Materials and methods
Selection and processing of pathomorphological samples
Cytological material from 78 patients with NSCLC was collected at the Department of Tumor Pathology and Pathomorphology, Oncology Center, Franciszek Lukaszczyk Memorial Hospital in Bydgoszcz, Poland. Informed consent for mutation testing was obtained from all the patients.
A total of 78 specimens were obtained following 67 FNA procedures, 7 bronchial brush procedures, 3 EBUS-TBNA procedures and 1 pleural liquid sampling. In the 78 patients, 75 adenocarcinoma subtypes were determined, and all were tested for the presence of EGFR mutations and the majority of samples were tested for KRAS mutations. Four cytological samples did not pass quality control steps (pathomorphological qualification or qualitative and quantitative DNA analysis). Biopsy samples were also stained with hematoxylin and eosin (H&E) for qualitative and quantitative analysis of tumor cells in the analyzed material (including macrodissection in marked out samples). Representative cytological smears were subjected to molecular oncological and genetic assessment of mutations in the EGFR and KRAS genes. Two clinical pathomorphologists (W.J. and C.J.), who were unaware of the patient characteristics, examined the cytological smears as follows. In each smear, two elements were evaluated by the number of neoplastic and non-neoplastic nucleated cells, using a subjective method of microscopic counting of 10 neighboring cells as the smallest virtual ‘decimal cell group’, dispersed uniquely within each tissue section, then 10-folded to the ‘100-fold’ and ‘1,000-fold’ cell groups in the two following steps, yielding the ‘10-fold’, ‘100-fold’ or ‘1,000-fold’ cell measures, respectively, to perform an approximate cell number estimation, when needed, with an accuracy of ~10%. Qualification of cytological material for molecular analysis was based on the following quantitative scale (QS): C1+, ≤100 tumor cells; C2+, >100–1,000 tumor cells; C3+, >1,000–5,000 tumor cells; C4+, >5,000–10,000 tumor cells; C5+, >10,000 (countless) tumor cells) (Table I). Two samples were not qualified due to the small number of cells (no more than 20 tumor cells) (Table II). The percentage of tumor cells (PTCs) was analyzed based on the number of neoplastic cells compared to all nucleated cells in the cytological specimens.
DNA isolation from biopsy smears (after H&E evaluation and image acquisition) was conducted by immersing cytological samples in xylene and by incubation at room temperature overnight or until cover slips were removed (up to 5 days). Once cover slips were taken off, the samples were incubated for 5 min in 100% ethanol, and then washed with 70% ethanol to rehydrate the tissue before the cells were scraped with a sterile scalpel into a tube. Subsequently, DNA was isolated using the QIAamp DNA FFPE tissue kit.
All isolation procedures were performed according to the manufacturer’s instructions. DNA was subjected to qualitative and quantitative analysis (NanoDrop; Thermo Scientific) and DNA isolates with an A260/A280 ratio ranging from 1.8 to 2.0 were tested for the presence of mutations in the EGFR and KRAS genes.
Detection of EGFR, KRAS and BRAF mutations
All 29 EGFR mutations in exons 18, 19, 20 and 21 were evaluated via the real-time polymerase chain reaction (PCR) methodology using mutation-specific oligonucleotides (EntroGen). A DNA quantity of 600–650 ng was adequate for the detection of 29 mutations in the samples of interest; however, in a few cases the total amount of DNA was reduced to 250 ng for the detection of the 29 mutations due to the high quality of DNA, even though a lower DNA yield was obtained from biopsy samples. Such amount of DNA was in accordance with the instructions for the detection of the 29 EGFR mutations, in which the recommended minimum DNA quantity was 50 ng. Mutation analysis was identical to the manufacturer’s protocol for EGFR mutation analysis using real-time PCR (EntroGen) performed using LightCycler® 480 II (Roche). For the purpose of validation, direct Sanger sequencing was performed, followed by real-time PCR of the most common mutations in exons 19 and 21. This assay reports the presence of the following EGFR mutations: 3 mutations in codon 719 in exon 18; 19 deletions in exon 19; 3 insertions, as well as p.T790M and p.S768I mutations in exon 20; p.L858R and p.L861Q mutations in exon 21. Each analysis was initiated by checking the endogenous control amplification plot of every sample (VIC detector). When endogenous Ctrl CT ranged from 22 to 32, FAM detector was used. When no FAM signal was present, the sample did not contain the selected mutation. Normal positive CT values (mutation detected) may range from 20 to 37. When the CT value was >38, the reaction alone or the entire procedure including DNA isolation was repeated. The EGFR mutation assay was established to have an analytical sensitivity of 1% based on dilution studies prepared by EntroGen.
In order to investigate the mutation status in codons 12 and 13 of the KRAS gene in 54 cytological samples, HybProbe and melting curve analysis (LightMix® Diagnostic kits; TIB MolBiol) were performed. LightCycler® FastStart DNA Master HybProbe was prepared according to the manufacturer’s instructions, along with the controls and the CTRL, LOW and HIGH reactions with the tested DNA. This assay reports the presence of KRAS mutations located in codons 12 and 13: c.34G>C, c.34G>A, c.34G>T, c.35G>A, c.35G>T, c.35G>C, c.37G>T and c.38G>A. Following real-time PCR analyses, melting curve analysis was performed according to the following protocol: the 13C reaction exhibits a peak at ~56°C, the 12C reaction exhibits a peak at ~68–70°C, the NTC reaction exhibits baseline value, the WT reaction exhibits a peak at ~64–65°C, the cWT reaction exhibits baseline value or a maximum of 10% value of the WT reaction signal. Reaction for a sample was analyzed when the control reaction met specific criteria. Wild-type was reported when a sample displayed no peak in the HIGH reaction and the LOW reaction displayed a wild-type-specific peak at the same time. The CTRL reaction also had to demonstrate a wild-type-specific peak. A mutation was reported when a sample displayed a clear peak in the HIGH reaction. The LOW reaction might return the same result or display two peaks corresponding to the wild-type and mutation. A mutation was also reported when a sample returned no peak in the HIGH reaction but had a clear peak at any temperature between 50 and 65°C in the LOW reaction. The CTRL reaction had to show a wild-type-specific peak or display two peaks corresponding to the wild-type and mutation.
Selected samples were confirmed using additional methods: EGFR mutation-positive samples via Sanger sequencing (ABI Prism 3130xl Genetic Analyzer), and KRAS-BRAF mutation-positive samples via Strip Assay, which also allowed the analysis of BRAF status in codon 600 (ViennaLab Diagnostics GmbH) according to the manufacturer’s instructions (30).
Discussion
The presence of an EGFR mutation in the tyrosine kinase domain constitutes an important predictive factor for the response to treatment with TKIs (9). Therefore, it is crucial to analyze the EGFR mutation status using the most reliable method, which permits detection of activating and inhibiting EGFR mutations.
In the real-time PCR and new generation sequencing era, small quantities of input material may be used for the analysis of numerous mutations. However, the heterogeneous character of tumors should not be underestimated. Therefore, quantitative estimation of cytological specimens should be performed in the preanalytical step, in order to determine mutations in as many tumor cells as possible. The findings of the present study suggest that EGFR mutations are detected in DNA isolated from heterogeneous cytological material containing ≥1,001 tumor cells. Therefore, the EGFR mutation analysis in cytological specimens may be performed as a routine evaluation in patients with inoperable NSCLC. In contrast, unequal cytological sampling could be potentially misleading. Any EGFR WT result should be carefully assessed as it might reflect sampling bias, particularly when the analysis was performed using DNA isolated from a small number of tumor cells (21–1,000). Taking into consideration the histological heterogeneity of NSCLC, genetic differentiation implied as local divergence of mutation status within a tumor may also be possible. Thus, the fact that cytological samples become less representative with a decreasing number of tumor cells cannot be excluded. In such a case, potential EGFR WT results obtained following analysis of cytological specimens with a small number of tumor cells might be considered as potentially containing false negatives, and different results might be obtained following the analysis of an additional FNA sample from the same patient. Cytological material may be analyzed using smeared slides or cytological cell blocks. This leads to the serious drawback of the time-consuming digital archiving of smear slides which is ‘sacrificed’ for DNA isolation.
Cytological smears constitute material more difficult to handle when compared to paraffin-embedded tissues; therefore, it is highly important to use well-validated, sensitive methods which do not provide false-positive results, and also decrease any discrepancies and false-negative results of the EGFR mutation status analysis, when DNA material derived from cytological samples is limited. Currently, different methods of EGFR status assessment are used, ranging from immunocytochemistry (10) and qualitative or quantitative PCR methods (11) to sequencing (including either Sanger or new generation sequencing methods). Taking into consideration that in most cases of inoperable NSCLC the existing cytological material is critical, we evaluated the real-time PCR detection of EGFR mutations in exons 18, 19, 20 and 21 of adenocarcinoma cells obtained via FNA, EBUS, TBNA, brushing procedures, and thoracentesis. In the present study, we found that 10% of the 77 analyzed adenocarcinoma samples harbored EGFR mutations, 35% of which carried KRAS mutations, while 51% of the 54 samples analyzed for KRAS mutations were negative regarding both mutation types. KRAS mutation results were moderately high for the detection of KRAS mutations with cytomorphological features of adenocarcinomas (12,13), but are in accordance with the recently reported prevalence of NSCLC patients with KRAS mutations (27%) detected using COLD-PCR (14) and 36.9% KRAS mutations detected in malignant pleural effusion of lung adenocarcinoma in a Dutch population (15).
EGFR mutations detected using an assay based on amplifying mutant-specific sequences (EGFR-RT52; EntroGen) were confirmed by Sanger sequencing. KRAS mutations were detected using high melting temperature (LightMix® kit; TIB MolBiol), a method known to provide false-positive results (16). Therefore all samples with detected KRAS mutations were verified using another method, KRAS Strip Assay. In fact, in the case of the two samples, the use of high melting temperature yielded borderline results [c.35G>A (G12D)], while following verification with a second method (ViennaLab Diagnostics GmbH), the samples were evaluated as KRAS-WT (data not shown). It should be taken into consideration that direct sequencing is not always sensitive enough to detect mutant DNA (17), and that high-resolution melting analysis may also give false-negative results (18). Notably, two false-negative results were found in cytological material with a small proportion of tumor cells, indicating that cytologists should not only distinguish benign and malignant samples, but they should also conduct proper qualification of the material for molecular analysis (18). The real-time PCR analysis of EGFR and KRAS mutations conducted in the present study, led to identification of EGFR c.2582T>A (L861Q) and KRAS c.35G>T (G12V), detected in 1,000–5,000 and 100–1,000 tumor cells, respectively. Thus, our results confirm that cytological samples characterized by QS=C3+ or even C2+ should not be evaluated as inadequate for EGFR mutation analysis using real-time PCR, but should be carefully selected by a cytopathologist.
Mutations in EGFR and KRAS detected in tumors embedded in paraffin blocks and fresh-frozen tumor specimens appear to be mutually exclusive (19). Our results obtained following analysis of cytological material are consistent with this observation (Table I and Fig. 1). Usually, EGFR mutations are characteristic of tumors in the non-smoker groups of patients, particularly among Asian women, while KRAS mutations are often detected in smoking-associated cancer types. Therefore, KRAS mutations have been suggested to constitute a primary mechanism of resistance to gefitinib or erlotinib in lung adenocarcinoma. This finding does not clarify whether this insensitivity is due to the presence of the mutated KRAS or the absence of the mutated EGFR(20). In contrast, the acquired resistance to TKIs has been studied to a great extent, particularly due to mutations in exon 20. The most frequent TKI mutation is EGFR c.2369C>T (T790M) (20), followed by insertions in exon 20 which may render the epidermal receptor approximately 100-fold less sensitive to erlotinib or gefitinib (21). Concerning the findings of the present study, 1 adenocarcinoma patient was found to carry both EGFR activating c.2573T>G and inhibiting c.2369C>T mutations following analysis of material isolated via the FNA procedure, indicating that the mutations were of activating and inhibiting types, respectively. The structure of the wild-type EGFR kinase domain has been published in both active and inactive conformations, while its crystal structure has been reported alone and in complex with erlotinib (22). Along with the discovery of the EGFR c.2573T>G (L858R) EGFR mutant crystal structure, which is TKI-sensitive, it was indicated that the substitution of arginine for leucine at position 858 activates the kinase. Comparison of the fold activity between the wild-type and the L858R mutant enzyme demonstrated an approximately 50-fold higher activity of the mutant conformation (23). Further studies of gefitinib revealed that this 4-anilinoquinazoline inhibitor, structurally similar to erlotinib, binds the EGFR c.2573T>G (L858R) mutant with a 20-fold higher affinity compared to the wild-type enzyme. This higher affinity for the EGFR c.2573T>G (L858R) mutant was explained by tighter binding to the active conformation of the kinase compared to the inactive conformation (23). In contrast, some patients were reported to become resistant following TKI treatment due to mutation in the ‘gatekeeper’ residue of threonine 790 (24,25). Another inhibiting mutation in exon 21 is the A→G change, which leads to the substitution of alanine for threonine at position 854 (5). Both mutations, the activating EGFR c.2573T>G (L858R) and the inhibiting EGFR c.2369C>T (T790M) were found in our cytological assessment.
The cytological samples of the present study obtained via FNA, were found to carry both mutations EGFR c.2573T>G and c.2369C>T (L858R/T790M), demonstrating the advantage of the FNA procedure and the ability of performing serial sampling of a given tumor to assess the efficacy of targeted therapy or identify genetic shifts of adenocarcinoma with EGFR mutations (3). This recommendation does not exclude patients from targeted therapies, since patients with the EGFR c.2369C>T (T790M) mutation of EGFR might still respond to erlotinib after gefitinib treatment (6). In fact, the patient with EGFR c.2573T>G (L858R) and EGFR c.2369C>T (T790M) was treated with erlotinib and a short-term improvement was observed. However, 7 months after the detection of the L858R/T790M mutation, a CT scan of the chest and abdomen revealed metastases to both lung and liver, resulting in the death of the patient.
According to a recent anticancer drug discovery using a high-throughput cell-based assay, inhibitors of the L858R/T790M mutant EGFR pathway were identified after screening 60,000 compounds (26). Those findings may also allow classification of novel inhibitors which suppress mutant EGFR c.2369C>T (T790M) signaling (26). An additional study on a classical protein kinase C inhibitor, revealed high potency against the mutated EGFR and significantly reduced tumor growth in an in vivo xenograft model employing an EGFR-mutant NSCLC cell line containing the EGFR c.2369C>T (T790M) mutation (27). Moreover, novel EGFR TKIs which bind irreversibly to EGFR-TK and form covalent cross-links with EGFR, such as afatinib (BIBW2992), have been demonstrated to be active against tumors resistant to reversible EGFR TKI (28,29).
In conclusion, cytological material is useful for the assessment of the EGFR mutation status for assessment of personalized therapy with EGFR-TKIs. We demonstrated that the real-time PCR methods used here may allow detection of activating and inhibiting mutations. Furthermore, in cytology material with >1,000 tumor cells in samples from a Polish Caucasian population, there was a frequency of 10% EGFR mutations, while in samples with a minimum of 5,000 tumor cells the frequency was 12,24%. Our data suggest the presence of false-negative cases within the EGFR WT results, obtained from the cytological specimens with a small number of tumor cells (even up to 1,000). Molecular EGFR and KRAS testing of cytological material could provide also further information to oncologists and pathomorphologists since the presence of EGFR mutations is mutually exclusive of EML4-ALK transcript and low percentage of KRAS mutation (11%) is found in NSCLC specimens carrying the EML4-ALK transcripts (31).