TIMELY REVIEWS Section Editor: Liron Pantanowitz, M.D.
EGFR Analysis: Current Evidence and Future Directions Claudio Bellevicine, M.D., Umberto Malapelle, Ph.D., Caterina de Luca, B.S., Antonino Iaccarino, Ph.D., and Giancarlo Troncone, M.D.,
Until a few years ago, only lung cancer histological specimens were considered suitable for testing epidermal growth factor receptor (EGFR) mutations. Then, several retrospective studies were designed to test EGFR mutation on a sizeable number of parallel cytological and histological samples obtained from the same patients and, even more recently, several institutions reported their prospective clinical experiences on routine specimens. Basing on these studies the College of American Pathologists, the International Association for the Study of Lung Cancer, and the Association for Molecular Pathology have recently considered cytological samples suitable for EGFR testing. Therefore, it seems timely to draw together the threads of this large body of information in order that cytopathologists can be knowledgeable partners in the multidisciplinary process of targeted cancer therapy and to help refine current testing guidelines. This review addresses (1) the more common proposed techniques including the use of direct cytologic smears cell blocks and liquid based cytology; (2) the issues related to current practice, which in Europe is external centralized testing that is usually done on samples containing very few cells; and (3) the future directions based on the implementation on lung cytology of next generation sequencing approaches. Diagn. Cytopathol. 2014;00:000–000. C 2014 Wiley Periodicals, Inc. V
Key Words: lung cancer; cytology; aspiration; EGFR; next generation sequencing
The Epidermal Growth Factor Receptor (EGFR) is a major target in the treatment of advanced-stage non-small cell lung cancer (NSCLC). The first generation of EGFR tyroDepartment of Public Health, Anatomic Pathology Section, University of Naples Federico II, Naples, Italy *Correspondence to Giancarlo Troncone, MD, PhD, Department of Public Health, Universita di Napoli Federico II, via Sergio Pansini 5, I-80131 Napoli, Italy. E-mail: [email protected] Received 23 April 2013; Revised 10 January 2014; Accepted 24 February 2014 DOI: 10.1002/dc.23142 Published online 00 Month 2014 in Wiley Online Library (wileyonlinelibrary.com). C 2014 WILEY PERIODICALS, INC. V
Ph.D.*
sine kinase inhibitors (TKI) included gefitinib (Iressa; AstraZeneca, London, UK) and erlotinib (Tarceva; Genentech, South San Francisco, CA, and OSI Pharmaceuticals, Long Island, NY).1,2 The efficacy of this treatment, as monotherapy in the first line setting, is restricted to those NSCLC harboring EGFR mutations.1 In particular, the two most common mutations in EGFR, the short in-frame deletions in exon 19 and the L858R point mutation in exon 21, are strongly correlated with improved quality of life, a high response rate (RR) and a longer progression-free survival (PFS). EGFR mutations are more commonly associated to female gender, adenocarcinoma (ADC) morphology, and Asian ethnicity in the absence of a smoking history.2,3 The first indications for anti-EGFR therapies are the locally advanced, recurrent, or metastatic disease settings; since these patients are not surgical candidates, tissue sampling for molecular testing through a minimally invasive approach, such as cytology, is ideal. As recently as 2009, however, a consensus statement from the European EGFR Workshop Group recommended that clinicians should be encouraged to provide tissue biopsies rather than cytological samples whenever possible, as further research was still necessary to fully understand the clinical reliability of mutational data obtained from cytological specimens.4 At that time, with a few rare notable exceptions,5,6 most of the reported institutional experiences with EGFR testing included only a few cytological samples. The small amount of tissue of most specimens, the need of optimizing cyto-preparation, the interference of background non-neoplastic cells, the necessity of sensitive molecular techniques and the lack of in vitro diagnostic tests licensed for cytology generated the diffuse perception that cytological material was unsuitable for effective molecular analysis and only sufficient to provide a generic diagnosis of NSCLC.7 Diagnostic Cytopathology, Vol. 00, No 00
1
Diagnostic Cytopathology DOI 10.1002/dc
BELLEVICINE ET AL.
this evidence, the international guidelines for lung cancer biomarker testing issued in 2013 by the College of American Pathologists (CAP), the International Association for the Study of Lung Cancer (IASLC), and the Association for Molecular Pathology (AMP) considered cytological samples suitable for EGFR testing.17 Hence, it seems the time is now ripe to review the more common techniques involving the use of different cytological preparations, the issues related to external centralized testing and the next generation sequencing approaches that hold promise for even further refinement of personalized treatment regimens for patients with lung cancer.
To eliminate such a perception and to assess the clinical reliability of mutational data obtained from cytological specimens and in particular, whether the mutation status assessed on cytological samples represented the mutation status of the entire tumor, retrospective studies were designed to test EGFR mutation on a sizeable number of parallel cytological and histological samples obtained from the same patients. Table I reports a list of these studies.8–15 The degree of concordance between paired samples was very high but not absolute, ranging between 82% and 100%. Discrepancies were more likely to reflect technical issues related to the different pre-analytical preparation rather than genuine intratumoral EGFR mutational heterogeneity. A detailed study by Yatabe et al., who tested 50 tumors in three different areas and 5 into 100 points, demonstrated that intratumoral EGFR mutational heterogeneity is extremely rare in lung adenocarcinomas.16 Accordingly, EGFR mutation patterns between primary and metastatic sites and between primary and recurrent neoplasms are similar;17 hence, testing of multiple tissue areas within a single tumor is not necessary.17 It thus conceivable that mutations detected on a cytological sample are representative of the entire tumor and, therefore, valuable to predict the response to treatment. Quite recently, several institutions reported their prospective clinical experiences on routinely collected cytological specimens.8,18–23 In general, the methodologies were sufficiently robust to test satisfactorily most of the routine cases; even limited cytological materials containing a small amount of extractable DNA were reliably tested.18 Different techniques, including direct sequencing and real time PCR (RT-PCR) based methods, applied to different sample types, including exfoliative and aspirative samples, and preparations, such as direct smears, cell-blocks (CB) and monolayer preparations, yielded data comparable to those obtained on larger tumor biopsies in terms of frequency and distribution of EGFR mutations according to sex, histology, and prior smoking history8,18–23 Basing on
EGFR Testing on Smears The smears used for diagnosis may be the only source of tumor cells testable for molecular studies, in cases with a small quantity of tissue material or when a cell-block is not available. Whenever a single slide is available, the molecular testing will destroy the evidence of tumor cells, which might have medico-legal consequences. To manage this risk, smears can be photographed or digitally scanned, to record the cytomorphology of a representative microscopic field for the archives.24,25 The DNA obtained from direct smears is of better quality than that extracted from matched cell blocks.26 Killian et al. reported that Diff-Quik should be preferred to Papanicolaou-stained slides.24 More recently, however, in cell line experiments, Papanicolaou-stained smears yielded optimal DNA yield and fragment length.27 As a matter of the fact, the studies performing EGFR testing on routine Diff-Quik9,23,28 or Papanicolaou8 smears did not show significant differences. The process of removing the coverslip of archival smears do not compromise the quality of the DNA isolated for molecular studies, but it is time-consuming. To avoid any delay, rapid on site evaluation (ROSE), at the time of the FNA procedure, can select a representative slide, that it is maintained uncoverslipped for immediate DNA extraction and EGFR
Table I. Summary of Studies That Analyzed EGFR Mutational Status in Paired Histological and Cytological Samples References
Type of cytological sample 8
Paired samples (n)
EGFR detection methods
Concordant/total (%) 100 100 100 100
Lozano et al. Bozzetti et al.9 Bruno et al.10 Chowdhury et al.11
Direct smears Direct smears or needle washing Direct smears Cell Block
20 9 50 12
Goto et al.12
Cell pellet from Bronchial Brushing
30
Mitiushkina et al.13
Direct smears
72
Sun et al.14 Kode et al.15
Direct smears and/or cell blocks Direct smears from the same site of the histology Direct smears from a different site
60 37
Direct sequencing Direct sequencing Direct sequencing LCM and direct sequencing or pyrosequencing PNA-LNA PCR clamp or ARMSSkorpion Exon 19 Fragment length assay; Exon 21 Allele-specific real-time Pyrosequencing Pyrosequencing
23
Pyrosequencing
Kode et al.15 a
29/30 case were evaluated; one test failure in histology sample. Discordant cases included two sputum specimens and three pleural effusions
b
2
Diagnostic Cytopathology, Vol. 00, No 00
96.6a 68/72; 94.4 55/60; 91.7b 36/37; 97 19/23; 82
Diagnostic Cytopathology DOI 10.1002/dc
EGFR TESTING ON CYTOLOGY
testing.26 Interestingly, da Cunha Santos et al. applied the “freezer method” to remove coverslips, reducing the time required for getting the slides ready for DNA extraction.29 The neoplastic can be scraped off with a tip of an insulin syringe needle, while examining the smear at the microscope.5,30,31 In their practice, Billah et al. carefully mapped the slides to define the location of tumor fragments or areas of higher tumor density, which enabled testing 93 of 99 archival smears.20 As a general rule, slides prepared from fine needle aspirates are easier to be microdissected than exfoliative cytology smears.32 In these latter, benign and neoplastic cells are more likely to co-exist, making tumor cells selection difficult. Manual microdissection, however, has intrinsic limitations. Lozano et al.8 only analyzed smears with >50% tumor cellularity, while Pang et al.22 were not able to manual microdissect 12 of 147 (8.1%) smears. In attempt to make manual microdissection more feasible, the novel cell-transfer technique was proposed.33 This technique employs the mounting media to detach the neoplastic area marked on the smears excluding non-neoplastic contaminant. Alternatively, laser capture microdissection (LCM) enables selective cancer cell dissection. In ours34 and other’s hands6,35 only 25 cancer cells are needed to reliably detect EGFR mutations. However, LCM requires specific training and it is very time consuming is an expensive technology that is not available to all laboratories. In a near future, computer-aided LCM will semi automate the microdissection process, allowing a more rapid retrieval of tumor cells, saving pathologist’s time and effort.36 Although morphology cannot reliable predict TKI responsiveness, Brechtel et al reported that on smears the cytomorphological features associated to former bronchioloalveolar carcinoma (BAC) such as flat monolayers, intranuclear cytoplasmatic inclusions, the absence of large nucleoli and extracellular mucin are linked to the presence of mutated EGFR cells.37 More recently, Marotti et al. confirmed these observations, also showing a link
between poor tumor differentiation, necrosis and squamoid features and the occurrence of KRAS mutations.38
EGFR Testing on Cell Blocks Cell block preparations allows for the obtainment of multiple serial sections for ancillary studies and a banked archive for future retrospective testing as new agents become available;17 furthermore, cell blocks are irreplaceable when the aspirated material, after the prolonged echo-endoscopic bronchial (EBUS) or digestive (EUS) procedures is clotted.39,40 Cell blocks may also be useful to provide useful morphological clues; for example, intercellular bridges can be highlighted on cell blocks, thus allowing for classification of otherwise non-small cell carcinomas as squamous cell carcinomas.41 Remarkably, cell blocks are recommended over smear preparations, by the recent CAP/IASLC/AMP guidelines.17 This may probably reflect, in part, the fact that most molecular pathologists may prefer dealing with preparations similar to those of traditional histological blocks. The disadvantages of cell blocks are also well known (Table II). The adequacy for tumor cells cannot be assessed immediately and deeper sections from the cell block may not contain the tumor cells observed on the originally prepared hematoxylin & eosin-stained slide.26 Cutting the cell block only once yielding sections both for microscopy and for genomic studies would be ideal to save as much tumor tissue as possible for molecular testing.42 Cell block cellularity may be variable, being dependent on the number of dedicated fine needle aspiration (FNA) passes and on the post-procedural handling of the needle rinse specimen.43 In routine practice, not infrequently cell blocks are either acellular or sparsely cellular.26 In the series reported by Pang et al. 8/51 (15%) cell blocks prepared from FNAs were rejected,22 while in the study by Cai et al the rejection rate was lower (9%).21 Since cell blocks are prepared from a pooled specimen, the neoplastic cells obtained by high yield needle passes are diluted by background benign elements from off target passes; as a consequence, tumor cell enrichment is cumbersome and LCM often necessary.11
Table II. Advantages and Disadvantages of Different Cytological Preparations for EGFR Mutation Testing Direct Smears
Cell Blocks
Liquid-Based Cytology
Advantages
High quality DNA; visualization of malignant cells; ROSE feasible
Diagnostic smears preserved TTF1 positive cells can be selected; recommended by CAP/IASLC/AMP Guidelines; useful for EBUS and EUS
Disadvantages
Loss of diagnostic material; delay due to coverslip removal; LCM may be needed
Poor DNA quality; ROSE unfeasible
High quality DNA; eliminates the need for slide preparation by clinicians; material maximized; sensitive methods can directly be applied to CytoLyt derived DNA ROSE unfeasible
Diagnostic Cytopathology, Vol. 00, No 00
3
Diagnostic Cytopathology DOI 10.1002/dc
BELLEVICINE ET AL.
EGFR Testing on Liquid Based Cytology Liquid-based cytology (LBC) makes material collection easier and eliminates the need for slide preparation by clinicians, which is valuable when it is not possible to have on-site an experienced cytopathologist.44 Additional advantages are reduction of background material such as blood and debris and the possibility of dividing the sample equally on several generally identical slides. LBC preparations are derived from homogenized, pooled liquid-based samples and thus, alike cell blocks, they are unsuitable for on-site evaluation and they may have variable neoplastic cell content.26 The most widespread used LBC fixatives are the methanol-based CytoLyt solution (Cytyc Corporation, Marlborough, MA) and the CytoRich Red collection fluid (Fisher Scientific UK Ltd, Loughborough, UK), that contains a small amount (0.4%) of formalin. In a recent study, we managed to extract DNA that was suitable for EGFR mutation analysis in CytoLyt-fixed clinical cytological samples from patients with non-small cell lung cancer.34 More recently two different studies both using human lung carcinoma cell lines carefully evaluated the effect of LBC fixatives on DNA quality. CytoLyt solution provides a higher DNA yield than CytoRich Red collection fluid,27 while both fixatives well preserved the DNA.27,45 The pre-analytical management of LBC sample depends on the sensitivity of the mutational assay. PCR and direct sequencing can reliably detect EGFR mutations only when slides are prepared from the collection fluid and when a pure population of neoplastic cell is selected by LCM.46 Conversely, LCM and direct sequencing may be replaced by more sensitive non-sequencing RT-PCR based assays directly performed on CytoLyt derived DNA.34 Malignant pleural effusion (MPE) is a frequent complication of peripheral lung adenocarcinoma.47 Testing EGFR on MPE obviates the need of tissue sampling. From a biological point of view, anchorage-independent and movable tumor cells from MPE are more representative of the invading tumor than a pleural nodule biopsy.48 However, technical issues are relevant. Kimura et al., extracted the DNA from 1.0 ml of pleural effusion fluid; mutations were detected even when the DNA yields were below the concentration measurable by spectrophotometry. It is conceivable that mutations can be detected in soluble DNA, due to the release of tumor-derived DNA into the pleural fluid through the apoptosis or necrosis of disseminated tumor cells.49,50 However, since the composition of MPE includes normal, inflammatory and reactive mesothelial cells, direct sequencing has an insufficient sensitivity. Strategies to avoid false-negative results include either the visualization of neoplastic cells on smears or cell blocks before DNA extraction or more sen4
Diagnostic Cytopathology, Vol. 00, No 00
sitive molecular methods. Noteworthy, the EGFR mutation status of MPE does not always re-capitulate that of the primary tumor tissue. Han et al. showed that 3 of 21 patients (14.3%) had a discordant MPE EGFR mutation status, probably reflecting technical issues.48
EGFR Testing in Centralized Laboratories The cytopathologist plays different roles in lung cancer biomarker testing. He is responsible of the preanalytical phases of testing; he has to review representative cell block sections or select best quality smears, to determine the cellularity and purity (mucin and necrosis may inhibit amplification) of the tumor sample being submitted for biomarker testing.43 Since various mutational assays have different analytic sensitivities, the cytopathologist (or the technician) should enrich for tumor content to a level that is acceptable for the assay being used.43 The cytopathologist may perform the molecular testing in-house, which is a frequent procedure in the United States,51 supervising the activity of his/her Clinical Laboratory Improvement Amendment (CLIA) accredited intramural molecular pathology laboratory, Conversely, in Europe not all pathology departments are equipped to run EGFR testing and a greater number of cytopathologists refers to outside molecular pathology laboratories.4 When determining what laboratory to use for outside EGFR testing, the cytopathologist should bear in mind the requirements that the CAP/IASLC/AMP guidelines set for testing laboratories.17 These have to ensure the use of validated methodologies able to detect, even in specimens with as little as 10% cancer cells, not only the two major mutations but all those occurring with a frequency of at least 1% of EGFR-mutated lung adenocarcinomas within a time around testing of 10 working days. Once the results of molecular analysis are received, the cytopathologist may compare the mutation signals with the extent of tumor cells in the tested specimen, carefully evaluating the quality processes employed to ensure confidence in the results.43 The minimal number of annual EGFR tests sufficient for guaranteeing the technical sufficiency of a laboratory was arbitrarily set at 50.52 The EGFR European Workshop Group recommend reference laboratories to be accredited either at the national or at the European level.4 In Italy, Greece, Holland, and Germany the national society of pathology organizes ring trials for external quality assurance assessment once or twice a year.17 The schemes organized by the European Society of Pathology evaluate not only EGFR but also KRAS and ALK biomarker testing in NSCLC. Similarly, a proficiency testing for EGFR and ALK testing is offered by the CAP. However, to date, only histological samples have been included in the proficiency testing schemes and assessment of the quality of testing on cytology remains problematic.
Diagnostic Cytopathology DOI 10.1002/dc
EGFR TESTING ON CYTOLOGY
External centralized testing is usually done on samples containing very few cells.19,20,22 In a reference molecular laboratory, the presence of both a pathologist, who is highly trained in selecting the best cytological specimen while considering the specific methods used by his laboratory, and a dedicated technician, who allows expeditious processing of cytological samples for molecular testing, is crucial, but not sufficient. In our experience, the smear sent to our centralized laboratory is not always the “good one” and is often paucicellular, since the cytopathologist who diagnosed the case is often reluctant “to sacrifice” the morphology of malignant cells for DNA extraction.25 In our laboratory practice, 16.2% (59 of 364) of the cytological samples were rejected because contained less than 25 neoplastic cells.25 In the study by Billah et al. in-house collected cytological material were more often (96, 3%; 160 of 166) adequate than outside cases from referred patients (83, 7%; 36 of 43).20 Pang et al. reported that 10% of outside samples were inadequate22 and despite the use of a very sensitive methodology (TheraScreen EGFR29 kit), 14% of samples received by Allegrini et al. were deemed inadequate.19 Thus, externalizing EGFR testing may be associated with a higher rate of inadequate samples. One of the basic requirements to improve specimen adequacy from participating sites is educational and efforts should be spent to provide the essential concepts of lung molecular pathology, whose understanding is necessary to properly handle and manage cytopathology specimens for molecular testing. To investigate the degree of reliability of treatment prediction based on EGFR mutational diagnosis performed by a centralized laboratory on cytological material we reviewed our records of the clinical outcomes of patients affected by lung cancer and compared the benefit of treatment in two groups of patients, the only between-group difference being the type of sample (histology or cytology) tested.25 In patients tested on histology the RR was 54% and the PFS 9.2 months. Patients selected on cytological samples had a similar outcome (RR 5 62%, PFS 5 8.6 months). Consequently, despite concern that less-than-optimal samples are sent to reference molecular laboratories, cytological samples can reliably predict treatment outcome. Thus, not only EGFR mutations can be reliably detected on paucicellular lung cancer cytology specimens,6,11,14,19,34,35 but also the results can reliably predict the efficiency of gefitinib treatment.25
Immunocytochemistry and EGFR Testing: Sample Prioritization Although knowledge of the specific NSCLC subtype would be essential prior to prescribing a number of important new drugs to treat NSCLC, including both cytotoxic chemotherapeutic and molecular targeted agents, this is not always possible on small tissue samples. These may only contain
undifferentiated cells that can express lineage markers associated with the differentiation type and using immunocytochemistry (ICC) the percentage of unclassifiable NSCLC can be limited to a very minority of cases.52 In the presence of clinical characteristics, such as young age and lack of smoking history, the CAP/IASLC/AMP international guidelines for lung cancer biomarker testing recommend to perform EGFR testing even in cytological cases showing squamous or small cell morphology, since an adenocarcinoma component cannot be completely excluded.17 Consequently, in these patients material should be prioritized for molecular analysis, since immunostaining cannot rule out EGFR testing. However, whenever possible both ICC and molecular studies need to performed, in order to better stratify patients for treatment. To this end, sufficient multidisciplinary team communication and ROSE by an experienced cytopathologist to consistently and correctly decide on handling of the specimens are possibilities to maximize available material.52
EGFR Mutation-Specific Monoclonal Antibodies Several studies suggested the usefulness of ICC assays using commercially available mutation-specific rabbit monoclonal antibodies directed against the most common mutated forms of EGFR: the 15-bp/5-amino-acid deletion (E746_A750del) in exon 19 (clone 6B6) and the L858R point mutation in exon 21. The high specificity (96– 100%) support their use as an initial screen; however, their lower sensitivity (47–92%) suggests the requirement of standard DNA-based assay to clarify EGFR status in specimens with equivocal or negative immunostaining results with these two mutation-specific monoclonal antibodies.53–60 Moreover, the 6B6 EGFR mutant antibody strongly reacts to EGFR proteins carrying the E746_A750del, whereas the immunoreaction is variable or absent in tumors affected by the less common 9-, 12-, 18- 24-, and 27-bp deletions.54 In paucicellular cytological specimens deemed inadequate for analysis, mutation-specific antibodies61 may be the most cost-effective option available to enable the specific detection of a few mutated cells in background rich of non-neoplastic cells.61 Before clinical application, however, the use these antibodies should be validated against a valid molecular assay. The impact of routine cytopreparation on the performance of EGFR mutationspecific antibodies should also be assessed; only one recent study human lung cancer cell culture as immunocytochemistry controls showed that formalin-based fixative ensure better results as compared with ethanol-based fixatives.45 However, the effect of different staining (DiffQuik or Papanicolaou) and preparation (smear or cell blocks) methods on the performance of EGFR mutationspecific antibodies remains to be determined. To date, most studies were carried out on cell blocks, since they Diagnostic Cytopathology, Vol. 00, No 00
5
Diagnostic Cytopathology DOI 10.1002/dc
BELLEVICINE ET AL.
represent the primary means by which additional cellular material is harvested for ancillary immunocytochemical studies.62–64 However, Hasanovic et al showed that monolayer preparations are more suitable than cell blocks62 and Kawara et al reported satisfactorily results on ethanol fixed Papanicolaou stained smears.56 Moreover, it is still unknown the reliability, in terms of evidence of clinical benefit from targeted therapy, of mutational data exclusively obtained on cytological specimens by EGFR mutation-specific antibodies.17
Testing Patients with Acquired Resistance Most (70%) of the patients harboring sensitizing EGFR mutations have significant clinical and radiographic responses to TKIs.65 However, almost all patients relapse or progress, a phenomenon termed acquired resistance. The most common mechanism is the emergence of a subclone of resistant cells carrying a secondary EGFR TK domain mutation. The most frequent (50%) of these resistance mutation is the T790M, that occurs in cis on the same allele as the original activating mutations and increases the affinity of the receptor for ATP, thereby reducing the TKI activity. While sensitizing EGFR mutations are present in every tumor cell, the clone carrying the T790M mutation is being selected under the TKI administration pressure and progressively expands.66 Second-generation EGFR TKIs are being intensively evaluated as further lines of treatment when AR resistance onsets. Afatinib (BIBW2992; Boehringer Ingelheim, Ingelheim, Germany) is a secondgeneration TKI, that irreversibly binds with a higher affinity to EGFR, even when the T790M occurs.67 To reliably detect the T790M the CAP/IASLC/AMP guidelines suggest to implement mutational assays sensitive enough to detect 5% of mutant cells.17 The slightly invasive type of FNA biopsies will be exploited in the future for serial tumor sampling, over time, to timely identify molecular markers of resistance.7
EGFR Testing by Next Generation Sequencing Next generation sequencing (NGS), also known as massive parallel sequencing, allows the analysis of a very large number of DNA molecules, that may represent a large number of genes or the in-depth parallel sequencing of a single gene. This latter application was exploited to increase the sensitivity of EGFR mutation detection on cytology by two different studies;68,69 these employed the 454 GS-Junior Next Generation sequencer (Roche Diagnostics, Mannheim, Germany), whose technology is a combination of emulsion PCR and pyrosequencing. In the largest of these studies performed on small tissue routine samples, including 66 cytology specimens, de Biase et al. demonstrated that 454 GS increases by 10% the number of cases harboring clinically relevant mutations detected by direct sequencing.68 The increase was mainly observed 6
Diagnostic Cytopathology, Vol. 00, No 00
in low cellular specimens (
EGFR Analysis: Current Evidence and Future Directions Claudio Bellevicine, M.D., Umberto Malapelle, Ph.D., Caterina de Luca, B.S., Antonino Iaccarino, Ph.D., and Giancarlo Troncone, M.D.,
Until a few years ago, only lung cancer histological specimens were considered suitable for testing epidermal growth factor receptor (EGFR) mutations. Then, several retrospective studies were designed to test EGFR mutation on a sizeable number of parallel cytological and histological samples obtained from the same patients and, even more recently, several institutions reported their prospective clinical experiences on routine specimens. Basing on these studies the College of American Pathologists, the International Association for the Study of Lung Cancer, and the Association for Molecular Pathology have recently considered cytological samples suitable for EGFR testing. Therefore, it seems timely to draw together the threads of this large body of information in order that cytopathologists can be knowledgeable partners in the multidisciplinary process of targeted cancer therapy and to help refine current testing guidelines. This review addresses (1) the more common proposed techniques including the use of direct cytologic smears cell blocks and liquid based cytology; (2) the issues related to current practice, which in Europe is external centralized testing that is usually done on samples containing very few cells; and (3) the future directions based on the implementation on lung cytology of next generation sequencing approaches. Diagn. Cytopathol. 2014;00:000–000. C 2014 Wiley Periodicals, Inc. V
Key Words: lung cancer; cytology; aspiration; EGFR; next generation sequencing
The Epidermal Growth Factor Receptor (EGFR) is a major target in the treatment of advanced-stage non-small cell lung cancer (NSCLC). The first generation of EGFR tyroDepartment of Public Health, Anatomic Pathology Section, University of Naples Federico II, Naples, Italy *Correspondence to Giancarlo Troncone, MD, PhD, Department of Public Health, Universita di Napoli Federico II, via Sergio Pansini 5, I-80131 Napoli, Italy. E-mail: [email protected] Received 23 April 2013; Revised 10 January 2014; Accepted 24 February 2014 DOI: 10.1002/dc.23142 Published online 00 Month 2014 in Wiley Online Library (wileyonlinelibrary.com). C 2014 WILEY PERIODICALS, INC. V
Ph.D.*
sine kinase inhibitors (TKI) included gefitinib (Iressa; AstraZeneca, London, UK) and erlotinib (Tarceva; Genentech, South San Francisco, CA, and OSI Pharmaceuticals, Long Island, NY).1,2 The efficacy of this treatment, as monotherapy in the first line setting, is restricted to those NSCLC harboring EGFR mutations.1 In particular, the two most common mutations in EGFR, the short in-frame deletions in exon 19 and the L858R point mutation in exon 21, are strongly correlated with improved quality of life, a high response rate (RR) and a longer progression-free survival (PFS). EGFR mutations are more commonly associated to female gender, adenocarcinoma (ADC) morphology, and Asian ethnicity in the absence of a smoking history.2,3 The first indications for anti-EGFR therapies are the locally advanced, recurrent, or metastatic disease settings; since these patients are not surgical candidates, tissue sampling for molecular testing through a minimally invasive approach, such as cytology, is ideal. As recently as 2009, however, a consensus statement from the European EGFR Workshop Group recommended that clinicians should be encouraged to provide tissue biopsies rather than cytological samples whenever possible, as further research was still necessary to fully understand the clinical reliability of mutational data obtained from cytological specimens.4 At that time, with a few rare notable exceptions,5,6 most of the reported institutional experiences with EGFR testing included only a few cytological samples. The small amount of tissue of most specimens, the need of optimizing cyto-preparation, the interference of background non-neoplastic cells, the necessity of sensitive molecular techniques and the lack of in vitro diagnostic tests licensed for cytology generated the diffuse perception that cytological material was unsuitable for effective molecular analysis and only sufficient to provide a generic diagnosis of NSCLC.7 Diagnostic Cytopathology, Vol. 00, No 00
1
Diagnostic Cytopathology DOI 10.1002/dc
BELLEVICINE ET AL.
this evidence, the international guidelines for lung cancer biomarker testing issued in 2013 by the College of American Pathologists (CAP), the International Association for the Study of Lung Cancer (IASLC), and the Association for Molecular Pathology (AMP) considered cytological samples suitable for EGFR testing.17 Hence, it seems the time is now ripe to review the more common techniques involving the use of different cytological preparations, the issues related to external centralized testing and the next generation sequencing approaches that hold promise for even further refinement of personalized treatment regimens for patients with lung cancer.
To eliminate such a perception and to assess the clinical reliability of mutational data obtained from cytological specimens and in particular, whether the mutation status assessed on cytological samples represented the mutation status of the entire tumor, retrospective studies were designed to test EGFR mutation on a sizeable number of parallel cytological and histological samples obtained from the same patients. Table I reports a list of these studies.8–15 The degree of concordance between paired samples was very high but not absolute, ranging between 82% and 100%. Discrepancies were more likely to reflect technical issues related to the different pre-analytical preparation rather than genuine intratumoral EGFR mutational heterogeneity. A detailed study by Yatabe et al., who tested 50 tumors in three different areas and 5 into 100 points, demonstrated that intratumoral EGFR mutational heterogeneity is extremely rare in lung adenocarcinomas.16 Accordingly, EGFR mutation patterns between primary and metastatic sites and between primary and recurrent neoplasms are similar;17 hence, testing of multiple tissue areas within a single tumor is not necessary.17 It thus conceivable that mutations detected on a cytological sample are representative of the entire tumor and, therefore, valuable to predict the response to treatment. Quite recently, several institutions reported their prospective clinical experiences on routinely collected cytological specimens.8,18–23 In general, the methodologies were sufficiently robust to test satisfactorily most of the routine cases; even limited cytological materials containing a small amount of extractable DNA were reliably tested.18 Different techniques, including direct sequencing and real time PCR (RT-PCR) based methods, applied to different sample types, including exfoliative and aspirative samples, and preparations, such as direct smears, cell-blocks (CB) and monolayer preparations, yielded data comparable to those obtained on larger tumor biopsies in terms of frequency and distribution of EGFR mutations according to sex, histology, and prior smoking history8,18–23 Basing on
EGFR Testing on Smears The smears used for diagnosis may be the only source of tumor cells testable for molecular studies, in cases with a small quantity of tissue material or when a cell-block is not available. Whenever a single slide is available, the molecular testing will destroy the evidence of tumor cells, which might have medico-legal consequences. To manage this risk, smears can be photographed or digitally scanned, to record the cytomorphology of a representative microscopic field for the archives.24,25 The DNA obtained from direct smears is of better quality than that extracted from matched cell blocks.26 Killian et al. reported that Diff-Quik should be preferred to Papanicolaou-stained slides.24 More recently, however, in cell line experiments, Papanicolaou-stained smears yielded optimal DNA yield and fragment length.27 As a matter of the fact, the studies performing EGFR testing on routine Diff-Quik9,23,28 or Papanicolaou8 smears did not show significant differences. The process of removing the coverslip of archival smears do not compromise the quality of the DNA isolated for molecular studies, but it is time-consuming. To avoid any delay, rapid on site evaluation (ROSE), at the time of the FNA procedure, can select a representative slide, that it is maintained uncoverslipped for immediate DNA extraction and EGFR
Table I. Summary of Studies That Analyzed EGFR Mutational Status in Paired Histological and Cytological Samples References
Type of cytological sample 8
Paired samples (n)
EGFR detection methods
Concordant/total (%) 100 100 100 100
Lozano et al. Bozzetti et al.9 Bruno et al.10 Chowdhury et al.11
Direct smears Direct smears or needle washing Direct smears Cell Block
20 9 50 12
Goto et al.12
Cell pellet from Bronchial Brushing
30
Mitiushkina et al.13
Direct smears
72
Sun et al.14 Kode et al.15
Direct smears and/or cell blocks Direct smears from the same site of the histology Direct smears from a different site
60 37
Direct sequencing Direct sequencing Direct sequencing LCM and direct sequencing or pyrosequencing PNA-LNA PCR clamp or ARMSSkorpion Exon 19 Fragment length assay; Exon 21 Allele-specific real-time Pyrosequencing Pyrosequencing
23
Pyrosequencing
Kode et al.15 a
29/30 case were evaluated; one test failure in histology sample. Discordant cases included two sputum specimens and three pleural effusions
b
2
Diagnostic Cytopathology, Vol. 00, No 00
96.6a 68/72; 94.4 55/60; 91.7b 36/37; 97 19/23; 82
Diagnostic Cytopathology DOI 10.1002/dc
EGFR TESTING ON CYTOLOGY
testing.26 Interestingly, da Cunha Santos et al. applied the “freezer method” to remove coverslips, reducing the time required for getting the slides ready for DNA extraction.29 The neoplastic can be scraped off with a tip of an insulin syringe needle, while examining the smear at the microscope.5,30,31 In their practice, Billah et al. carefully mapped the slides to define the location of tumor fragments or areas of higher tumor density, which enabled testing 93 of 99 archival smears.20 As a general rule, slides prepared from fine needle aspirates are easier to be microdissected than exfoliative cytology smears.32 In these latter, benign and neoplastic cells are more likely to co-exist, making tumor cells selection difficult. Manual microdissection, however, has intrinsic limitations. Lozano et al.8 only analyzed smears with >50% tumor cellularity, while Pang et al.22 were not able to manual microdissect 12 of 147 (8.1%) smears. In attempt to make manual microdissection more feasible, the novel cell-transfer technique was proposed.33 This technique employs the mounting media to detach the neoplastic area marked on the smears excluding non-neoplastic contaminant. Alternatively, laser capture microdissection (LCM) enables selective cancer cell dissection. In ours34 and other’s hands6,35 only 25 cancer cells are needed to reliably detect EGFR mutations. However, LCM requires specific training and it is very time consuming is an expensive technology that is not available to all laboratories. In a near future, computer-aided LCM will semi automate the microdissection process, allowing a more rapid retrieval of tumor cells, saving pathologist’s time and effort.36 Although morphology cannot reliable predict TKI responsiveness, Brechtel et al reported that on smears the cytomorphological features associated to former bronchioloalveolar carcinoma (BAC) such as flat monolayers, intranuclear cytoplasmatic inclusions, the absence of large nucleoli and extracellular mucin are linked to the presence of mutated EGFR cells.37 More recently, Marotti et al. confirmed these observations, also showing a link
between poor tumor differentiation, necrosis and squamoid features and the occurrence of KRAS mutations.38
EGFR Testing on Cell Blocks Cell block preparations allows for the obtainment of multiple serial sections for ancillary studies and a banked archive for future retrospective testing as new agents become available;17 furthermore, cell blocks are irreplaceable when the aspirated material, after the prolonged echo-endoscopic bronchial (EBUS) or digestive (EUS) procedures is clotted.39,40 Cell blocks may also be useful to provide useful morphological clues; for example, intercellular bridges can be highlighted on cell blocks, thus allowing for classification of otherwise non-small cell carcinomas as squamous cell carcinomas.41 Remarkably, cell blocks are recommended over smear preparations, by the recent CAP/IASLC/AMP guidelines.17 This may probably reflect, in part, the fact that most molecular pathologists may prefer dealing with preparations similar to those of traditional histological blocks. The disadvantages of cell blocks are also well known (Table II). The adequacy for tumor cells cannot be assessed immediately and deeper sections from the cell block may not contain the tumor cells observed on the originally prepared hematoxylin & eosin-stained slide.26 Cutting the cell block only once yielding sections both for microscopy and for genomic studies would be ideal to save as much tumor tissue as possible for molecular testing.42 Cell block cellularity may be variable, being dependent on the number of dedicated fine needle aspiration (FNA) passes and on the post-procedural handling of the needle rinse specimen.43 In routine practice, not infrequently cell blocks are either acellular or sparsely cellular.26 In the series reported by Pang et al. 8/51 (15%) cell blocks prepared from FNAs were rejected,22 while in the study by Cai et al the rejection rate was lower (9%).21 Since cell blocks are prepared from a pooled specimen, the neoplastic cells obtained by high yield needle passes are diluted by background benign elements from off target passes; as a consequence, tumor cell enrichment is cumbersome and LCM often necessary.11
Table II. Advantages and Disadvantages of Different Cytological Preparations for EGFR Mutation Testing Direct Smears
Cell Blocks
Liquid-Based Cytology
Advantages
High quality DNA; visualization of malignant cells; ROSE feasible
Diagnostic smears preserved TTF1 positive cells can be selected; recommended by CAP/IASLC/AMP Guidelines; useful for EBUS and EUS
Disadvantages
Loss of diagnostic material; delay due to coverslip removal; LCM may be needed
Poor DNA quality; ROSE unfeasible
High quality DNA; eliminates the need for slide preparation by clinicians; material maximized; sensitive methods can directly be applied to CytoLyt derived DNA ROSE unfeasible
Diagnostic Cytopathology, Vol. 00, No 00
3
Diagnostic Cytopathology DOI 10.1002/dc
BELLEVICINE ET AL.
EGFR Testing on Liquid Based Cytology Liquid-based cytology (LBC) makes material collection easier and eliminates the need for slide preparation by clinicians, which is valuable when it is not possible to have on-site an experienced cytopathologist.44 Additional advantages are reduction of background material such as blood and debris and the possibility of dividing the sample equally on several generally identical slides. LBC preparations are derived from homogenized, pooled liquid-based samples and thus, alike cell blocks, they are unsuitable for on-site evaluation and they may have variable neoplastic cell content.26 The most widespread used LBC fixatives are the methanol-based CytoLyt solution (Cytyc Corporation, Marlborough, MA) and the CytoRich Red collection fluid (Fisher Scientific UK Ltd, Loughborough, UK), that contains a small amount (0.4%) of formalin. In a recent study, we managed to extract DNA that was suitable for EGFR mutation analysis in CytoLyt-fixed clinical cytological samples from patients with non-small cell lung cancer.34 More recently two different studies both using human lung carcinoma cell lines carefully evaluated the effect of LBC fixatives on DNA quality. CytoLyt solution provides a higher DNA yield than CytoRich Red collection fluid,27 while both fixatives well preserved the DNA.27,45 The pre-analytical management of LBC sample depends on the sensitivity of the mutational assay. PCR and direct sequencing can reliably detect EGFR mutations only when slides are prepared from the collection fluid and when a pure population of neoplastic cell is selected by LCM.46 Conversely, LCM and direct sequencing may be replaced by more sensitive non-sequencing RT-PCR based assays directly performed on CytoLyt derived DNA.34 Malignant pleural effusion (MPE) is a frequent complication of peripheral lung adenocarcinoma.47 Testing EGFR on MPE obviates the need of tissue sampling. From a biological point of view, anchorage-independent and movable tumor cells from MPE are more representative of the invading tumor than a pleural nodule biopsy.48 However, technical issues are relevant. Kimura et al., extracted the DNA from 1.0 ml of pleural effusion fluid; mutations were detected even when the DNA yields were below the concentration measurable by spectrophotometry. It is conceivable that mutations can be detected in soluble DNA, due to the release of tumor-derived DNA into the pleural fluid through the apoptosis or necrosis of disseminated tumor cells.49,50 However, since the composition of MPE includes normal, inflammatory and reactive mesothelial cells, direct sequencing has an insufficient sensitivity. Strategies to avoid false-negative results include either the visualization of neoplastic cells on smears or cell blocks before DNA extraction or more sen4
Diagnostic Cytopathology, Vol. 00, No 00
sitive molecular methods. Noteworthy, the EGFR mutation status of MPE does not always re-capitulate that of the primary tumor tissue. Han et al. showed that 3 of 21 patients (14.3%) had a discordant MPE EGFR mutation status, probably reflecting technical issues.48
EGFR Testing in Centralized Laboratories The cytopathologist plays different roles in lung cancer biomarker testing. He is responsible of the preanalytical phases of testing; he has to review representative cell block sections or select best quality smears, to determine the cellularity and purity (mucin and necrosis may inhibit amplification) of the tumor sample being submitted for biomarker testing.43 Since various mutational assays have different analytic sensitivities, the cytopathologist (or the technician) should enrich for tumor content to a level that is acceptable for the assay being used.43 The cytopathologist may perform the molecular testing in-house, which is a frequent procedure in the United States,51 supervising the activity of his/her Clinical Laboratory Improvement Amendment (CLIA) accredited intramural molecular pathology laboratory, Conversely, in Europe not all pathology departments are equipped to run EGFR testing and a greater number of cytopathologists refers to outside molecular pathology laboratories.4 When determining what laboratory to use for outside EGFR testing, the cytopathologist should bear in mind the requirements that the CAP/IASLC/AMP guidelines set for testing laboratories.17 These have to ensure the use of validated methodologies able to detect, even in specimens with as little as 10% cancer cells, not only the two major mutations but all those occurring with a frequency of at least 1% of EGFR-mutated lung adenocarcinomas within a time around testing of 10 working days. Once the results of molecular analysis are received, the cytopathologist may compare the mutation signals with the extent of tumor cells in the tested specimen, carefully evaluating the quality processes employed to ensure confidence in the results.43 The minimal number of annual EGFR tests sufficient for guaranteeing the technical sufficiency of a laboratory was arbitrarily set at 50.52 The EGFR European Workshop Group recommend reference laboratories to be accredited either at the national or at the European level.4 In Italy, Greece, Holland, and Germany the national society of pathology organizes ring trials for external quality assurance assessment once or twice a year.17 The schemes organized by the European Society of Pathology evaluate not only EGFR but also KRAS and ALK biomarker testing in NSCLC. Similarly, a proficiency testing for EGFR and ALK testing is offered by the CAP. However, to date, only histological samples have been included in the proficiency testing schemes and assessment of the quality of testing on cytology remains problematic.
Diagnostic Cytopathology DOI 10.1002/dc
EGFR TESTING ON CYTOLOGY
External centralized testing is usually done on samples containing very few cells.19,20,22 In a reference molecular laboratory, the presence of both a pathologist, who is highly trained in selecting the best cytological specimen while considering the specific methods used by his laboratory, and a dedicated technician, who allows expeditious processing of cytological samples for molecular testing, is crucial, but not sufficient. In our experience, the smear sent to our centralized laboratory is not always the “good one” and is often paucicellular, since the cytopathologist who diagnosed the case is often reluctant “to sacrifice” the morphology of malignant cells for DNA extraction.25 In our laboratory practice, 16.2% (59 of 364) of the cytological samples were rejected because contained less than 25 neoplastic cells.25 In the study by Billah et al. in-house collected cytological material were more often (96, 3%; 160 of 166) adequate than outside cases from referred patients (83, 7%; 36 of 43).20 Pang et al. reported that 10% of outside samples were inadequate22 and despite the use of a very sensitive methodology (TheraScreen EGFR29 kit), 14% of samples received by Allegrini et al. were deemed inadequate.19 Thus, externalizing EGFR testing may be associated with a higher rate of inadequate samples. One of the basic requirements to improve specimen adequacy from participating sites is educational and efforts should be spent to provide the essential concepts of lung molecular pathology, whose understanding is necessary to properly handle and manage cytopathology specimens for molecular testing. To investigate the degree of reliability of treatment prediction based on EGFR mutational diagnosis performed by a centralized laboratory on cytological material we reviewed our records of the clinical outcomes of patients affected by lung cancer and compared the benefit of treatment in two groups of patients, the only between-group difference being the type of sample (histology or cytology) tested.25 In patients tested on histology the RR was 54% and the PFS 9.2 months. Patients selected on cytological samples had a similar outcome (RR 5 62%, PFS 5 8.6 months). Consequently, despite concern that less-than-optimal samples are sent to reference molecular laboratories, cytological samples can reliably predict treatment outcome. Thus, not only EGFR mutations can be reliably detected on paucicellular lung cancer cytology specimens,6,11,14,19,34,35 but also the results can reliably predict the efficiency of gefitinib treatment.25
Immunocytochemistry and EGFR Testing: Sample Prioritization Although knowledge of the specific NSCLC subtype would be essential prior to prescribing a number of important new drugs to treat NSCLC, including both cytotoxic chemotherapeutic and molecular targeted agents, this is not always possible on small tissue samples. These may only contain
undifferentiated cells that can express lineage markers associated with the differentiation type and using immunocytochemistry (ICC) the percentage of unclassifiable NSCLC can be limited to a very minority of cases.52 In the presence of clinical characteristics, such as young age and lack of smoking history, the CAP/IASLC/AMP international guidelines for lung cancer biomarker testing recommend to perform EGFR testing even in cytological cases showing squamous or small cell morphology, since an adenocarcinoma component cannot be completely excluded.17 Consequently, in these patients material should be prioritized for molecular analysis, since immunostaining cannot rule out EGFR testing. However, whenever possible both ICC and molecular studies need to performed, in order to better stratify patients for treatment. To this end, sufficient multidisciplinary team communication and ROSE by an experienced cytopathologist to consistently and correctly decide on handling of the specimens are possibilities to maximize available material.52
EGFR Mutation-Specific Monoclonal Antibodies Several studies suggested the usefulness of ICC assays using commercially available mutation-specific rabbit monoclonal antibodies directed against the most common mutated forms of EGFR: the 15-bp/5-amino-acid deletion (E746_A750del) in exon 19 (clone 6B6) and the L858R point mutation in exon 21. The high specificity (96– 100%) support their use as an initial screen; however, their lower sensitivity (47–92%) suggests the requirement of standard DNA-based assay to clarify EGFR status in specimens with equivocal or negative immunostaining results with these two mutation-specific monoclonal antibodies.53–60 Moreover, the 6B6 EGFR mutant antibody strongly reacts to EGFR proteins carrying the E746_A750del, whereas the immunoreaction is variable or absent in tumors affected by the less common 9-, 12-, 18- 24-, and 27-bp deletions.54 In paucicellular cytological specimens deemed inadequate for analysis, mutation-specific antibodies61 may be the most cost-effective option available to enable the specific detection of a few mutated cells in background rich of non-neoplastic cells.61 Before clinical application, however, the use these antibodies should be validated against a valid molecular assay. The impact of routine cytopreparation on the performance of EGFR mutationspecific antibodies should also be assessed; only one recent study human lung cancer cell culture as immunocytochemistry controls showed that formalin-based fixative ensure better results as compared with ethanol-based fixatives.45 However, the effect of different staining (DiffQuik or Papanicolaou) and preparation (smear or cell blocks) methods on the performance of EGFR mutationspecific antibodies remains to be determined. To date, most studies were carried out on cell blocks, since they Diagnostic Cytopathology, Vol. 00, No 00
5
Diagnostic Cytopathology DOI 10.1002/dc
BELLEVICINE ET AL.
represent the primary means by which additional cellular material is harvested for ancillary immunocytochemical studies.62–64 However, Hasanovic et al showed that monolayer preparations are more suitable than cell blocks62 and Kawara et al reported satisfactorily results on ethanol fixed Papanicolaou stained smears.56 Moreover, it is still unknown the reliability, in terms of evidence of clinical benefit from targeted therapy, of mutational data exclusively obtained on cytological specimens by EGFR mutation-specific antibodies.17
Testing Patients with Acquired Resistance Most (70%) of the patients harboring sensitizing EGFR mutations have significant clinical and radiographic responses to TKIs.65 However, almost all patients relapse or progress, a phenomenon termed acquired resistance. The most common mechanism is the emergence of a subclone of resistant cells carrying a secondary EGFR TK domain mutation. The most frequent (50%) of these resistance mutation is the T790M, that occurs in cis on the same allele as the original activating mutations and increases the affinity of the receptor for ATP, thereby reducing the TKI activity. While sensitizing EGFR mutations are present in every tumor cell, the clone carrying the T790M mutation is being selected under the TKI administration pressure and progressively expands.66 Second-generation EGFR TKIs are being intensively evaluated as further lines of treatment when AR resistance onsets. Afatinib (BIBW2992; Boehringer Ingelheim, Ingelheim, Germany) is a secondgeneration TKI, that irreversibly binds with a higher affinity to EGFR, even when the T790M occurs.67 To reliably detect the T790M the CAP/IASLC/AMP guidelines suggest to implement mutational assays sensitive enough to detect 5% of mutant cells.17 The slightly invasive type of FNA biopsies will be exploited in the future for serial tumor sampling, over time, to timely identify molecular markers of resistance.7
EGFR Testing by Next Generation Sequencing Next generation sequencing (NGS), also known as massive parallel sequencing, allows the analysis of a very large number of DNA molecules, that may represent a large number of genes or the in-depth parallel sequencing of a single gene. This latter application was exploited to increase the sensitivity of EGFR mutation detection on cytology by two different studies;68,69 these employed the 454 GS-Junior Next Generation sequencer (Roche Diagnostics, Mannheim, Germany), whose technology is a combination of emulsion PCR and pyrosequencing. In the largest of these studies performed on small tissue routine samples, including 66 cytology specimens, de Biase et al. demonstrated that 454 GS increases by 10% the number of cases harboring clinically relevant mutations detected by direct sequencing.68 The increase was mainly observed 6
Diagnostic Cytopathology, Vol. 00, No 00
in low cellular specimens (
