2.1 Experimental strategy and patient characteristics
Sections were cut from the FFPE block of the primary tumor tissue biopsy for all 51 patients and were subjected to DNA extraction. For 39 patients, the same DNA sample was analyzed by both assays. For 12 patients, different DNA samples (extracted from different sections of the same FFPE block) were analyzed by only one assay, either the Ion Ampliseq Colon Lung v2 assay or the NIPD Genetics custom tumor profile assay (Figure 1A, C). Out of 51 patients, 39 patients had samples that met quality control criteria and were subjected to targeted sequencing with the two different assays (Figure 1C). A total of 12 DNA samples were excluded from analysis by both assays due to low or poor DNA quality. The average age of patients was 60.8 years (±9.2). Out of 39 patients, 38.5% were smokers. Most of the patients were diagnosed with adenocarcinoma (74.3%) and 25.7% of them with squamous cell carcinoma. Five patients (12.8%) were diagnosed with stage IV disease, 35.9% of the patients were diagnosed with stage III, 25.6% with stage II and 25.7% with stage I. Out of 29 adenocarcinoma patients 18 (62%) and out of 10 squamous cell carcinoma 8 (80%) patients had either lymph node or distant metastasis (Table 1).
2.2. Concordance analysis
Overall 31 patients (79.5%) had mutations identified in their tumor specimens by either assay, 27 of which had mutations identified by both assays (69.23%) (Figure 2A). Concordance analysis was performed on sequencing data generated by both assays as described above for the genomic region commonly covered by both methods using the same DNA samples (originating from the same FFPE sections). This commonly covered region consists of hotspot regions of 18 genes with a total genomic size of 8.6 Kb (Figure 1A, B). Thirty-four variants in 25 patients were identified by both methods while one extra variant -the KRAS G13C at 5.9% VAF- was identified only by the NIPD tumor profile assay in patient 1 (Table 2). The IonAmpliseq assay failed to detect this variant above the minimum acceptable threshold of 5%, however, it was detected at VAF=3.78%. Hence, concordance between the two assays was estimated at 97.14% (Figure 2B). The frequencies of the concordant mutations in both assays were highly similar (r2=0.9156, Figure 2B).
2.3. Intra-tumor variability
The processing of DNA samples originating from different sections at different layers of the same patients’ FFPE tissue biopsy block enabled the investigation of intra-tumor heterogeneity. As shown in Figure 2D, two commonly covered variants, TP53 V227G and BRAF V600E were not identified in all sections tested from the same tumor (Figure 2D, Table S1). Given that the two methods have shown performance similarities, this variability does not necessarily reflect differences in the sensitivity of each method but instead could be a result of the heterogeneity of the tissue biopsy samples with different spatial origin within the primary tumor. However, to conclusively characterize the extend of intra-tumor heterogeneity and investigate its clinical importance a large-scale study is required.
2.4. Assessment of the molecular profile for NSCLC patients
Data from all the regions covered from the two assays combined (including overlapping and exclusively covered regions) were used to assess the mutation profile of these patients. A total of 52 mutations were identified in 31 patients. Sixteen patients (51.6%) had just 1 mutation identified in their tumor biopsy (Figure 3A). The total number of mutations did not show significant correlation with tumor content in the FFPE specimens (Supplemental figure 1A). However, there’s a statistically significant increase allele frequency in FFPE samples with tumor content >70% as compared with FFPE samples with tumor content ≤70% (Supplemental figure 1B). The most frequent mutations were identified in TP53 (21 patients, 40.4%), KRAS (7 patients, 13.46%), EGFR (5 patients, 9.62%) and PIK3CA (4 patients, 7.7%) (Figure 3B). Mutations in TP53 were predominantly identified in the DNA binding domain (amino acids 98-292)20, a region known to harbor the majority of deleterious mutations in this gene (Figure 4A). KRAS mutations were predominantly localized on codon 12, a widely studied recurrent region in multiple cancer types while 1 mutation was identified in codon 13 (Figure 4B). All KRAS mutations identified were associated with reduced sensitivity to TKI inhibitors21. EGFR mutations were identified in the tyrosine kinase domain (Figure 4C); four in-frame exon19 deletions and one exon 21 L858R substitution, all known to confer sensitivity to EGFR inhibitors such as gefitinib, afatinib and erlotinib22.
Eleven mutations were identified in regions exclusively covered by only one assay. A MET N375S mutation was identified in patient 42 by the Ion Amliseq assay, while 10 more mutations of clinical significance in STK11, RET, PTEN, GNAS, TP53, BRCA2, PALB2, CHEK2, and PIK3CA were exclusively covered and identified by the NIPD tumor profile assay in 10 patients (Figure 3A, Table S2). Moreover, oncodriver mutations in BRAF, EGFR, KRAS and exon 10/exon21 PIK3CA were found to be mutually exclusive in this cohort (Table 2, Figure 3).
2.5. Comparison of clinical utility
In total, 52 unique mutations were identified, 50 mutations with the NIPD tumor profile assay and 40 mutations with the Ion Ampliseq assay (Table 2, Table S1, Table S2). These, mutations were assessed based on i) their clinical utility including their association with sensitivity to an approved therapy, ii) resistance to an approved therapy (contraindicated therapy), iii) association with an approved therapies in a different cancer type and iv) their investigation in clinical trials (either for their prognostic or potential therapeutic significance). The NIPD Genetics custom tumor profile assay identified more mutations associated with approved therapy in NSCLC as compared to the Ion Ampliseq (6 versus 5 mutations respectively). In addition, 7 mutations associated with resistance to approved therapy (contra-indicated for use) were identified by the NIPD Genetics assay as compared to 6 mutations with the Ion Ampliseq assay. Furthermore, the NIPD Genetics assay identified 8 mutations associated with approved therapy for a different cancer type compared with 3 mutations identified with the Ion Amlpiseq assay. Thirteen mutations identified by the NIPD assay were included in NCCN guidelines for NSCLC as compared to 11 mutations identified with the Ion Ampliseq assay. Finally, 49 mutations associated with clinical trials were identified by the NIPD Genetics assay compared to 40 mutations identified by the Ion Amliseq assay (Figure 5, Table S3).
2.6. Association of mutations with patient survival and clinicopathological characteristics
The prognostic significance with respect to overall survival (OS) of the mutations identified in these patients was explored. Patients were divided in early (stage I-II) and late (III-IV) stage NSCLC with or without mutations in TP53 or KRAS. Although there was no statistically significant difference between the 4 groups, a trend was observed for improved OS in late-stage patients with wt KRAS or TP53 (Figure 6). Due to the limited number of patients in the cohort it was not possible to investigate the prognostic significance of each gene separately. Next, associations of mutations in the 4 most frequently mutated genes with clinicopathological characteristics such as age, histology, smoking status were evaluated. A statistically significant association between KRAS status and histology was observed (p=0.03099). Mutated KRAS was positively correlated with adenocarcinoma as opposed to squamous cell carcinoma (Table 3). However, due to the small size of the patient cohort, the results should be interpreted with caution.