3.1 Patient Characters
Between 2016 and 2018, 1755 lung cancer patients treated at Henan Cancer Hospital underwent genetic testing. Patients who had received systemic treatment, i.e., chemotherapy or targeted therapy, or being diagnosed as other histological subtypes of lung cancer, were excluded, leaving 853 treatment-naive LUAD patients in this Henan Cancer Hospital (HNCH) cohort. The cohort included 450 (52.8%) males and 403 females (47.2%), with over 90% of the patients having stage-3 (19.2%, 164/853) or stage-4 (74.3%, 634/853) disease at diagnosis. Patients of ≥60 years old at diagnosis account for 59.3% (506/853) of the HNCH cohort, and 40.4% (345/863) of the patients had a smoking history (Table 1). Clinical characteristics of the patients in our HNCH cohort are strikingly different from the TCGA study with 77.6% of the patients presenting with early-stage disease, 69.7% of the patients presenting with LUAD ≥60 years, and 82.8% of the patients having a history of smoking. Although the disease incidence of both HNCH and TCGA patients peak at 61-70 years old, the secondary peak occurs at 51-60 yeas old in HNCH, and at 71-80 in TCGA patients, respectively; HNCH patients present the disease at a significantly earlier age (Supplementary Figure 1A and B). Secondly, smoking is more prevalent in TCGA patients than HNCH patients, particularly in females (1.5% vs 77.8%) (Supplementary Figure 1C).
3.2 Mutation Landscape of the TKI-treatment-related Genes in HNCH Patients
The average sequencing depth was 917×, and over 97.75% of the bases had a coverage of over 50×. In total, 574 single nucleotide variants (SNVs), 270 indels and 88 amplifications were identified in 857 treatment-naïve HNCH patients (Supplementary Tables 2 and 3). Of these patients, 209 (24.5%) did not carry any mutation in the driver gene hotspots, 419 (49.1%) had one mutation and 225 (26.4%) had multiple mutations (Figure 1). The proportion of TCGA patients carrying no mutations was significantly higher (36% vs 24.5%; χ2 test, p=.000), while fewer TCGA patients carried multiple mutations (19.7% vs 26.4%; χ2 test, p=0.006). Overall, we observed a lower average mutation count of 0.93 mutation per patient (m/p) in TCGA patients versus 1.09 m/p in the HNCH patients (Figure 1). Subgroups comparisons revealed a higher average mutation count in HNCH females than males (1.16 vs 1.03 m/p) and in non-smokers than smokers (1.14 vs 1.02 m/p). In the TCGA study, however, this number was higher in males than females (0.95 vs 0.91 m/p) and in smokers than non-smokers (0.96 vs 0.79 m/p). In both cohorts, the mutation frequency was higher in high-stage patients as compared to low-stage patients.
We further investigated whether detected variants were enriched for any of the known mutation signatures. In HNCH patients, smokers and males showed a high prevalence of C>A conversions, a signature known associated with tobacco smoking, while non-smokers and females showed a unique, albeit lower, prevalence of T>A conversions. C>T mutations were quite common in all cases, without a clear trend with gender and smoking. In the TCGA study, C>A conversions were common in both males and females and more enriched in smokers vs non-smokers. C>T conversions, which were prevalent in HNCH patients, were rare in TCGA patients (Supplementary Figure 2). These differences might all be attributed to the differences in smoking prevalence between males and females in the HNCH cohort that was not seen in the TCGA cohort.
3.3 SNVs and Small Indels Detected in This Study
In general, the variants found in the recommended eight genes were mutually exclusive, with EGFR and KRAS being the most commonly affected genes, accounting for 58.7% (501/853) and 12.4% (106/853) of the cases respectively (Figure 2A). EGFR mutations were more common in non-smoking females, whereas KRAS mutations were more common in males who smoked. In the TCGA study, KRAS mutations were prevalent in both males and females (Figure 2B). In both HNCH and TCGA, a binary logistic regression model identified being female and non-smoking as risk factors for carrying EGFR mutations and smoking as a risk factor for carrying KRAS mutations (Figure 2C).
Targetable EGFR mutations were detected in 54.7% (467/853) of HNCH patients, with Exon19 deletions (Exon19del) and L858R being the most common ones, accounting for 38.0% (194/511) and 37.6% (193/511) of all EGFR mutations, respectively (Supplementary Figure 3A). E746_A750del is the dominant Exon19del subtype (75.2%, 146/194). Uncommon EGFR activating mutations, e.g., Exon 20ins (4.7%, 24/511), G719X (2.9%, 15/511) and S768I (1.6%, 8/511), were rare in the HNCH cohort. The T790M mutation, a commonly observed EGFR-TKI resistant marker, was detected in 5 out of 853 (0.6%) treatment-naïve patients. Of note, the oncogenic potential is still unclear for the 44 uncommon EGFR mutations identified in this study. KRAS mutations, detected in 12.4% (106/853) of HNCH patients, were predominantly located in exon 2 with recurrent conversions resulting in a change of G12 to A/C/D/F/L/V and the AMG510-targeting G12C is dominant (Supplementary Figure 3B). Oncogenic variants in BRAF were identified in 49 patients, with the canonical V600E mutation being observed in 20.4% (10/49) of the cases. MET exon14 skipping mutations were identified in 3 cases. SNVs and small indels in ALK, ERBB2, MET, RET and ROS1 were less common in our cohort (Supplementary Table 2).
3.4 Sensitivity of EGFR rare Mutations to Different EGFR-TKIs in-vivo
EGFR rare mutations represent a heterogeneous subgroup of genetic aberrations that their response to different EGFR-TKIs varies, in which the compound mutations, i.e., a common mutation plus an uncommon mutation, or two uncommon mutations in cis, are more complicated. We randomly selected 8 out of 44 EGFR rare mutations detected in this study, including 2 single mutations (V765L or Q849K) and 6 compound mutations (Exon19del plus L747S/I744V, L858R plus V843I/T854A/G873E, and E709A plus G719A), to determine their response to different EGFR-TKIs in vivo. Genetically tailored PC9 cells were established by CRISPR/Cas9 editing and the mutations introduced were validated by Sanger sequencing (Supplementary Figure 4). The cells were treated with different EGFR-TKIs at a dose ranging from 0.6 to 6,000 uM for 72h. MTT assays showed that PC9 cells carrying one of the above-mentioned EGFR variants were overall more sensitive to afatinib and osimertinib as compared to gefitinib and erotinib; particularly as compared to osimertinib, afatinb showed a more constant and consistent efficacy in all these rare and compound mutations, and the cells appeared to respond to afatinib at lower concentrations (Figure 4).
3.5 CNV Changes and Chromosome Rearrangements
Amplifications of EGFR, MET and ERBB2 gene loci were identified in 6.8% (58/853), 2.6% (22/853) and 0.9% (8/853) of the HNCH patients respectively, including three patients with dual amplifications (Supplementary Table 3). EGFR amplifications were observed in 7/410 EGFR wild-type patients and in 51/443 EGFR mutant patients. ALK rearrangements were detected in 59 patients with EML4 being the most common translocation partner (53 cases). Other rare ALK fusion partners included ATIC, CCNY, DCTN1, ERBP1, KLC1 and LOC730100. ROS1 rearrangements were observed in 19 cases, with CD74 being the fusion partner in 10 cases. Other less common fusion partners included MYO5C, SLC34A2, SDC4, TCOF1 and YWHAZ. RET rearrangements were observed in 9 cases, with KIF5B as the most frequent fusion partner, followed by CCD6, EPS8 and TBC1D32 (Figure 5). Taken together, structural variants were identified in 169 patients, accounting for about 20% of our cohort. An attractive exploration in the future would be clarifying the functional consequences of rare fusion partners.