Competitive evolution of NSCLC tumor clones and the drug resistance mechanism of first-generation EGFR-TKIs in Chinese NSCLC patients

Abstract

Purpose: Although many studies have reported on the resistance mechanism of first-generation EGFR TKIs (1^st EGFR TKIs) treatment, large-scale dynamic ctDNA mutation analysis based on liquid biopsy for non-small cell lung cancer (NSCLC) in the Chinese population is rare. Using in-depth integration and analysis of ctDNA genomic mutation data and clinical data at multiple time points during the treatment of 53 NSCLC patients, we described the resistance mechanisms of 1^st EGFR TKIs treatment more comprehensively and dynamically. The resulting profile of the polyclonal competitive evolution of the tumor provides some new insights into the precise treatment of NSCLC.

Experimental design: A prospective study was conducted in patients with advanced NSCLC with acquired resistance to erlotinib, gefitinib or icotinib. By liquid biopsy, we detected mutations in 124 tumor-associated genes in the context of drug resistance. These 124 genes covered all tumor therapeutic targets and related biological pathways. During the entire course of treatment, the interval between two liquid biopsies was two months.

Results: Unlike the common mutations tested in tissue samples, our data showed a higher coverage of tumor heterogeneity (32.65%), more complex patterns of resistance and some new resistance mutation sites, such as EGFR p.V769M and KRAS p.A11V. The major resistance-associated mutations detected were still EGFR p.T790M (45.28%), other point mutations in EGFR (33.9%), and KRAS and NRAS mutations (15.09%). These mutation ratios might be considered as a preliminary summary of the characteristics of Chinese patients. In addition, starting from the two baseline mutations of the EGFR gene (19del vs. L858R), we first described the detailed mutation profile of the EGFR gene. Although there was no significant difference in the number of patients with EGFR p.19del and EGFR p.L858R baseline mutations (24% vs. 16%, P = 0.15), patients from the EGFR p.19del baseline group were much more likely to develop EGFR p.T790M resistance mutations (62.1% vs. 19.3%, P = 0.007). Through careful integration of gene mutation information and clinical phenotype information, an interesting phenomenon was found. Although the variant allele fraction (VAF) of the EGFR p.T790M mutation was significantly linearly correlated with that of the EGFR drug-sensitive mutation (r = 0.68, P = 0.00025), neither VAF was associated with the tumor volume at the advanced stage. It was shown that other tumor clones might contribute more to the resistance to 1^st EGFR TKIs treatment than tumor clones carrying the EGFR p.T790M mutation when resistance developed. By further analysis, we found that, in some patients, when the primary tumor clones detected were those carrying EGFR-/- mutations (both types the EGFR p.19del/p.L858R and EGFR p.T790M mutation types were missing), most of them showed a poor prognosis and ineffective late treatment, indicating that EGFR-/- played a more important role than EGFR p.T790M in the process of NSCLC drug resistance in these patients. From the perspective of the clonal evolution of NSCLC tumor cells, these phenomena could be explained by the competitive evolution between different tumor clones. In addition, two new mutations, KRAS p.A11V and EGFR p.V769M, emerged significantly during drug resistance in NSCLC patients and had shown obvious competitive clonal evolution characteristics. Combined with clear clinical drug resistance phenotypic information, we believed that these two new mutations might be related to new drug resistance mechanisms and deserve further study. We have also seen an interesting phenomenon. In some patients undergoing 1^st EGFR TKIs treatment, the EGFR p.T790M mutation appeared, disappeared, and reappeared, and this spatial and temporal diversity of the EGFR p.T790M mutation was regulated by targeted drug and chemotherapy and was correlated with the individual tumor mutation profile.

Conclusions: The constitution and competitive evolution of the tumor clones have a decisive influence on treatment and can be regulated by targeted drugs and chemotherapy. Additionally, EGFR p.T790M spatial and temporal diversity during treatment warrants more attention, and this spatial and temporal diversity may be useful for the choice of treatment strategies for certain NSCLC patients. Through longitudinal cfDNA sample analysis, the resistance mechanism and dynamic clinical features of Chinese NSCLC patients are systematically established as reliable and meaningful to understand acquired resistance and make further personalized treatment decisions dynamically. Two new potential drug resistance-associated mutations in EGFR and KRAS have been found and are worthy of further study. Finally, our research shows that the evolutionary process of tumor cloning can be artificially regulated and intervened, possibly providing a new way to treat tumors.

Keywords: Oncology.

Fig. 1

Change of Sum of longest diameters (SLD) during from baseline to optimal response during 1st EGFR TKI treatment and summary of the temporal collection of liquid biopsies. (A) According to the standard of the Drug Reaction Assessment (RECIST 1.1) handbook, the red dotted line indicates that the tumor shrank by 30%, representing the drug clinical response of a PR. (B) 32 patients had first cfDNA liquid biopsies at progression. 13 patients started with ARMs method for baseline activating mutation detection and collection cfDNA samples at two-month interval until progression. 8 patients had baseline cfDNA samples and collected cfDNA samples at two month interval until progression. All patients kept collecting after progression at the same time interval if possible.

Fig. 2

Statistics of genes and mutations at drug resistance. (A) The number of mutant genes detected in all patients with resistance was marked in red, and the number of nonsynonymous mutations was marked in green. (B) Percentage distribution of mutant genes in all patients when resistant. (C) Distribution of base substitution types of non-synonymous mutations in all patients when resistant. (D) Recurrent nonsynonymous mutation distribution in all patients.

Fig. 3

Detailed mutation profiles of patients with EGFR p.V769M and KRAS p.A11V mutations. Recurrent and acquired mutations are marked in red during treatment. (A) The mutation profile of three ctDNA assays in P5. (B) Overview of two ctDNA series mutations in P61. (C) The mutation profile of two ctDNA assays in P48. (D) ctDNA mutation profile in progressive P44 with advanced disease. (E) Baseline ctDNA mutation profile in P44.

Fig. 4

Drug resistance mutation profile. (A) Acquired resistance mutant pie at the time of progression. Compared with the baseline mutation profile, the percentage of acquired resistance mutations was described. The blue wedge indicates the proportion of acquired EGFR p.T790M mutations accompanied with other point mutations. The dark blue wedge represents the EGFR p.T790M mutation with EGFR amplification. The red wedge represents the EGFR p.T790M mutation accompanied with EGFR and ERBB2 gene amplification. The bright green wedge represents the proportion of non-synonymous mutations in the PIK3CA gene. The dark red wedge represents the proportion of non-synonymous mutations in KRAS and NRAS genes. The yellow wedge indicates the proportion of non-synonymous mutations in BRAF gene. The black wedge represents the proportion of non-synonymous mutations in ERBB2 gene and the white wedge represents the mutation ratio of ERBB2 gene amplification. The pale green wedge represents the ratio of mutations in MET gene amplification and EGFR gene amplification. The gray wedge represents the insertion mutation ratio of the 20 exons of the EGFR gene. The pink wedge indicates the proportion of EGFR point mutations that do not contain EGFR p.T790M. (B) Statistical pie chart of multi-gene resistance mutation. The yellow wedge indicates the proportion of mutations that occur in multiple genes. The green wedge indicates that multiple mutation ratios occur in a single gene. The blue wedge indicates a single mutation in a single gene. (C) Heat maps of the VAF values for resistance mutations in 49 patients. The patient ID is the y-axis, and the corresponding amino acid change is the x-axis. Three patients with extremely high mutation counts are marked with the blue line.

Fig. 5

Clinical data of 3 patients with an extremely high mutation burden. (A1, A2) Prior to 1^st EGFR TKI treatment, P15 had small tumor foci with bone metastases. (A3) After progression, the ctDNA mutation profile showed an extremely high mutation burden and the corresponding tumor foci filled the entire right lung. (B1) Prior to 1^st EGFR TKI treatment, P16 showed small tumor foci in the left lung with bone metastases. (B2) After progression, the ctDNA mutation profile showed an extremely high mutation burden, and the corresponding tumor foci filled the entire left lung. (C1, C2) P20 with bone metastases had a smaller initial lesion. (C3) Larger tumor lesions at progression. (D1) The number of different ctDNA mutations in P20 at 5 time points. (D2) Carcinoembryonic antigen (CEA) was significantly increased during 1^st EGFR TKI treatment, and the change in CEA paralleled the number of point mutations.

Fig. 6

Mutation profile and clinical features between two baseline mutation patient groups. (A) Kaplan-Meier plot without progression between baseline EGFR drug-sensitive mutation patient (19del vs. L858R) groups. (B) Distribution of 1^st TKI clinical response between two baseline-mutated patient groups (* indicates P < 0.05 between the two groups). (C) Acquired resistant mutational pie chart for patient group with baseline EGFR p.19del mutation. (D) Acquired resistance mutation pie chart in the patient group with baseline EGFR p.L858R mutation. (E) Acquired EGFR p.T790M mutation ratio distribution between two patient groups with baseline EGFR drug-sensitive mutations (19del vs. L858R, ** indicates P < 0.01). (F) Distribution of 1^st TKI clinical response between EGFR p.T790M positive and negative patient groups at progression (* indicates P < 0.05 between the two groups).

Fig. 7

Analysis of the EGFR mutation profile between the EGFR p.T790M positive and negative patient groups and analysis of the relationship between the VAF of the EGFR mutations and tumor size. (A) A heat map of the VAF of the EGFR gene mutation in the 42 patients after drug resistance. (B) Linear relationship between the VAF of EGFR drug-sensitive mutations and VAF of EGFR p.T790M after resistance in 21 patients (r = 0.68, p = 0.00025). (C) The VAF of EGFR p.T790M versus tumor size in 18 patients after drug resistance (r = 0.22, P = 0.48). (D) The VAF of EGFR drug-sensitive mutations and VAF of EGFR p.T790M in 18 patients and tumor volume maps in resistance. The red line represents the VAF value of EGFR p.T790M in 18 patients. Dark green indicates the VAF of EGFR drug-sensitive mutations. Light green represents the tumor volume at drug resistance. The purple dotted line represents the estimated tumor volume. (E) Estimated linear relationship between the ratio of EGFR −/− tumor subclones and tumor size in 18 EGFR p.T790M-positive patients (y = 0.956x + 5.9465; r = 0.93, P = 2.2 * 10⁻⁸; The green dots represent the contributions from the tumor subclone EGFR −/−, and the black dots represent the contributions from the tumor subclone EGFR +/+).

Fig. 8

Dynamic analysis of ctDNA revealed spatial and temporal diversity of EGFR p.T790M mutations. Throughout the monitoring process, the red line represents the mutation trend of EGFR p.T790M, the green line represents the mutation trend of EGFR p.19del, the blue line represents the mutation trend of EGFR p.L858R, the light blue line represents the mutation trend of KRAS p.G12C, and the dark red line represents the mutation trend of EGFR p.A839T. The black dashed lines indicate the time points of drug resistance, the red stars indicate baseline drug-sensitive mutations as detected by the ARMs method. (A) The EGFR p.T790M mutation occurred in the sixth month after pathologically confirmed resistance in P50. (B) The EGFR p.T790M mutation occurred in the eighth month after pathologically confirmed resistance in P39. (C) The EGFR p.T790M mutation occurred in the sixth month after pathologically confirmed resistance in P60. (D) P5 received 1^st EGFR TKI treatment as an adjunctive treatment after surgery, and new metastatic lesions appeared in the left lung during drug resistance. Four months later, an EGFR p.T790M mutation was observed. (E) P20 showed that the EGFR p.T790M mutation occurred in the early treatment phase and was then inhibited and eventually disappeared in drug resistance. At the same time, P20 also demonstrated three mutation trends of the EGFR gene and the corresponding competitive evolution of tumor subclones. (F) P49 showed that EGFR p.T790M appeared 2 months prior to pathologically confirmed drug resistance, and the VAF value rose further when resistant.