OUP user menu

The Application of Molecular Diagnostic Studies Interrogating EGFR and KRAS Mutations to Stained Cytologic Smears of Lung Carcinoma

Bryan L. Betz PhD, Michael H. Roh MD, PhD, Helmut C. Weigelin MLS(ASCP), Jeremiah B. Placido MD, Lindsay A. Schmidt MD, Sara Farmen MD, PhD, Doug A. Arenberg MD, Gregory P. Kalemkerian MD, Stewart M. Knoepp MD, PhD
DOI: http://dx.doi.org/10.1309/AJCP84TUTQOSUONG 564-571 First published online: 1 October 2011


EGFR and KRAS mutation analyses are of increasing importance for guiding the treatment of non–small cell lung carcinomas. Insufficient cellularity of cell blocks can represent an impediment to the performance of these tests. We investigated the usefulness of cytologic direct smears as an alternative specimen source for mutation testing. Tumor cell–enriched areas from freshly prepared and archived rapid Romanowsky–stained direct smears in 33 cases of lung carcinoma were microdissected for DNA isolation and evaluated for EGFR and KRAS mutations. EGFR mutations were detected in 3 adenocarcinomas; 2 tumors had the L858R substitution and 1 an exon 19 deletion. KRAS mutations affecting codon 12, 13, or 61 were detected in 11 cases (8 adenocarcinomas and 3 non–small cell carcinomas). EGFR and KRAS mutations were mutually exclusive. Hence, archived and freshly prepared direct smears represent a robust and valuable specimen source for molecular studies, especially when cell blocks exhibit insufficient cellularity.

Key Words:
  • Lung cancer
  • EGFR
  • KRAS
  • Cytology
  • Direct smear
  • Fine-needle aspiration
  • Non–small cell carcinoma
  • Adenocarcinoma

Lung cancer is one of the most commonly diagnosed malignancies and represents the highest cause of cancer mortality in the world.1,2 Lung cancer is a histologically diverse disease dichotomized into 2 general categories: small cell carcinoma and non–small cell lung carcinoma (NSCLC). NSCLCs can be further subclassified into major subtypes, including adenocarcinoma, squamous cell carcinoma, large cell carcinoma, and large cell neuroendocrine carcinoma.2

Adenocarcinoma represents the most common subtype of lung cancer.1 During the past decade, our understanding of the molecular pathogenesis of pulmonary adenocarcinoma has been improved and refined. This cancer is no longer regarded simply as a histologically diverse disease composed of distinct variants—such as bronchioloalveolar, papillary, and acinar—as recent studies have provided increased insight into the various molecular aberrations that underlie the pathogenesis of adenocarcinomas in different demographic groups. Two salient examples include mutations in the epidermal growth factor receptor (EGFR) gene, seen more commonly in East Asian women with a negative smoking history, and mutations in KRAS, observed more commonly in Caucasian male smokers.3

Approximately 10% to 15% of unselected NSCLCs have EGFR mutations, with the highest frequency occurring in adenocarcinoma.4,5 The most common EGFR mutations are small in-frame deletions in exon 19 and a point mutation in exon 21 resulting in an L858R substitution.6 These collectively account for approximately 90% of all EGFR mutations and phenotypically result in hyperactivation of the EGFR. Lung adenocarcinomas driven by these mutations are sensitive to EGFR tyrosine kinase inhibitors (TKIs) such as gefitinib (Iressa) and erlotinib (Tarceva).79 Furthermore, when treated with these TKIs, patients who have tumors with EGFR mutations have longer progression-free survival than do patients with tumors without mutations.10,11

Oncogenic KRAS mutations are found in approximately 15% to 30% of NSCLCs.3,5 These point mutations most frequently involve codon 12 or 13; rarely, mutations occur at codon 61.12 They result in the impairment of the intrinsic GTPase activity of KRAS and are insensitive to GTPase-activating proteins, thereby leading to the accumulation of activated, GTP-bound KRAS protein. The occurrence of KRAS and EGFR mutations is mutually exclusive.13,14 Furthermore, as KRAS represents a downstream effector of EGFR, lung adenocarcinomas with mutations in KRAS are refractory to EGFR TKIs.15,16

The importance of identifying EGFR and KRAS mutations in lung cancer highlights an increasing emphasis on personalized medical care as molecular diagnostic testing to interrogate these mutations enables physicians to optimize therapeutic regimens. This is especially important for patients with lung cancer who have advanced-stage cancer at diagnosis and, as a result, are not candidates for surgical resection. In these patients, small biopsy specimens, often obtained by fine-needle aspiration (FNA) for cytologic examination, represent the only opportunity to obtain tissue material for diagnosis and ancillary molecular diagnostic studies. Consequently, there is increasing demand on cytopathologists to maximize the efficacy with which this limited tissue material is handled and analyzed.

Cell blocks prepared from lung and mediastinal lymph node fine-needle aspirates are routinely used for ancillary studies. Unfortunately, in some cases, insufficient cellularity of cell blocks represents an impediment to the performance of these studies, which can necessitate a repeated FNA during which the risks and complications are not negligible. There are limited reports in the literature demonstrating the usefulness of cytologic smears of lung cancer in the interrogation of EGFR mutations.1719 These studies primarily used decoverslipped, archived smears as a source of DNA that was subjected to EGFR mutation testing via high-resolution melting analysis. We extended these findings by performing EGFR and KRAS mutation testing on fresh and archived direct smears via polymerase chain reaction (PCR)-based fragment analysis and direct sequencing–based tests. To validate this approach, we tested corresponding resection tissue or cytologic blocks in addition to the direct smears to correlate results.

Materials and Methods

Case Selection

The study was approved by the institutional review board at the University of Michigan, Ann Arbor. In 7 cases, air-dried, cytologic direct smears were prepared fresh from scraped tumor tissue from surgically resected lung masses and stained with rapid Romanowsky, without coverslipping. In 5 cases, air-dried, rapid Romanowsky–stained smears were prepared fresh from cytologic specimens, without coverslipping. For 21 cases, rapid Romanowsky–stained air-dried smears were obtained from the archive (1 smear per case). Coverslipped slides were incubated in xylene for 3 days, after which the coverslips were gently removed. Decoverslipped slides were then allowed to dry and were submitted for DNA isolation after the appropriate area containing tumor was selected (see “DNA Isolation”). The corresponding formalin-fixed, paraffin-embedded tissue block (7 cases) or cytology cell block (26 cases) was also retrieved in each case for comparison testing. The cellularity of the cell blocks was assessed by examining the H&E-stained sections and estimating the total number of tumor cells and their percentage of total cellularity (ie, percentage of tumor cells out of total cells including benign nucleated cells). The total number of tumor cells were semiquantitatively designated as follows: acellular, no tumor cells (score 0); sparsely cellular, fewer than 50 tumor cells (1+); moderately cellular, approximately 50 to 300 tumor cells (2+); and abundantly cellular, more than 300 tumor cells (3+).

DNA Isolation

All stained smears were reviewed by 2 cytopathologists (M.H.R. and S.M.K.), and the area containing the highest proportion of tumor cells was marked on the underside of each slide using a marking pen Image 1. Genomic DNA was extracted from the marked region (area range, 16–80 mm2) using the Pinpoint Slide DNA Isolation Kit (Zymo Research, Irvine, CA) according to the manufacturer’s instructions and including the optional purification step. DNA was eluted in a final volume of 25 μL of TE buffer (10 mmol/L tris(hydroxymethyl)aminomethane-hydrochloride and 0.5 mmol/L EDTA; pH 9.0). Genomic DNA extraction from paraffin-embedded tissue and cytology cell blocks was performed on the BioRobot EZ1 (Qiagen, Valencia, CA) using the paraffin section protocol. For each block, 3 to 5 sections of 10-μm thickness were used for extraction. DNA was eluted in a final volume of 100 μL of TE buffer.

Mutation Testing

KRAS codons 12, 13, and 61 were evaluated for mutations using direct sequencing. Briefly, 2 fragments containing these codons were amplified using the following primer pairs: 5′-GTGTGACATGTTCTAATATAGTCA-3′ (forward) and 5′-GAATGGTCCTGCACCAGTAA-3′ (reverse) for codons 12 and 13 and 5′-GCACTGTAATAATCCAGACTG-3′ (forward) and 5′-CAATTTAAACCCACCTATAATGGT-3′ (reverse) for codon 61. Each 25-μL PCR reaction contained 5 μL of purified DNA, 500-nmol/L concentrations of each primer, and 1× Phusion HF mastermix (Finnzymes, Thermo Scientific, Vantaa, Finland). Cycling conditions consisted of denaturation at 98°C for 30 seconds followed by 40 amplification cycles: 99°C for 5 seconds, 60°C for 20 seconds, and 72°C for 20 seconds. An aliquot of each PCR product was confirmed by gel electrophoresis. The remainder was purified by using the QIAquick PCR purification kit (Qiagen) and subjected to bidirectional sequencing with ABI BigDye v1.1 terminators (Applied Biosystems, Carlsbad, CA) and the following nested sequencing primers: 5′-ATGTTCTAATATAGTCACATTTTC-3′ (forward) and 5′-GTCCTGCACCAGTAATATGC-3′ (reverse) for codons 12 and 13 and 5′-ATCCAGACTGTGTTTCTCCC-3′ (forward) and 5′-CCACCTATAATGGTGAATATC-3′ (reverse) for codon 61. Sequence products were purified using the DyeEx Spin Kit (Qiagen) and analyzed on the ABI 3130xl (Applied Biosystems). The forward and reverse sequence chromatograms were reviewed for mutations with software-assisted analysis (Mutation Surveyor, SoftGenetics, State College, PA).

Image 1

Five representative decoverslipped rapid Romanowsky–stained slides after microdissection for DNA extraction. Areas on slides containing optimal tumor cellularity and density were selected and indicated on the underside of each slide with a marking pen. These areas are visualized as clear spaces on the slide because cellular contents in these areas have been removed for DNA isolation. Note the tumor percentage written in the upper right corner of each slide.

EGFR mutations (exon 19 deletions and the exon 21 L858R substitution) were evaluated by using a PCR-based fragment analysis assay. Mutations were independently confirmed by direct sequencing using the following PCR primer pairs: 5′-TGGCACCATCTCACAATTGC-3′ (forward) and 5′-GAAAAGGTGGGCCTGAGGTT-3′ (reverse) for exon 19; and 5′-AGAGCTTCTTCCCATGATGATC-3′ (forward) and 5′-CAGCCTGGTCCCTGGTGTC-3′ (reverse) for exon 21. PCR conditions were as described for KRAS, except 300-nmol/L concentrations of each primer were used and the annealing temperature was 65°C. Amplification products were sequenced with the following nested sequencing primers: 5′-CCATCTCACAATTGCCAGTT-3′ (forward) and 5′-TGGGCCTGAGGTTCAGAG-3′ (reverse) for exon 19; and 5′-TCTTCCCATGATGATCTGTCC-3′ (forward) and 5′-GGTCCCTGGTGTCAGGAAA-3′ (reverse) for exon 21.


To determine if air-dried, rapid Romanowsky–stained cytologic smears represent a viable specimen source for DNA purification and mutational analysis, direct smears were first prepared with scraped tumor cells from 7 lung adenocarcinomas that were surgically resected Table 1 and Table 2 (cases 1–7). The smears were stained in rapid Romanowsky, and genomic DNA was extracted from tumor-enriched areas for EGFR and KRAS mutation testing. The extracted area ranged from 16 to 80 mm2, and the DNA yield averaged 1.02 μg (range, 0.2–2.82 μg). All cases exhibited robust PCR amplification in both mutation assays indicating that an adequate yield of high-quality DNA was extracted. The EGFR L858R substitution mutation was detected in 2 adenocarcinomas Figure 1. A KRAS G12D mutation was detected in another case Figure 2.

Next, for 26 cases, rapid Romanowsky–stained smears that were prepared from cytologic specimens were analyzed for EGFR and KRAS mutations. In 5 cases, the stained direct smears were prepared fresh and not coverslipped (Tables 1 and 2, cases 8–12). In the remaining 21 cases, archived stained direct smears were decoverslipped in xylene (Tables 1 and 2, cases 13–33). Areas enriched with tumor cells were microdissected for DNA purification in each case. The cytologic diagnoses for the 26 cases included the following: adenocarcinoma, 14; NSCLC, not otherwise specified, 10; squamous cell carcinoma, 1; and small cell carcinoma, 1. The extracted area ranged from 24 to 64 mm2, and DNA yield averaged 0.54 μg (range, 0.08–5.37 μg). Mutation analysis was successful in all cases. An EGFR exon 19 deletion was detected in 1 adenocarcinoma. KRAS mutations were detected in 7 cases of adenocarcinoma and 3 cases of NSCLC, not otherwise specified.

View this table:
Table 1

To validate the mutation results obtained from the stained smears, corresponding paraffin-embedded tissue blocks (cases 1–7) or cytologic cell blocks (cases 8–33) were obtained. Adequate tumor for mutation testing was present in each of the 7 tissue blocks. However, only 17 of 26 cell blocks contained sufficient cellularity; 9 cell blocks were completely acellular and, therefore, were not evaluated (Table 2). Overall, 23 of 24 cases demonstrated concordant EGFR and KRAS mutation results between the stained direct smear and the corresponding tissue or cell block. The discordant case (Table 2, case 28) was one in which concurrent KRAS codon 12 and 13 mutations were detected in the direct smear but not the corresponding cell block. Follow-up review indicated that the cell block contained sparse clusters of adenocarcinoma with an overall tumor cellularity of less than 10%. Consequently, the lack of detectable KRAS mutations in the cell block is likely a false-negative result due to lack of sufficient tumor cellularity in this specimen.

Overall, EGFR mutations were detected in 3 (14%) of 21 adenocarcinomas and not in the remainder of the cases examined. Of the 3 patients, 2 were men with a positive smoking history; the third patient was a woman with no smoking history Table 3. KRAS mutations were observed in 8 (38%) of 21 adenocarcinomas and 3 (30%) of 10 NSCLCs, not otherwise specified. All 11 patients with tumors with KRAS mutations had a positive smoking history. Finally, EGFR and KRAS mutations were mutually exclusive.

View this table:
Table 2
Figure 1

Representative sequencing results of EGFR mutations detected in direct smears. Two cases (A, Case 6; B, Case 7) with the EGFR L858R mutation, confirmed by sequencing of exon 21, are shown.


The advent of targeted therapeutics is changing the approach to management of patients with pulmonary adenocarcinoma. Fundamental to this was the discovery that EGFR mutations predict response and survival benefit in patients treated with EGFR-targeted therapies.10,11,15,20 KRAS mutation status has also gained relevance since lung tumors with these mutations are resistant to EGFR-targeted therapy.15,21 As the number of patients who receive targeted therapies increases, so too does the need to interrogate the mutation status of the associated molecular markers on small biopsy specimens of primary and metastatic pulmonary tumors. The possible inadequacy of specimens to perform these studies is problematic; additional biopsy procedures are not without risks and lead to delays in treatment.

Figure 2

Representative sequencing results of KRAS mutations detected in direct smears. Four cases (A, Case 1; B, Case 20; C, Case 11; D, Case 33) with KRAS mutations (G12D, G13D, Q61H, and G12C) are shown.

View this table:
Table 3

To date, few studies have examined the use of immediately acquired or archived direct smears from FNA material for use in molecular analysis. The use of direct smears for molecular studies provides numerous possible advantages over the use of cell blocks, including the ability to ascertain at the time of the procedure whether sufficient material is available for subsequent analysis; less processing, resulting in decreased turnaround time and technical-related expenses; tumor cell enrichment enabling DNA isolation from cellular material obtained from a single pass as opposed to a cumulative mixture of cellular material from multiple passes; and obtaining high-quality non–cross-linked DNA owing to absence of exposure to formalin. The use of direct smears also provides advantages over some other suggested methods such as dividing a specimen into aliquots in a microcentrifuge tube22,23 or applying an aliquot to a filter paper card.24 In contrast with the latter 2 methods, the use of stained smears for molecular testing provides the opportunity for pathologist-based review of tumor cell adequacy before extraction.

This activity is advocated by many researchers and clinicians in the molecular pathology testing community. For example, in the most recent College of American Pathologists survey, 121 of 130 laboratories performing KRAS mutation testing require pathologist review of the specimen before testing.25 This is an especially important quality assurance step given the limited sensitivity of some molecular tests. It also provides the opportunity for tumor cell enrichment by microdissection in cases with limited tumor cellularity. Given these advantages and the chief drawback of cell blocks (ie, the lack of sufficient material in a significant number of cases), the specific goals in this study were as follows: (1) investigate the potential for developing a real-time method whereby direct smears obtained during FNAs are effectively triaged by pathologists for subsequent EGFR and KRAS mutation testing, (2) demonstrate and ensure that such material obtained during the procedure would prove to be adequate and reliable for subsequent molecular analysis, and (3) ascertain whether mutational analysis could be performed retrospectively using archival cases.

In our study, the use of cytologic direct smears for molecular testing proved to be a robust and reliable method. Mutation testing was successful in freshly obtained and archived rapid Romanowsky–stained smears and yielded results that were concordant with those obtained from corresponding tissue blocks and cell blocks. Of note is that we had 1 discordant case in which KRAS mutations were identified in the direct smear but not in the corresponding cell block owing to limited tumor cellularity. This case highlights a potential limitation of cell blocks that may contain tumor cell clusters that are not evenly dispersed throughout the block, thereby limiting the ability to enrich for tumors. The cell block in this case contained sparse tumor cell clusters with an overall tumor cell percentage of less than 10%. This proportion of tumor cells is below what is required for the direct sequencing assay we used for KRAS mutation testing, likely leading to the false-negative result. In contrast, the area selected and extracted on the direct smear contained 80% tumor cells, which allowed detection of the mutation in this case. Also of significance is that a large percentage of cell blocks in our study were acellular (9/26 [35%]) and, thus, proved inadequate for molecular testing. Direct smears were successfully tested in each of these cases, underscoring the usefulness of these preparations in cases in which cell blocks contain inadequate cellularity.

Because the procedure of removing coverslips is time-consuming, it is more convenient for pathologists to anticipate that additional cellular material will be needed for molecular analysis and to maintain at least 1 uncoverslipped stained slide at the time of the FNA procedure that can be immediately sent for EGFR and/or KRAS testing. Nevertheless, it is helpful to know that coverslipping and subsequently decoverslipping the smears does not compromise the quality of DNA that is isolated for molecular studies. Hence, if molecular studies are required retrospectively after the cytologic diagnosis is made, coverslipped smears can be effectively used in a scenario in which the cell block exhibits insufficient cellularity.

A related key finding in our study is that rapid Romanowsky–stained slides can provide high-quality DNA, even if archived for a prolonged period. Specifically, in our study, we were able to isolate high-quality DNA from smears that were up to 5 years old. A recent study by Killian and coworkers26 reached the same conclusion by demonstrating that DNA obtained from rapid Romanowsky–stained direct smears may be used successfully in high-resolution comparative genomic hybridization arrays, DNA methylation assays, and single nucleotide polymorphism genotyping platforms. These findings are noteworthy in that, for these analyses, high-quality, high-molecular-weight DNA was required. Their study revealed that DNA is stable in the oldest tested specimens (ie, 10 years old) and that Papanicolaoustained slides surprisingly yielded DNA of reduced quality. They postulated that hematoxylin, which is included in the Papanicolaou stain, could lead to DNA degradation in these specimens, which correlates with findings in previous studies.27 These findings, in conjunction with our results, are serendipitous in that rapid Romanowsky–stained slides are optimal for immediate on-site triaging at the time of the FNA procedure. This staining method is inexpensive and rapid and allows for easy visualization of tumor cells to be extracted. Furthermore, tumor-enriched areas can be easily identified and marked before triaging for molecular studies.

The ability to visualize and pathologically vet material marked for subsequent molecular analysis is also an important feature in the present study that warrants further emphasis. In the recent study by da Cunha Santos and coworkers,24 archived material stored on Whatman filter papers (FTA cards), although sufficient for providing high-quality DNA, did not always show the same molecular mutational profile as material harvested from corresponding cell blocks in the same FNA specimens. The authors pointed out that different tumor areas may demonstrate distinct molecular profiles (ie, tumor heterogeneity), and this offers an explanation for discordant findings between cell blocks and Whatman FTA samples because they were obtained from different passes from the same specimens. However, another possibility is that tumor cells were not well-represented in cellular material stored on the Whatman FTA cards from discordant samples; rather, these samples may have predominantly contained contaminating benign cells such as inflammatory cells, bronchial epithelial cells, or histiocytes.

In the present study, these variables were controlled because the presence and proportion of tumor cells were verified on the direct smear before extraction. The area on the smear that is selected for DNA extraction ideally contains a large number and high representation of tumor cells (>40%). However, the minimum required proportion of tumor cells will depend on the analytic sensitivity of the specific mutation assay used by the testing laboratory and could range from about 1% to 40%. By using our PCR-based fragment analysis and direct sequencing–based tests, we were able to detect an EGFR mutation in a case with 15% tumor cells (Table 1, case 10) and a KRAS mutation in a case with 5% tumor cells (Table 1, case 23). Rapid Romanowsky–stained direct smears provide a feasible, robust source of DNA for use in EGFR and KRAS mutation analyses. Given the ability to immediately triage material and ensure adequacy for molecular ancillary studies, this method potentially represents an improved paradigm in acquiring and archiving FNA material for these studies.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.
  16. 16.
  17. 17.
  18. 18.
  19. 19.
  20. 20.
  21. 21.
  22. 22.
  23. 23.
  24. 24.
  25. 25.
  26. 26.
  27. 27.
View Abstract