology with correct diagnosis of ADC in concurrent biopsy. Cytology (A) shows carcinoma cells with dense blue cytoplasm mimicking SQCC. Concurrent biopsy (B) shows poorly differentiated non-small cell carcinoma, which was diffusely positive for TTF-1 (not shown) supporting ADC. C and D, SQCC identified as ADC by cytology with correct diagnosis of SQCC in concurrent biopsy: Cytology (C) shows three-dimensional clusters of nonkeratinizing cells with prominent nucleoli, closely mimicking ADC. (D) Poorly differentiated carcinoma with focal squamous whorls diagnostic of SQCC is seen on the corresponding biopsy. E and F, ADC identified as SQCC by biopsy with correct diagnosis of ADC in concurrent cytology: Polarized (eccentric) nuclei and granular cytoplasm support ADC in cytology (E). Tumor cells in concurrent biopsy (F) have abundant pink cytoplasm and sharp cell borders mimicking SQCC. This tumor was found to be diffusely positive for TTF-1 and negative for p63, supporting ADC (not shown). A, C, and E, Papanicolaou stain, 400. B, D, and F, H&E stain, 400. H&E (a routine stain in histology) does not. This accounts for greater ability of routine cytology to identify squamous differentiation without the aid of IHC. Another advantage of cytology is that the three-dimensional cell arrangements of ADC may be better visualized in cytology than histology. Finally, cytology lacks formalin-fixation artifact and has less crush artifact compared with small biopsy, which allows for greater resolution of nuclear and cytoplasmic detail. This latter factor accounts for a well-known advantage of cytology over small biopsy in the diagnosis of small cell lung carcinoma. 32,33 Our data support the conclusion that cytology offers a similar advantage in the subtyping of NSCLC. A novel approach used in this study was to use IHC (in addition to final diagnosis in subsequent resection or autopsy) as a means of verifying the accuracy of the small specimen diagnoses. Several recent studies have shown that ADC and SQCC have distinct profiles of TTF-1 and p63 expression, 13,17,20 22 and this can be used to aid in the diagnosis of morphologically challenging cases. The recently developed IHC algorithm is becoming increasingly incorporated into clinical practice (as discussed earlier) and provides an alternative gold standard for measuring the accuracy of small specimens. The validity of this approach is supported by recent studies showing that tumor type identified by IHC in preoperative specimens predicts with excellent accuracy the resection diagnosis of ADC versus SQCC. 13,20,23 The study design, which included specimens obtained during a single procedure, was chosen to allow the most direct comparison between cytology and biopsy specimens; however, this approach has several limitations. The main disadvantage of this approach is that paired specimens may not be entirely representative of the individual, isolated performance of either method. For example, the total number of FNA passes performed by an interventional radiologist may be reduced if a core biopsy is obtained concurrently, resulting in a less cellular cytology specimen. Conversely, FNA may be more likely to be obtained in cases where a core biopsy is judged to be suboptimal, particularly if friable or minute. The second limitation is that for some paired specimens pathologists may elect to render a more generic diagnosis (such as NSCLC-NOS) in cytology, and defer the IHC work-up and further classification to a biopsy. Therefore, the relative rate of IHC utilization and the rate of unclassified specimens in this study may not reflect a full potential of cytology for NSCLC subtyping, which is more accurately reflected in a study where cytology is the only diagnostic modality. 23 The third limitation of the study is that verification of diagnosis (in the form of resection, autopsy or IHC) was not available for nearly half of the patients. As a result, the primary conclusion from this study is the high level of concordance in the diagnoses, whereas the conclusion for comparable accuracy is based on only a subset of cases. In addition to tumor subtyping, another important consideration in the method comparison is diagnostic yield (i.e., the rate of successful sampling of a mass lesion). Because in this study only cases that were diagnostic by both modalities were selected, the comparative diagnostic Copyright 2011 by the International Association for the Study of Lung Cancer 1853

6 Sigel et al. Journal of Thoracic Oncology Volume 6, Number 11, November 2011 FIGURE 2. Definitive morphologic evidence of squamous or glandular differentiation is frequently more apparent in cytologic preparation than biopsies. Cytology specimen (A) shows definitive keratinization (dense red or orange cytoplasm) diagnostic of squamous cell carcinoma. In contrast, keratinization is inapparent in concurrent biopsy (B) which required immunohistochemistry (IHC, positive p63; C) for the diagnosis. IHC for TTF-1 was negative (not shown). Cytology specimen (D) shows eccentric nuclei with flat luminal edge (arrow) diagnostic of adenocarcinoma. In contrast, morphologic differentiation is inapparent in concurrent core biopsy (E) and required IHC (positive TTF-1; F) for the diagnosis. A and D, Papanicolaou stain, 400. B and E, H&E stain, 400. C and F: Immunohistochemistry for p63 and TTF-1, respectively, 400. yield of biopsy versus cytology cannot be addressed. However, prior studies show that diagnostic yield is comparable for transthoracic biopsy and FNA, 26,27,34 36 whereas results vary for different bronchoscopic methods. 28,36 Importantly, several studies show significant improvement in the diagnostic yield by combining biopsy and cytology specimens in a single bronchoscopic 28,37 39 or transthoracic 26,27,40 procedure. It is fairly standard in bronchoscopic procedures to obtain cytological preparations concurrently with a biopsy, whereas for transthoracic approaches, the material obtained is more variable. 41 When a concurrent FNA is obtained with a transthoracic core biopsy, it is a common view that the utility of cytology is limited to on-site adequacy assessment to confirm the successful targeting of a lesion, but that cytology does not have an important role in reaching the final diagnosis. In contrast, we show a distinct value added by cytologic specimens, in that in a subset of cases a more specific and accurate diagnosis is achieved in cytology. Together with prior studies showing the advantage of paired specimens for maximizing the overall diagnostic yield, an attempt should be made to obtain paired specimens whenever clinically possible. Another essential consideration for small specimens in the age of targeted therapies is suitability for molecular testing. Although this question was not addressed in the current study, several prior studies have shown that cytology specimens, particularly FNAs are highly suitable for EGFR (and KRAS) mutation analysis. 16,23,42 45 Recent studies from our institution suggest that overall small biopsies and FNAs have a comparable rate of yielding sufficient material for molecular testing, 23,46,47 although further studies are needed to directly compare specific specimen types. In particular, when deciding on a method of tissue sampling, an important consideration is that standard molecular studies and IHC generally require cellular paraffin-embedded material (formalin-fixed paraffinembedded tissue for biopsy and cell block for cytology). Although FNAs typically yield enough material for a cell block, other cytologic specimens (sputum, bronchial brush/wash/lavage) are usually insufficient for a cell block and cannot be used for routine IHC or molecular studies. In particular, core touch preparations (core imprints), 48 which are currently used at some institutions, including ours, for immediate on-site assessment of adequacy of a core biopsy, should not be regarded as an FNA-equivalent as they do not yield a cell block and cannot be used for routine ancillary studies. For both biopsies and FNAs, obtaining tissue in excess of what traditionally has been sufficient for a minimal morphologic diagnosis of non-small cell carcinoma versus small cell carcinoma (which may require only a few tumor cells) is now necessary to ensure that cellularity is sufficient for potential IHC or molecular studies. This can be achieved by performing additional FNA passes and maximizing the amount of material in the needle-rinse fluid, which is used as a source of a cell block, or by obtaining an extra biopsy, if clinically feasible. Close interdisciplinary communication is encouraged to ensure that the specimens are providing sufficient material for the optimal diagnosis and predictive marker testing. In summary, in our routine practice, small biopsy and cytology specimens achieve comparable specificity and accuracy of NSCLC subtyping, and optimal results are obtained when the findings from both modalities are combined. We therefore recommend obtaining concurrent biopsy and cytology samples, whenever clinically feasible, and correlating the results from both methods to ensure the greatest diagnostic accuracy Copyright 2011 by the International Association for the Study of Lung Cancer

7 Journal of Thoracic Oncology Volume 6, Number 11, November 2011 NSCLC Subtyping by Biopsy versus Cytology REFERENCES 1. Travis WD, Rekhtman N, Riley GJ, et al. Pathologic diagnosis of advanced lung cancer based on small biopsies and cytology: a paradigm shift. J Thorac Oncol 2010;5: Travis WD, Rekhtman N. Pathological diagnosis and classification of lung cancer in small biopsies and cytology: strategic management of tissue for molecular testing. Semin Respir Crit Care Med 2011;32: Mok TS, Wu YL, Thongprasert S, et al. Gefitinib or carboplatin-paclitaxel in pulmonary adenocarcinoma. N Engl J Med 2009;361: Ladanyi M, Pao W. Lung adenocarcinoma: guiding EGFR-targeted therapy and beyond. Mod Pathol 2008;21(Suppl 2):S16 S Shaw AT, Yeap BY, Mino-Kenudson M, et al. Clinical features and outcome of patients with non-small-cell lung cancer who harbor EML4- ALK. J Clin Oncol 2009;27: Scagliotti G, Brodowicz T, Shepherd FA, et al. Treatment-by-histology interaction analyses in three phase III trials show superiority of pemetrexed in nonsquamous non-small cell lung cancer. J Thorac Oncol 2011;6: Johnson DH, Fehrenbacher L, Novotny WF, et al. Randomized phase II trial comparing bevacizumab plus carboplatin and paclitaxel with carboplatin and paclitaxel alone in previously untreated locally advanced or metastatic non-small-cell lung cancer. J Clin Oncol 2004;22: Bass AJ, Watanabe H, Mermel CH, et al. SOX2 is an amplified lineage-survival oncogene in lung and esophageal squamous cell carcinomas. Nat Genet 2009;41: Weiss J, Sos ML, Seidel D, et al. Frequent and focal FGFR1 amplification associates with therapeutically tractable FGFR1 dependency in squamous cell lung cancer. Sci Transl Med 2010;2:62ra Schrump DS, Altorki NK, Henschke CL, et al. Non-small cell lung cancer. In VT DeVita, S Hellman, SA Rosenberg (Eds.), Cancer: Principles and Practices of Oncology. Philadelphia, PA: Lippincott Williams & Wilkins, Pp Ou SH, Zell JA. Carcinoma NOS is a common histologic diagnosis and is increasing in proportion among non-small cell lung cancer histologies. J Thorac Oncol 2009;4: Edwards SL, Roberts C, McKean ME, et al. Preoperative histological classification of primary lung cancer: accuracy of diagnosis and use of the non-small cell category. J Clin Pathol 2000;53: Rekhtman N, Ang DC, Sima CS, et al. Immunohistochemical algorithm for differentiation of lung adenocarcinoma and squamous cell carcinoma based on large series of whole tissue sections with validation in small specimens. Mod Pathol. [Epub ahead of print May 27, 2011] DOI: /modpathol Terry J, Leung S, Laskin J, et al. Optimal immunohistochemical markers for distinguishing lung adenocarcinomas from squamous cell carcinomas in small tumor samples. Am J Surg Pathol 2010;34: Rossi G, Papotti M, Barbareschi M, et al. Morphology and a limited number of immunohistochemical markers may efficiently subtype non-small-cell lung cancer. J Clin Oncol 2009;27:e141 e142; author reply e143 e Nicholson AG, Gonzalez D, Shah P, et al. Refining the diagnosis and EGFR status of non-small cell lung carcinoma in biopsy and cytologic material, using a panel of mucin staining, TTF-1, cytokeratin 5/6, and P63, and EGFR mutation analysis. J Thorac Oncol 2010;5: Mukhopadhyay S, Katzenstein AL. Subclassification of non-small cell lung carcinomas lacking morphologic differentiation on biopsy specimens: utility of an immunohistochemical panel containing TTF-1, napsin A, p63, and CK5/6. Am J Surg Pathol 2011;35: Loo PS, Thomas SC, Nicolson MC, et al. Subtyping of undifferentiated non-small cell carcinomas in bronchial biopsy specimens. J Thorac Oncol 2010;5: Khayyata S, Yun S, Pasha T, et al. Value of P63 and CK5/6 in distinguishing squamous cell carcinoma from adenocarcinoma in lung fine-needle aspiration specimens. Diagn Cytopathol 2009;37: Pelosi G, Rossi G, Bianchi F, et al. Immunhistochemistry by means of widely agreed-upon markers (cytokeratins 5/6 and 7, p63, thyroid transcription factor-1, and vimentin) on small biopsies of non-small cell lung cancer effectively parallels the corresponding profiling and eventual diagnoses on surgical specimens. J Thorac Oncol 2011;6: Bishop JA, Benjamin H, Cholakh H, et al. Accurate classification of non-small cell lung carcinoma using a novel microrna-based approach. Clin Cancer Res 2010;16: Righi L, Graziano P, Fornari A, et al. Immunohistochemical subtyping of nonsmall cell lung cancer not otherwise specified in fine-needle aspiration cytology: a retrospective study of 103 cases with surgical correlation. Cancer. [Epub ahead of print January 18, 2011] DOI: /cncr Rekhtman N, Brandt SM, Sigel CS, et al. Suitability of thoracic cytology for new therapeutic paradigms in non-small cell lung carcinoma: high accuracy of tumor subtyping and feasibility of EGFR and KRAS molecular testing. J Thorac Oncol 2011;6: Travis WD, Brambilla E, Noguchi M, et al. International association for the study of lung cancer/american thoracic society/european respiratory society international multidisciplinary classification of lung adenocarcinoma. J Thorac Oncol 2011;6: Landis JR, Koch GG. An application of hierarchical kappa-type statistics in the assessment of majority agreement among multiple observers. Biometrics 1977;33: Aviram G, Greif J, Man A, et al. Diagnosis of intrathoracic lesions: are sequential fine-needle aspiration (FNA) and core needle biopsy (CNB) combined better than either investigation alone? Clin Radiol 2007;62: Yamagami T, Iida S, Kato T, et al. Combining fine-needle aspiration and core biopsy under CT fluoroscopy guidance: a better way to treat patients with lung nodules? AJR Am J Roentgenol 2003;180: Matsuda M, Horai T, Nakamura S, et al. Bronchial brushing and bronchial biopsy: comparison of diagnostic accuracy and cell typing reliability in lung cancer. Thorax 1986;41: Nizzoli R, Tiseo M, Gelsomino F, et al. Accuracy of fine needle aspiration cytology in the pathological typing of non-small cell lung cancer. J Thorac Oncol 2011;6: Thomas JS, Lamb D, Ashcroft T, et al. How reliable is the diagnosis of lung cancer using small biopsy specimens? Report of a UKCCCR Lung Cancer Working Party. Thorax 1993;48: Suh J, Rekhtman N, Ladanyi M, et al. Testing of new IASLC/ATS/ERS criteria for diagnosis of lung adenocarcinoma (AD) in small biopsies: minimize immunohistochemistry (IHC) to maximize tissue for molecular studies. Mod Pathol 2011;24:404A 431A, 424A. 32. Zakowski MF. Pathology of small cell carcinoma of the lung. Semin Oncol 2003;30: Rekhtman N. Neuroendocrine tumors of the lung: an update. Arch Pathol Lab Med 2010;134: Bocking A, Klose KC, Kyll HJ, et al. Cytologic versus histologic evaluation of needle biopsy of the lung, hilum and mediastinum. Sensitivity, specificity and typing accuracy. Acta Cytol 1995;39: Klein JS, Salomon G, Stewart EA. Transthoracic needle biopsy with a coaxially placed 20-gauge automated cutting needle: results in 122 patients. Radiology 1996;198: Schreiber G, McCrory DC. Performance characteristics of different modalities for diagnosis of suspected lung cancer: summary of published evidence. Chest 2003;123:115S 128S. 37. Jones AM, Hanson IM, Armstrong GR, et al. Value and accuracy of cytology in addition to histology in the diagnosis of lung cancer at flexible bronchoscopy. Respir Med 2001;95: Govert JA, Dodd LG, Kussin PS, et al. A prospective comparison of fiberoptic transbronchial needle aspiration and bronchial biopsy for bronchoscopically visible lung carcinoma. Cancer 1999;87: Dobler CC, Crawford AB. Bronchoscopic diagnosis of endoscopically visible lung malignancies: should cytological examinations be carried out routinely? Intern Med J 2009;39: Boiselle PM, Shepard JA, Mark EJ, et al. Routine addition of an automated biopsy device to fine-needle aspiration of the lung: a prospective assessment. AJR Am J Roentgenol 1997;169: Yung RC. Tissue diagnosis of suspected lung cancer: selecting between bronchoscopy, transthoracic needle aspiration, and resectional biopsy. Respir Care Clin N Am 2003;9: Boldrini L, Gisfredi S, Ursino S, et al. Mutational analysis in cytological specimens of advanced lung adenocarcinoma: a sensitive method for molecular diagnosis. J Thorac Oncol 2007;2: Copyright 2011 by the International Association for the Study of Lung Cancer 1855

8 Sigel et al. Journal of Thoracic Oncology Volume 6, Number 11, November Horiike A, Kimura H, Nishio K, et al. Detection of epidermal growth factor receptor mutation in transbronchial needle aspirates of non-small cell lung cancer. Chest 2007;131: Smouse JH, Cibas ES, Janne PA, et al. EGFR mutations are detected comparably in cytologic and surgical pathology specimens of nonsmall cell lung cancer. Cancer 2009;117: Billah S, Stewart J, Staerkel G, et al. EGFR and KRAS mutations in lung carcinoma: molecular testing by using cytology specimens. Cancer Cytopathol 2011;119: Arcila ME, Oxnard GR, Nafa K, et al. Rebiopsy of lung cancer patients with acquired resistance to EGFR inhibitors and enhanced detection of the T790M mutation using a locked nucleic acid-based assay. Clin Cancer Res 2011;17: Solomon SB, Zakowski MF, Pao W, et al. Core needle lung biopsy specimens: adequacy for EGFR and KRAS mutational analysis. AJR Am J Roentgenol 2010;194: Chandan VS, Zimmerman K, Baker P, et al. Usefulness of core roll preparations in immediate assessment of neoplastic lung lesions: comparison to conventional CT scan-guided lung fine-needle aspiration cytology. Chest 2004;126: Copyright 2011 by the International Association for the Study of Lung Cancer