Somatic mutation detection efficiency in EGFR: a comparison between high resolution melting analysis and Sanger sequencing

A prospective series study in which consecutive FFPE tumor samples, sent to Molecular Oncology Diagnostics Laboratory (MODL) for SEQ of the EGFR gene TKD region and mutation analysis were subjected to HRM. Samples irrespective of the tissue of origin, stage of the disease and demographic characteristics were included in the study. Following the screening of 275 FFPE samples, a total of 116 samples were taken for the comparative study. Pathogenic variations in EGFR TKD were identified in 75 out of 275 samples by SEQ. HRM was performed in 67 of the 75 positive samples, and first 49 of the 200 negative samples, in chronological order. Eight positive mutant samples were not included for comparative analysis due to low quantity of DNA.

Tumor tissue samples obtained from patients were subjected to grossing and dissection, fixed in neutral buffered formalin solution, embedded in paraffin, sectioned, and stained. The tissues in paraffin blocks (FFPE) were sectioned and subjected to histopathological examination by the pathologists in the Department of Pathology. Following the determination of the histological as well as immunohistochemical characteristics of the tumor, and the subsequent assessment of the adequacy of the cellular content in the sections, region of interest from each tumor tissue was selected to enrich tumor cell content to about 50% for molecular genetic analysis. Approximately 10 sections of the selected regions of each tumor tissue, with a thickness of 10 μm, were obtained in a 1.5 ml tube and transferred to MODL at room temperature for mutation analysis.

The DNA from FFPE sections was extracted using QIAmp® FFPE DNA extraction kit (Qiagen, USA), according to the procedure described by the manufacturer. DNA quantification was performed by checking the absorbance at 260 nm and 280 nm by spectrophotometric analyzer (Thermo Fisher Scientific, USA), and fluorometric method using Qubit3.0® fluorometer (Thermo Fisher Scientific, USA). Downstream processing of the extracted DNA was performed only when the ratio of absorbance at 260/280 was ≥1.8 with the concentration of DNA ≥ 5 ng/μl.

In a 20 μl assay, 1X Emerald GT PCR master mix (Takara/Clontech, USA) was added, along with m13-tagged forward and reverse target primers (5 μM). Approximately 50 ng of template DNA is added, in a typical assay, and made up with distilled water. Primers for exons 18, 19, 20, and 21 of the EGFR gene (NCBI Genbank Accession ID: NM_005228.3) were synthesized (Merck-Sigma, Bangalore, India). Design and characterization of the primer sequences for both sequencing and HRM were obtained from a previously published literature [18]. Thermal cycler settings included an initial denaturation of 95 °C for 15 min, followed by 45 cycles of denaturation at 94 °C for 45 s, annealing at 58 °C for 45 s, extension at 72 °C for 45 s and a final extension at 72 °C for 10 min. The amplicons were assessed using 2% Agarose gel (SeaKem® LE Agarose, Lonza, USA). The PCR products were then subjected to post-PCR clean up to remove residual primers and other enzyme proteins using HighPure® PCR product purification kit (Roche Molecular Diagnostics, Switzerland).

In a 20 μl assay, 1X of High-Resolution Master mix (Roche Molecular Diagnostics, Switzerland), 300nΜ each of forward and reverse primers, and 2.5 mM of MgCl₂ were added. As described in the previous study [18], 5 ng of template DNA was added, and the assay was made up with PCR-grade distilled water. The assay strip tubes were loaded onto the LightCycler480®. The assay was optimized based on previously described method [18] with minor modifications. The standardized thermal cycler settings include - an initial denaturation at 95 °C for 15 min followed by 50 cycles of denaturation at 95 °C for 10 s, annealing at 65 °C for 10 s and extension at 72 °C for 30 s, with initial 10 cycles of touchdown (1 °C /cycle). This is followed by final denaturation at 95 °C for 1 min and cooling at 4 °C for 2 min. The high-resolution melting was performed from 65 °C to 95 °C at a ramp rate of 0.02 °C/s with 25 fluorescence data acquisition points, followed by cooling to 4 °C for 30 s.

DNA sequencing using Sanger’s dideoxy method was performed using BigDye® Terminator cycle sequencing kit v3.1 compatible with ABI 3500® Genetic analyzer. The original reaction setup, recommended by the manufacture, was optimized with the following modifications. In a total assay volume of 10 μl of separate forward and reverse reactions, BigDye® Terminator, 1X sequencing buffer, 0.8 μM of m13 forward or reverse primer were added to the respective reaction assays. Post-PCR purified DNA amplicon was added, and the volume was made up with distilled water. Assay conditions were setup by the manufacturer recommendations with the following modifications. Initial denaturation at 96 °C for 1 min, 15 cycles of denaturation (at 96 °C for 10 s), annealing (at 50 °C for 5 s) and extension (at 60 °C for 75 s), followed by final extension (2 set of 5 cycles with expanding the extension phase 60 °C to 90 s and 2 min subsequently). The amplified product was stored at 4 °C. Post-sequencing PCR clean-up was done using Sephadex® G-50 medium (molecular weight cut-off of 30,000 Mr) (GE Lifesciences), and Whatman UNIFILTER® filtration plates (Sigma, St. Louis MO, USA). The products were then subjected to clean up by Sephadex gel column filtration using Sephadex® G-50 medium. Capillary electrophoresis was performed using ABI 3500® Genetic analyzer (Thermo Fisher Scientific Inc. USA). The electropherogram obtained from the ABI 3500 Genetic Analyzer is exported to the sequence analysis software, Codoncode® aligner program version 7.0. Comparison of sequences in the contig automatically identifies the variation in the sample sequence when aligned to a reference sequence, and located in the genome by Basic Local Alignment Search Tool (BLAST) [19] from National Center for Biotechnology Information (NCBI). Pathogenicity of variants was identified from different public databases such as dbSNP (NCBI, NIH USA), ClinVar (NCBI, NIH USA), and/or COSMIC database (Sanger Institute, UK). The pathogenicity of variants that were not reported in public databases or previous publications, were tested using computer-aided public accessible prediction tools such as MutationTaster [20], Polyphen [21] or Sorting Intolerant from Tolerant – SIFT- algorithm [22].

Real Time PCR was performed using a CE/IVD approved commercial assay kit (TruPCR EGFR kit v2 from 3B BlackBio Biotech Ltd., Bhopal, India) for the detection of exon 19 deletion. In a 20 μl assay, 10 μl of 1X Multiplex Master mix and 5 μl of a probe mix was added along with 5 ng of template DNA (final concentration: 0.25 ng/μl). The probemix contains FAM labelled primer-probe specific for detection of exon 19 deletion and VIC labelled primer probe for exon 2 region of the EGFR gene, for detection of wildtype DNA as internal control. The assay conditions were as follows: Initial denaturation of 94 °C for 10 min, followed by 10 cycles of denaturation at 94 °C for 15 s and annealing at 68 °C for 30 s. This was followed by 40 cycles of denaturation at 94 °C for 15 s and annealing at 60 °C for 1 min. The assay was performed in LightCycler® 480 real-time PCR machine (Roche Molecular Diagnostics). The cycle threshold (C_t) value for both FAM and VIC dyes were obtained for further analysis.

HRM was performed using the LightCycler® 480 real-time PCR and data was analyzed in LightCycler® 480 Gene scanning software 1.5, Windows version. In a typical HRM, during the melting phase, with increasing temperature, fluorescence decreases as the saturated dye detaches from the denatured amplicon DNA. The change in fluorescence per unit change in temperature (dF/dT) is plotted. Since the fluorescence intensity changes (HRM signal curve) for different samples have different end points, the data is normalized using the pre-melt and post-melt temperature ranges. Hence, each species of DNA is delineated according to the melting temperature. In the difference plot, the pre-assigned wildtype DNA is set as the baseline curve and the DNA from tumor tissues are plotted with different colors from that of the wildtype DNA. Significant deviation from the baseline curve is indicative of and assigned as mutant species by the software. The standard sensitivity is kept as 0.3 and as the sensitivity is increased, smaller deviations can be identified as mutant. The difference plot of heteroduplexes (where one strand is mutant and other strand is wildtype) shows maximum deviation from the wildtype species, while that of homoduplex mutant DNA shows intermediate deviation. Samples with an aberrant HRM curve are identified as mutant species. HRM results were compared with the results from SEQ for validating HRM analysis in the detection of EGFR mutation (Fig. 1).

Statistical analysis was performed using IBM SPSS Windows version 20.0 software. Categorical variables are expressed using frequency and percentage. Diagnostic measures such as sensitivity, specificity and accuracy were calculated. McNemar’s test was used to test the statistical significance of the difference between HRM compared to standard test of SEQ.

Initial amplification PCR of exons 18–21 was standardized by performing a series of gradient PCRs to better accommodate assays for all four exons in a single thermal cycler setting. We found that at a lower annealing temperature than that described in previous studies, samples with low yield were better amplified, due to controlled reduction in specificity of the primers (see details in the discussion section). The standardization of the annealing temperature was performed using FFPE tissue-derived wildtype DNA (Fig. 2). At 58 °C, we observed satisfactory amplification for all four exons, following which all SEQ initial amplification PCR was conducted at this temperature.

With SEQ, 275 FFPE tumor samples received at MODL were screened for sequence variations in EGFR TKD (exons 18, 19, 20, and 21). About 27% of the samples (75/275) contained at least one pathogenic variant in any of the four exons. However, samples with well-differentiated adenocarcinoma of lung showed a 37% mutation occurrence rate. On the other hand, in poorly differentiated adenocarcinoma, mutation was detected in only 11% of the samples. Among the different exons of EGFR TKD which had mutation, exon 19 variations constituted about 69% (52/75) of the cases. Mutation in exon 21 was found in about 21% of the cases (16/75), and exons 18 and 20 variations were detected in one and eight samples, respectively.

The DNA concentration for a typical SEQ was 2.5 ng/μl. In order to determine the lower limit of detection (LOD) of MAF, we used two commercially available standards of EGFR TKD mutants (mutDNA), 1) isolated DNA containing the exon 19 deletion, p.E746_A750del (Ex19_std) with 50% MAF (# HD251, Horizon Discovery Ltd., Ireland, UK), each vial contains 1 μg DNA in Tris-EDTA buffer (pH: 8.18, concentration: 50 ng/μl) with 50% mutant allele fraction of EGFR ΔE746-A750 (SNP ID: rs121913421) validated by digital droplet PCR, and 2) FFPE tissue containing the exon 21 mutation, L858R (Ex21_std) with 50% MAF (#HD130, Horizon Discovery Ltd., Ireland, UK) each vial contains one section FFPE cell pellet of human cell lines, 15-20 μm thick, with an approximate cell density of 3.5 × 10⁵ cells/section, containing roughly 400 ng total DNA with 50% mutant allele fraction of EGFR L858R (SNP ID: rs121434568) validated by digital droplet PCR.

The LOD of MAF of Ex19_std and Ex21_std was separately determined by performing a series of assays with different total DNA concentrations ranging from 0.25 ng/μl (5 ng/assay of mutDNA + wtDNA) to 2.5 ng/μl (50 ng/assay of mutDNA + wtDNA). Figure 3a and b show the chromatograms of Ex19_std at two different total DNA concentrations (a: 2.5 ng/μl, b: 0.25 ng/μl), and Fig. 3c and d show the chromatograms of Ex21_std at two different total DNA concentrations (c: 2.5 ng/μl, d: 0.25 ng/μl). As seen from Fig. 3b and d, there was a sudden absence of the mutant peak at 5% MAF for Ex19_std and 2.5% MAF for Ex21_std at 0.25 ng/μl assay. However, in the 2.5 ng/μl, the mutant peak was not detected at 0.5% MAF for Ex19_std and 0.125% for Ex21_std. This shows that the LOD of MAF for a 0.25 ng/μl assay was 10% for Ex19_std, and 5% for Ex21_std for a 0.25 ng/μl assay. On the other hand, in a 2.5 ng/μl assay, lowest LOD for Ex19_std and Ex21_std were 1% and 0.25% respectively (Fig. 3a & c).

Since the origins of wtDNA and mutDNA were different, there can be a difference in amplification due to the quality of DNA. In order to address this, we performed SEQ with both wtDNA and mutDNA separately as well as in 1:1 mix. At 25% MAF, the wtDNA: mutDNA ratio was 1:1, for a total DNA concentration of 2.5 ng/μl. The peak heights of both wtDNA and mutDNA species were found to be equal, irrespective of cell-derived or tissue derived DNA (See Fig. 3b, d and Supplementary Fig. S1c). Moreover, the amplification peak heights of both tissue-derived wtDNA and cell-derived mutDNA, when sequenced separately were comparable as that of 1:1 mix (Fig. S1a and b). This shows that there was minimal difference between the quality of DNA obtained from both sources.

With the intention of assessing both mutant and wildtype DNA amplification in mixed assays used in SEQ, multiplex real time PCR with allele-specific primer-probe for the detection of exon 19 deletion was performed. MAFs of 10%, 5%, 1% and 0.1% were assessed in triplicates in mixed DNA assays. Similarly, Ex19_std and wtDNA were also separately assessed. The multiplex allele-specific primer-probe is incorporated with two fluorescent dyes, FAM and VIC, detecting mutant and wildtype alleles, respectively. Accordingly, the amplifications of both the alleles were determined by the C_t values obtained for both dyes, respectively, in each assay. According to the manufacturer’s instruction, ΔC_t (The difference between C_t of mutant and C_t of wildtype) ≤ 12, indicates positive for the deletion of exon 19. The FAM C_t values showed wide variation across all MAFs, the lowest being detected in 50% MAF, while the highest in 0.1% MAF (Fig. 4 and Table 1). On the other hand, VIC C_t was comparable in all assays between all the MAFs (Fig. 4c). ΔC_t of all the MAFs, except 0.1% MAF, were detected as positive for exon 19 deletion (Table 1). Thus, the wildtype DNA as well as the mutant standard DNA were amplified in the mixed assays, thereby validating the lowest MAF detected in the identical assay using SEQ.

In order to assess the HRM detectability, a previous study by Do et al. included 5 ng of DNA per 20 μl assay (0.25 ng/μl final concentration) with 50% MAF [18]. We examined the LOD of initial DNA concentration by performing a dilution curve of mutDNA standard mixed with wtDNA in a real-time PCR assay using the same HRM conditions. Different mutDNA standard concentrations from 0.5 pg/μl to 0.25 ng/μl at 50% MAF were employed and assessed against C_t of real-time PCR assay. C_t value increases with decreasing concentration (Fig. 5). We found that the relative change in C_t values (ΔC_t) was proportional to the natural logarithm of DNA concentration/assay as per the equation given below.

By applying the above equation, the saturation concentration of DNA for no change in ΔC_t value was ~ 0.249 ng/μl. Hence, the total DNA concentration/assay was fixed as 0.25 ng/μl for heteroduplex HRM analysis. This concentration is 10 times lower than the amount of DNA required per assay for SEQ.

In order to examine whether the LOD of MAF decreases with increasing concentration of total DNA (mutDNA+ wtDNA) per assay, we titrated mutDNA with wtDNA in a series of separate HRM assays containing varying total DNA concentrations. HRM MAF titration was conducted by incorporating decreasing MAF in a mixed DNA sample from 10% to 0.005% of both Ex19_std & Ex21_std, in an increasing total DNA concentration/assay from 0.25 ng/μl to 2.5 ng/μl (Figs. 6 and 7). For both the mutDNA standards, lowest LOD observed was 0.25% in a 2.5 ng/μl HRM assay, while in a 0.25 ng/μl assay, 5% MAF was the lowest LOD detected with the Ex19_std, and 10% with the Ex21_std.

Of the 116 samples that were included in the study, 67 samples were mutation positive using the SEQ method (Table 2). Forty-seven of the 67 positive samples had deletion mutation in exon 19, and substitution mutations were detected in 15 and five samples in exon 21 (Table 3), and exon 20 respectively. None of the samples in this group were found to have a mutation in exon 18. Hence, the performance characteristics for exons 18 and 20 were not separately included in the comparative performance analysis. HRM positive samples were distributed as 72% positive and 28% negative in SEQ samples with the exon 19 mutation (Table 3). Furthermore, HRM positivity was distributed as 60% positive and 40% negative in SEQ samples with the exon 21 mutation (Table 3). The sensitivity of detection was 100% for both exon 19 and exon 21 mutations, while specificity was higher with the exon 21 variants compared to exon 19 variants (90% vs. 74%). McNemar’s comparison of these two methods was significant (p value < 0.01) for both the groups, and overall performance differences between these two methods. Overall specificity of HRM over SEQ was about 67% with 86% accuracy (Table 2). The positive predictive value (PPV) was 60% for exon 21 variants, about 72% for exon 19 variants, with an overall 80% PPV. However, the negative predictive value (NPV) was 100% in individual variant groups and overall values as well.