Subtyping of nonsmall cell lung cancer on ... - Wiley Online Library

ORIGINAL ARTICLE

Subtyping of Nonsmall Cell Lung Cancer on Cytology Specimens: Reproducibility of Cytopathologic Diagnoses on Sparse Material Omland Silje Haukali, M.D,1* Hager Henrik, M.D,2 Ege Karen Olsen, M.D., Ph.D,3 Thomsen Birthe, M.D,1 and Skov Birgit Guldhammer , M.D., Ph.D.1

Cytologic examination of fine-needle aspiration (material is increasingly used in diagnosing lung cancer. High interobserver agreement in distinguishing small-cell lung cancer from nonsmall-cell lung cancer (NSCLC) on cytologic material has been demonstrated. Because of new treatment-modalities, subclassification of NSCLC into squamous cell carcinoma (SQC) and nonSQC has clinical impact. Subclassification based on morphology alone may be difficult, but applying immunohistochemistry (IHC) to clot-material has proved helpful. When insufficient material is available to make a clot from the aspirate, cytoscrape (CS) can convert cytologic material into tissue fragments useful for IHC. The purpose of this study was to test the reproducibility of pulmonary malignant diagnoses, in particular distinction between subgroups of NSCLC, based on smeared material and IHC on CS. A consecutive series of May–Grunwald–Giemsa (MGG) stained smears and CS with IHC on material from 79 patients suspected of having lung cancer was included. The material was circulated twice to four pathologists. The diagnoses were categorized in five groups: SQC, adenocarcinoma of the lung, non-SQC, benign lesion and other forms of malignancy, including metastases. Reproducibility was analyzed using Kappa statistics. Interobserver reproducibility of the diagnoses in round 1 was good to very good (kappa 0.57–0.71) and very good in round 2 (0.63–0.80). Reproducibility of subclassification of NSCLC based on MGG stained smear and IHC on CS, was very good among experienced pathologists. With only sparse material

1 Department of Pathology, Rigshospitalet, Copenhagen University Hospital, Denmark 2 Department of Pathology, Aarhus University Hospital, Aarhus, Denmark 3 Department of Pathology, Odense University Hospital, Odense, Denmark *Correspondence to: Silje Haukali Omland, M.D., Rigshospitalet, Copenhagen University Hospital, Department of Pathology, Blegdamsvej 9, 2100 Copenhagen, Denmark. E-mail: [email protected] Received 13 June 2012; Accepted 14 February 2013 DOI: 10.1002/dc.22995 Published online 17 May 2013 in Wiley Online Library (wileyonlinelibrary.com).

C 2013 WILEY PERIODICALS, INC. V

available, CS should be used to achieve reproducible diagnoses, including subtyping of NSCLC. Diagn. Cytopathol. 2014;42:105–110. VC 2013 Wiley Periodicals, Inc. Key Words: reproducibility; cytopathology; nonsmall cell lung cancer; subtyping; immunohistochemistry

Cytologic examination of fine-needle aspiration material (FNA) and other cytological specimens is increasingly used in the diagnosis and staging of lung cancer as well in the diagnosis of potential metastases to the lung. More than 70% of lung cancers are diagnosed and staged using small biopsies and/or cytological specimens as the majority of patients with lung cancer present themselves in advanced stages.1 Because of new treatment modalities, pathologists have been required to provide more detailed diagnoses, in particular to distinguish pulmonary adenocarcinomas (ACL) from squamous cell carcinomas (SQC). The epidermal growth factor receptor (EGFR)directed tyrosine kinase inhibitors erlotinib and gefitinib are effective in patients with EGFR mutation, most frequently observed in ACL.2,3 Patients with adenocarcinomas have more benefit of pemetrexed than patients with squamous cell histology,4 and bevacizumab should not be given to patients with SQC primarily because of the risk of fatal pulmonary hemorrhage.5 A high reproducibility of the distinction between SCLC and nonsmall-cell lung cancer (NSCLC) on cytological specimens has been observed.6 The distinction between subgroups of NSCLC, based on morphology alone, may be challenging in small biopsies and/or cytology specimens, and the accuracy as well as the reproducibility of subtyping NSCLC has been only moderate and insufficient for proper treatment of these subgroups of NSCLC.7,8 However, by applying a (small) panel of immunohistochemical antibodies, the distinction between ACL and SQC improves to a clinically relevant level.9–11 Strategic Diagnostic Cytopathology, Vol 42, No 2

105

Diagnostic Cytopathology DOI 10.1002/dc

OMLAND ET AL.

use of antibodies for accurate diagnosis of lung cancer is important in order to spare material for further analyses including molecular studies.1,10 Immunohistochemistry (IHC) on conventional smears from cytological material is often nonconclusive because of unspecific background staining of disrupted cells and membrane fragments sticking to slides or bloody smears. In contrast, cellblocks provide optimal material for ancillary studies, including IHC.12,13 Several studies have demonstrated that applying IHC to cellblocks improves diagnoses based on histological as well as on cytological material.9–11 The optimal way to obtain a cellblock is to make a clot directly from the aspirated material and convert this cytological material into a cellblock, which can be handled as a biopsy specimen. However, the aspirated material may be too sparse to form a clot. In this case a cytoscrape (CS) can be made. CS is a technique that has been shown to improve the diagnostic information from FNA.14 The utility of the method has been evaluated by assessing the sensitivity and specificity of the method and by quantifying the extra diagnostic information obtained by the method in comparison with information obtained by smears alone. In aspirates from 47 patients with possible malignant infiltrates, malignant tumor cells were identified in the CS material in 91% of cases. CS material stained with thyroid transcription factor-1 (TTF1) and mucin improved the diagnoses made on smears in 72% of cases. Thus, the accuracy of the test has been proved by providing more precise subtyping of NSCLC or proper identification of primary tumor in cases of metastases to the lung.14 The purpose of this study was to test the reproducibility among pathologists of subtyping NSCLC on sparse cytological material using a combination of May–Grunwald–Giemsa (MGG) stained smeared material and IHC on CS.

Materials and Methods The material was a consecutive series of cytological material from 85 patients admitted to the Department of Respiratory Diseases, Gentofte University Hospital, Copenhagen during the period May 2008 to August 2010. The patients had been referred to the hospital with infiltrates/tumors of the lung or possible metastases. The aspirated material was smeared onto glass slides, airdried, and stained with MGG for cytologic evaluation. To be included in this study, at least two cellular slides were required to be available for diagnostic purposes. From each aspirate the two most optimal slides were selected by one of the authors (BGS). This selection was mainly based on the preservation state of the cells and a high number of malignant tumor cells. One of these slides was used for morphological evaluations of the cytologic material. The other slide(s) was (were) used for making a CS. The CS were 106

Diagnostic Cytopathology, Vol 42, No 2

prepared in the following way: after the slides were read, the cells were gently scraped off the slides with a surgical blade. No destaining was done prior to this procedure. Clots were made by taking the cells from the blade and putting them into tissue cassettes covered by Millipore paper. In some cases three drops of human plasma and one drop of thrombin were added to the cells to form a clot. Afterwards the cells in the cassettes were processed as routine paraffinembedded material.14 In six cases, no malignant cells were present in the CS and these cases were excluded from the study. Thus, a series of MGG stained smears and CS from 79 patients was included in the present study. The material derived from FNA from the lung (N 5 30), EBUS and EUS from mediastinal lymph nodes/tumors (N 5 42) and cytologic material from other regions (liver, pleura and pericardium) (N 5 7). The CS were stained by haematoxylin-eosin and a panel of immunohistochemical antibodies considered relevant after cytology evaluation and according to clinical information including previous malignant diseases. This evaluation was performed by one of the authors (BGS), an experienced cytopathologist who was not included in the reproducibility part of the study. Mucin stain and TTF1, CK7, CK5/6, and p63 were applied to the majority of CS (Figs. 1,2). The following markers were applied in a variable number of cases according to the clinical information and the cytomorphological features on the smears: CD56, CK18, CK19, CK20, ER, CD45, HMB45, Melan-A, Chromogranin, Synaptophysin, Vimentin, S100, CD20, CEAM, KI67, PSA, CK AE1/AE3, CK 34BE12, CDX2. On average, four immunohistochemical stainings per case were available. For all immunohistochemical staining the standard method from our laboratory was applied. The sets of slides (the MGG stained slide and CS with IHC from each patient) were coded randomly and circulated twice with 2 months between the rounds to four pathologists. One pathologist was inexperienced with 6 months’ experience in pathology and limited experience in lung pathology, and three were very experienced lung pathologists with a minimum of 10 years’ experience within this field. The inexperienced pathologist was included in the panel to assess the influence of experience. There was a short learning session between the two rounds for the inexperienced pathologist. This session was prepared and directed by the pathologist not involved in the reproducibility study (BGS) and consisted of a short introduction to diagnostic lung cytology including reading of IHC on CS. For the experienced pathologists, a short instruction session was conducted between the two rounds on how to read the IHC on CS as these pathologists were unfamiliar with the use of CS. As described,14 the morphology of the malignant cells in CS is different from the morphology in ordinary


REPRODUCIBILITY OF CYTOPATHOLOGIC DIAGNOSES

Fig. 1. Adenocarcinoma, MGG stained smear. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Fig. 3. SQC, MGG stained smear. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Fig. 2. Same case as Fig. 2, CS, TTF1 positive. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

Fig. 4. Same case as Fig. 3, CS, CK5/6 positive. [Color figure can be viewed in the online issue, which is available at wileyonlinelibrary.com.]

clots. Each observer was informed that the material was from a patient suspected of lung cancer and the sampling region was stated (Figs. 3,4). The pathologists were presented with both the MGG stained smear and immunohistochemically stained CS material in both rounds, and they were asked to classify each set of slides for every single patient included in the study (MGG stained smeared material and CS with IHC) in one of the following categories: benign-granuloma, benign nongranuloma, malignant tumor (NOS), carcinoma (NOS), NSCLC (NOS), SCLC, combined carcinoma, SQC, adenocarcinoma, non-SQC, large cell carcinoma, large cell neuroendocrine carcinoma, sarcomatoid carcinoma, salivary gland tumor, carcinoid tumor, lymphoma (NOS), malignant melanoma, sarcoma (NOS) and others. The four pathologists were asked to diagnose in accordance with cytopathologic guidelines.15 Several studies

have demonstrated that applying IHC to cellblocks improves diagnoses based on histological as well as on cytopathologic material.9–11 The aim of this study was however to assess reproducibility and not accuracy of diagnosis. For statistical purposes all answers were categorized in five groups: (1) SQC, (2) adenocarcinoma of the lung, (3) non-SQC, (4) benign lesion and (5) malignancy other than the former, including metastases from other organs. For the second round, all sets of slides (MGG stained smeared material and CS stained by IHC) were recoded randomly. The pathologists made their diagnoses without knowing the original diagnosis, and (for the second round) without knowing their own or the others’ diagnoses from the first round. The study did not require approval from The Danish Committee on Biomedical Research Ethics. Diagnostic Cytopathology, Vol 42, No 2

107


OMLAND ET AL. Table I. Numbers of Diagnoses in Five Categories by the Four Pathologists in the First and Second Round

Observer A Observer B Observer C Observer D

Round

Adenocarcinoma of the lung

1 2 1 2 1 2 1 2

29 34 25 23 27 31 33 40

SQC

Nonsquamous

Benign

Other malignant

9 11 12 11 13 12 9 12

0 0 0 2 0 0 0 0

1 0 0 0 1 1 1 1

40 34 42 43 38 35 36 26

Statistics/Analysis The Kappa statistic was applied for analysis of the interobserver variation of the assessment of the diagnoses using the statistic package JMP v. 6 (SAS, Cary, NC).16 For measurement of concordance between the four pathologists the evaluation of each pathologist was compared with those of each of the other members of the panel. For each pair of observers a diagnostic table was composed and the specific Kappa value was calculated. The Kappa statistic is a widely used index for measuring chance-corrected agreement on a nominal scale.16 Values close to one denote almost perfect agreement and a value greater than 0.80 is for general purposes representative of excellent agreement. Values between 0.80 and 0.60 represent very good agreement, values between 0.60 and 0.40 represent good agreement, values between 0.40 and 0.20 represent fair to good agreement and finally a value below 0.20 represents poor agreement. The Kappa value reflects the reproducibility alone and gives no indication of how many cases are affected by the disagreement between observers.

Results The diagnoses in the five categories by the four observers in the first and second rounds are summarized in Table I. Category 3, non-SQC, was not used at all in round 1 and only used by one observer twice in round 2. Kappa value and 95% confidence intervals for the diagnoses in five categories by the four observers in the first and second rounds are summarized in Table II. The Kappa coefficients in the first round were good to very good (Kappa 0.57–0.71). The only value under 0.60 involved the inexperienced pathologist (observer D). In the second round Kappa values were very good (0.63–0.80) and four out of six kappa values were higher in round 2 although not statistically significant. The lowest Kappa value (0.63) in round 2 involved the inexperienced pathologist.

Discussion In this study, we demonstrated that even if only sparse cytologic material is available, the reproducibility of 108


Table II. Overall Kappa Value and 95% Confidence Interval (CI) for the Four Pathologists in the First and Second Round

Observer

Observer

Kappa value round 1

A A A B B C

B C D C D D

0.70 0.71 0.62 0.69 0.65 0.57

95% CI round 1

Kappa value round 2

0.56–0.84 0.57–0.85 0.47–0.77 0.55–0.83 0.50–0.79 0.41–0.73

0.67 0.80 0.69 0.74 0.63 0.66

95% CI round 2 0.54–0.81 0.68–0.92 0.55–0.83 0.61–0.86 0.49–0.77 0.52–0.80

clinical relevant diagnoses is very good, when smeared MGG stained material is combined with IHC on CS. Reproducibility was 0.57–0.71 in the first round and 0.63–0.80 in the second round. Although FNA in many cases provides the pathologist with enough material including clots for IHC for diagnostic purposes, this is not always the case. The method described here can be used as a supplement when there is insufficient aspirated material to make blood clots from the cannula or syringe. The final diagnoses should be made in combination with conventional smears in which the morphology is better preserved. All kinds of smeared material are useful for making CS. The limiting factor is the number of malignant cells present on the smear. Smears with only a few malignant cells are unsuitable for making a CS. Several immunohistochemical markers, including TTF1, CK7, P63, CK5/6, desmocollin, p40, and 34betaE12, have been investigated for NSCLC subtyping, and IHC has been shown to reduce the number of inconclusive diagnoses (NSCLC-NOS) to a variably low percentage. However, as only sparse material often is available for morphologic diagnosis as well as for molecular diagnosis including testing for EGFR mutations and EML4-ALK translocation, it is crucially important that the tissue is handled optimally in terms of doing clinical relevant analyses on the right material. For this reason, a small panel of markers has been proposed.1,9,17,18 In terms of sensitivity and specificity for subclassification of NSCLC into adenocarcinoma and SQC, the relative expression of TTF-1 and p63 seems to provide the most acceptable results on cytological block material as well as on small biopsies.10,11 On the basis of this data it seems reasonable to limit immunophenotyping of NSCLC to between two and four markers.1 This was also the case in this study in which only sparse material was available. In cases where there was suspicion of metastatic disease in the lung, IHC with other antibodies including ER, CD20, CDX2, PSA, MelanA among many others should be applied according to the morphology and clinical information. Our personal experience is that all relevant antibodies can be applied to CS, and even molecular testing including testing for EGFR mutations and EML4-ALK translocation can be performed on CS (data not shown).


REPRODUCIBILITY OF CYTOPATHOLOGIC DIAGNOSES

Non-SQC is a diagnosis that should be avoided by pathologists. It is a category used by clinicians to define groups of patients that can be treated in a similar manner despite several histological types. In this study the category is only used twice. Paech DC et al.19 recently presented a systematic review of the interobserver variability in the differentiation between squamous and nonsquamous NSCLC in histology. Six studies were included with Kappa values ranging from 0.48 to 0.88. The included material was biopsies and resection material, with one report also including cytologic material. The material was stained with haematoxylin-eosin, with three studies allowing mucin stains on request but no immunohistochemical staining. The authors concluded that on the basis of morphology alone, pathologists can reliably classify the great majority of NSCLC into the broad groups of nonsquamous and squamous NSCLC, but only moderate agreement was found in the subclassification of NSCLC. This underlines the need for further pathology tools, including IHC to help in the clinically important subtyping of NSCLC. The usefulness of a diagnostic test rests with its clinical significance, its accuracy, and its reproducibility. Regarding clinical significance, numerous studies indicate that FNA including EUS-FNA and EBUS-TBNA of mediastinal lymph nodes is a safe, minimally invasive technique improving the clinical diagnoses and staging of patients with NSCLC and of patients suspected to have metastases to the lung.13,20–29 The accuracy of a diagnostic test, positive predictive value and negative predictive value has been the subject of many studies.13,30,31 However, when evaluating the accuracy of a diagnostic test, a gold standard is needed. In many cases this can be defined only with great difficulty. The purpose of this study was not to evaluate accuracy. What can be evaluated and what is necessary to investigate for the usefulness of a diagnostic test is reproducibility. Reproducibility is defined as closeness of agreement among repeated measurements under the same operating conditions. Interobserver variation has been estimated for diagnoses made on cytological as well as on histological material.6,32–34 Variation exists in most, probably all, diagnostic tests. In a reproducibility study the question is whether the interobserver variation for the diagnostic test is acceptably small so that it can be relied on in the daily routine. In this study, an improvement in four out of six Kappa values from the first to the second round was found, though not statistically significant. There was no significantly higher improvement for the inexperienced pathologist than for the experienced observers. In both rounds the lowest Kappa value involved the inexperienced observer, which reflects the importance of experience. As there were training sessions between the two rounds of assessment we did

not analyze intraobserver reproducibility (successful training might in effect depress the intraobserver agreement). The diagnostic situation in this study was different from the daily routine in that the pathologists had CS with a varying number of IHC and only one slide stained by MGG (preselected among all available diagnostic smears) per aspirate. This selection of slides was introduced in order to reduce the time constraints for the study. However, the diagnostic slides had been preselected by a very experienced lung (cyto)pathologist, and all available material was identical for the four pathologists. Thus, the test situation should be regarded as optimal for a reproducibility study. If the aim of this study had been an evaluation of the accuracy of the method, such a reduction of the evaluated material might have led to false-negative diagnoses. However, the aim of the study was to analyze reproducibility. Whether having more slides improves the interobservation of pathologic diagnoses has yet to be tested. From our point of view it seems reasonable to believe, that more slides per case would increase the probability of obtaining a good diagnostic slide, thereby improving our results. In conclusion, the reproducibility of subtyping NSCLC on sparse cytologic material that includes one MGG-stained smear and IHC on CS is very good among experienced cytopathologists. When insufficient cytologic material is available for making a clot directly from the aspirate, CS is a method to achieve material for IHC, useful for making clinical relevant diagnoses, including the important subtyping of NSCLC.

References 1. 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;6:244–285. 2. Mok TS, Wu YL, Thongprasert S, et al. Gefitinib or carboplatinpaclitaxel in pulmonary adenocarcinoma. N Engl J Med 2009;361:947–957. 3. Ku GY, Haaland BA, de Lima LG, Jr. Gefitinib vs. chemotherapy as first-line therapy in advanced non-small cell lung cancer: Meta-analysis of phase III trials. Lung Cancer 2011 Dec;74: 469–473. 4. Scagliotti GV, Parikh P, von PJ, et al. Phase III study comparing cisplatin plus gemcitabine with cisplatin plus pemetrexed in chemotherapy-naive patients with advanced-stage non-small-cell lung cancer. J Clin Oncol 2008;26:3543–3551. 5. Sandler A, Gray R, Perry MC, et al. Paclitaxel-carboplatin alone or with bevacizumab for non-small-cell lung cancer. N Engl J Med 2006;355:2542–2550. 6. Skov BG, Baandrup U, Jakobsen GK, et al. Cytopathologic diagnoses of fine-needle aspirations from endoscopic ultrasound of the mediastinum: Reproducibility of the diagnoses and representativeness of aspirates from lymph nodes. Cancer 2007;111:234–241. 7. Edwards SL, Roberts C, McKean ME, Cockburn JS, Jeffrey RR, Kerr KM. Preoperative histological classification of primary lung cancer: Accuracy of diagnosis and use of the non-small cell category. J Clin Pathol 2000;53:537–540. Diagnostic Cytopathology, Vol 42, No 2

109


OMLAND ET AL. 8. Burnett RA, Swanson BJ, Howatson SR, et al. Observer variability in histopathological reporting of malignant bronchial biopsy specimens. J Clin Pathol 1994;47:711–713.

22. Bardales RH, Stelow EB, Mallery S, Lai R, Stanley MW. Review of endoscopic ultrasound-guided fine-needle aspiration cytology. Diagn Cytopathol 2006;34:140–175.

9. 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;35:15– 25.

23. Varela-Lema L, Fernandez-Villar A, Ruano-Ravina A. Effectiveness and safety of endobronchial ultrasound-transbronchial needle aspiration: A systematic review. Eur Respir J 2009 May;33:1156– 1164.

10. 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;5:436–441. 11. Loo PS, Thomas SC, Nicolson MC, Fyfe MN, Kerr KM. Subtyping of undifferentiated non-small cell carcinomas in bronchial biopsy specimens. J Thorac Oncol 2010;5:442–447. 12. 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;6:451–458. 13. Tournoy KG, Carprieaux M, Deschepper E, van Meerbeeck JP, Praet M. Are EUS-FNA and EBUS-TBNA specimens reliable for subtyping non-small cell lung cancer? Lung Cancer 2012;76:46–50. 14. Skov BG, Kiss K, Ramsted J, Linnemann D. A technique to improve diagnostic information from fine-needle aspirations: Immunohistochemistry on cytoscrape. Cancer Cytopathol 2009;117:120– 127. 15. Johnston WW, Elson CE. Respiratory tract. In: Bibbo M, Wilbur D, editors. Comprehensive cytopathology. 3rd ed. Elsevier: Saunders; 2008. p 303–372. 16. Cohen J. A coefficient of agreement for nominal scale. Cohen J Educ Psychol Meas 1960;20;37–46.

24. Levine MS, Weiss JM, Harrell JH, Cameron TJ, Moser KM. Transthoracic needle aspiration biopsy following negative fiberoptic bronchoscopy in solitary pulmonary nodules. Chest 1988;93:1152– 1155. 25. Annema JT, Versteegh MI, Veselic M, Voigt P, Rabe KF. Endoscopic ultrasound-guided fine-needle aspiration in the diagnosis and staging of lung cancer and its impact on surgical staging. J Clin Oncol 2005;23:8357–8361. 26. Eloubeidi MA, Tamhane A, Chen VK, Cerfolio RJ. Endoscopic ultrasound-guided fine-needle aspiration in patients with non-small cell lung cancer and prior negative mediastinoscopy. Ann Thorac Surg 2005;80:1231–1239. 27. Detterbeck FC, Jantz MA, Wallace M, Vansteenkiste J, Silvestri GA. Invasive mediastinal staging of lung cancer: ACCP evidencebased clinical practice guidelines (2nd edition). Chest 2007;132(3 Suppl):202S–220S. 28. Tournoy KG, Annema JT, Krasnik M, Herth FJ, van Meerbeeck JP. Endoscopic and endobronchial ultrasonography according to the proposed lymph node map definition in the seventh edition of the tumor, node, metastasis classification for lung cancer. J Thorac Oncol 2009 Dec;4:1576–1584. 29. Annema JT, van Meerbeeck JP, Rintoul RC, et al. Mediastinoscopy vs endosonography for mediastinal nodal staging of lung cancer: A randomized trial. JAMA 2010;304:2245–2252.

17. Righi L, Graziano P, Fornari A, et al. Immunohistochemical subtyping of nonsmall cell lung cancer not otherwise specified in fineneedle aspiration cytology: A retrospective study of 103 cases with surgical correlation. Cancer 2011;117:3416–3423.

30. 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;6:451–458.

18. Rosell R, Perez-Roca L, Sanchez JJ, et al. Customized treatment in non-small-cell lung cancer based on EGFR mutations and BRCA1 mRNA expression. PLoS One 2009;4:e5133.

31. Rivera MP, Mehta AC. Initial diagnosis of lung cancer: ACCP evidence-based clinical practice guidelines (2nd edition). Chest 2007;132(3 Suppl):131S–148S.

19. Paech DC, Weston AR, Pavlakis N, et al. A systematic review of the interobserver variability for histology in the differentiation between squamous and nonsquamous non-small cell lung cancer. J Thorac Oncol 2011;6:55–63.

32. Fassina A, Fedeli U, Corradin M, Da FM, Fabbris L. Accuracy and reproducibility of pleural effusion cytology. Leg Med (Tokyo) 2008;10:20–25.

20. Stoll LM, Yung RC, Clark DP, Li QK. Cytology of endobronchial ultrasound-guided transbronchial needle aspiration versus conventional transbronchial needle aspiration. Cancer Cytopathol 2010;118:278–286. 21. Cameron SE, Andrade RS, Pambuccian SE. Endobronchial ultrasound-guided transbronchial needle aspiration cytology: A state of the art review. Cytopathology 2010;21:6–26.

110


33. Noguchi M, Minami Y, Iijima T, Matsuno Y. Reproducibility of the diagnosis of small adenocarcinoma of the lung and usefulness of an educational program for the diagnostic criteria. Pathol Int 2005;55:8–13. 34. Skov BG, Lauritzen AF, Hirsch F, Nielsen HW. The histopathological diagnosis of malignant mesothelioma v. pulmonary adenocarcinoma: Reproducibility of the histopathological diagnosis. Histopathology 1994;24:553–557.