Tumor biopsy specimens from 59 patients with advanced NSCLC with EGFR mutations were tested, and these patients had previously experienced pd on 1g/2g TK. The median TTP for EGFRT790m.1g/2g was detected in 38 (63%) patients at 10.3 months (range 1.3-75.8 months), and the t790m group and the t790m group were 7.4 months and 13.6 months (hr, 1.64; 95%CI, 0.95-2.83; p=0.07; Supplementary Figure s5). Between the t790m and t790m groups,

There was no difference in previous treatment, including EGFR tki (65.8% vs. 61.9%, p=0.76), chemotherapy used previously (76% vs. 66%, p=0.75, wilcoxon test), or duration of chemotherapy exposure (mean 215 days vs. 175 days, p=0.75). Consistent with previous reports, histological transformation under microscopy was rare, with one case of squamous transformation (a450) and another case of mixed adenocarcinoma small cell histology (a092) observed both at baseline and after resistance.

Genome maps of t790m and t790m? Disease status

We first evaluated the differences in the prevalence of somatic genomic alterations in our cohort (except EGFR    T790M) and the early treatment cohort (Wes data) in 38 patients with advanced EGFR mutations. Tp53    SNV alterations were more common in TKI resistant population (63%) than in the early treatment population (45%). In addition, we detected that genomic amplification rates of tert and three genes encoded in 14q22 (PTGDR, Sav1 and Sos2) in the treatment-resistant tumor compared with the early treatment tumor. Next,

By comparing T790M and T790M, we sought to identify previously reported and identify potential novel treatments that can handle drug resistance mechanisms. 5 (8%) patients showed mett alterations, mainly occurring in the T790M patient cohort. In one patient (A056), a splice site deletion that caused the 14 jump of Met exon 14 (metex14), and the transcriptional deletion at exon 14 was detected by RNASEQ data. This patient also exhibited chromosomal arm 7p deletion (deletion of EGFR mutation) and 12q amplification (including mdm2 and cdk4,

Typically amplified with metex14). Similarly, we did not detect primary exon 19 deletion/l858r mutation in another T790m? patient (a058), indicating that loss of activated EGFR mutation is a truly rare mechanism for acquired resistance. Pik3ca mutations (n=10,17%) and Her2 amplifications (n=4,7%) are common, but there is no difference based on T790m status (p=0.69 and p=0.54, respectively), although the frequency of PIK3ca changes in drug-resistant tumors (17%) is generally higher than that in advanced treatment without EGFR mutations

Patients with tumors (5%, p=0.09). In the PIK3CA mutations, it is worth noting that most are cloned (n=5/7), and annotated by clinvar as potentially pathogenic (n=6/7). In summary, this suggests that the role of PI3K signaling in mediating tki resistance is independent of the T790m state. Given the potential role of keap1 and NFE2L2 co-mutations in the production of therapeutic resistance. However, we did not detect any NFE2L2 mutations in the cohort, and we only detected one keap1 mutation in the T790m tumor (a449).

Tp53 changes are early clonal events found in about half of EGFR-mutated tumors. Compared with t790m patients, tp5    3 gene mutations in t790m patients were significantly richer (86% vs. 5    0%, p=0.0    1), and most of them (n=36/37, 97%) were early clonal events (n=3   6/37, 97%), which were significant compared with t790m patients (p=0.0    1). In addition, mdm2 and yeats4 amplifications were at t799 More common in 0m patients and is mutually exclusive to tp53 mutations (p=0.004). These two genes are located at the same chromosomal location and negatively regulate tp53. As we and others have previously reported, the more mutations of known cancer driver genes, the shorter the ttp in multivariate analysis (hr, 1.41; 95%CI, 1.09-1.81; p=0.008; Supplementary Figure S8a). In particular, the concurrent RB1/TP53 alteration (n=5) is closely related to TTP without t790m status, consistent with a recent report.

Copy number analysis found that WGD is common, comparable to the ratio observed in early treatment of EGFR-mutant tumors (88% vs. 89%). However, in 7 (12%) patients without WGD, all were t790m (embedded image test, p=0.04) and there was a deletion of EGFR exon 19, indicating that the deletion of WGD predicts the emergence of t790m drug-resistant diseases. At t790m and t790m? There was no difference between tumors (gii; medians were 52% and 55%, respectively), and no co-mutation of tp53 was found to be associated with increased gii. The tumor mutation burden in patients with t790m was higher than that of patients with t790m (median tmb 2.36       1.66 mutations/mb, p=0.02, Figure 1c; Supplementary Figure 1c, S9a), which may be due to the large number of smokers in the t790m cohort.

Next, we examined the recurring focus amplification and deletion events associated with the T790m status of patients 3q23 (including PIK3CB, MRAs and Foxl2 genes) and 14q21 (including FOXA1 and NKX2-1) of patients t790m status revealed recurring focus amplification and deletion events associated with amplification (Supplementary Figure S10). It is worth noting that the 3q    arm amplification was higher in the t790m group (57%) than in the t790m group (13%) and was the only statistically significant arm-level event (Figure 1e, embedded    image-test    p≤0.001). We confirmed the chromosome 3q gene (including sox 2) There was higher expression in patients with 3q arm increase (Supplementary Fig. s11c). To understand the extent to which these chromosome levels may be obtained after TK treatment, we again compared with the pre-treatment cohort. In the treatment-initial cohort, 3q increase (Supplementary Fig. s3) tumors occurred in 7 of 38 cases (18%). This result was significantly lower than t790m? (p=0.003), but similar to those of t790m? patients (p=0.53), suggesting that 3q amplification may be obtained specifically from t790m? tumors. Interestingly, chromosome 3q contains the squamous cell line transcription factors Tp63 and Sox2, which are the main characteristics of lung squamous cell carcinoma but have not been associated with lung adenocarcinoma.

Next, we studied mutation processes related to EGFR     Tki resistance. Recognized mutation characteristics were found in lung adenocarcinoma, namely aging, apobec, DNA double-strand break repair, smoking and DNA mismatch repair. Compared with t790m?, EGFR    T790m drug-resistant tumors have a higher relative contribution rate to aging signals, which is the highest of a[c>t]g in the context of EGFR    T790m point mutation. Probability nucleotide changes were consistent (Supplementary Figures 3S12 and S13). In contrast, the proportion of non-aging features (smoking, apolipoprotein bec and DNA repair) in t790m compared with t790m tumors was significantly higher than that in t790m tumors, suggesting another mutational process driving t790m resistance. Overall, this suggests that t790m resistance may be unlikely to occur in tumors with a larger proportion of active non-aging mutational features (e.g. smoking/apobec characteristics).

We used the Wes data to estimate the proportion of cancer cells (purity) of each tumor sample and used the transcriptome deconvolution method Tumera to estimate and compare gene expression in cancer and interstitial (nonmalignant) cells in t790m and t790m tumors. It is worth noting that we speculate that in cancer cells of t790m tumors, the expression of lung adenocarcinoma markers such as napsa, nkx2-1, sfta2 and sfta3 is common and almost completely lost. In addition, we observed an increased expression of histological markers of squamous cell carcinoma or neuroendocrine carcinoma in a small fraction of t790m tumors (n=4, 27%. These findings were orthogonal validation of multiple immunofluorescence, It was confirmed that the expression of napsa and nkx2-1 (ttf-1) in t790m cancer cells was reduced compared to t790m tumors (Figure 2D; Supplementary Figure S14). It is worth noting that analysis of three untreated NSCLC adenocarcinoma cohorts showed that the expression of low adenocarcinoma marker genes (Napsa and nkx2-1) was very rare and was only observed in EGFR wild-type tumors (Figure 2F). In summary, these data highlight the previously underestimated acquired lineage plasticity after TK resistance, especially in t790m tumors, coexisting with 3q amplification and non-aging mutation signature processes, potentially promoting epidermal growth factor receptor-independent signaling mechanisms.

Given the lack of efficacy of checkpoint inhibitors in EGFR-mutated NSCLC, we sought to describe the immune environment associated with EGFR     TKI resistance, initially stratifying tumors based on the “T cell inflammatory gene expression profile” (Gep) characteristics.

We then used a published calculation method (teder; ref) to further elucidate the subpopulation of infiltrated immune cells associated with TK resistance. This showed that the hypothetical levels of mdscs in immune t790m? were higher than those in immune t790m? (p=0.04, t test) and lower levels of tmm2 (p=0.003, t test). The expression of pd-l1, foxp3 and ido in immune t790m? was also significantly higher than that of immune t790m? (Figure 3B, pd-l1 and toxp 3 Multiple immunofluorescence staining see Figures s15b and s15c). Next, we investigated whether the immunophenotype during drug resistance was related to the previous duration of 1g/2g    TKI. Interestingly, the total TTP of the immune T790m tumor was the shortest (Figure 3c), with half (5/10 patients) having a total TTP of less than 3 months. In contrast, the overall survival of the immune T790m tumor was the longest (median TTP20.6 months; range 8.2 months to 76.8 months), compared with the immunorefrigerated T790m tumor (median TTP4 .January; Range 1.3-December; Heart rate 11.78; p=0.004; 95%CI    3.01-46.2 months; p=0.001; Figure 3c). Meta analysis highlighted the lack of efficacy of single-dose immune checkpoint inhibitors in non-small cell lung cancer with EGFR mutation (59 cases), 7/8 (88%, 4/8        hot, 2/8         cold, 2/8      unknown) patients, which is consistent with meta-analysis (Supplementary Table s5). However, a patient with immune T790m (a096 ) was treated with a combination of nivolumab-ipilimumab immune checkpoint inhibitor (60 cases) in clinical trials and obtained a stable condition of 8.9 months. Taken together, our data suggest a potential role of inflammatory chemokine, for example, CXCL9-probably driven by MDSSCS- plays a role in mediating T790M?tki resistance. Furthermore, our data highlight the significant heterogeneity of TME components in Gep "hot" tumors, indicating the need to inquire in more detail to delineate specific immune targets.

Although third-generation EGFR tkis are increasingly being adopted in front-line settings, this clinical practice is partly due to the inability to predict resistance trajectories in individual patients. After identifying new molecular features (3q amplification, transcriptional subtype, loss of adenocarcinoma lineage marker and inflammation) associated with different EGFR tki resistance statuses, we attempted to establish a model to predict the emergence of t790m. We speculate that these genomes, chromosomal levels and transcriptional characteristics may exist at baseline levels and may also represent changes obtained during treatment (Figure 4a). To further explore this, we identified three trunk features that may exist before the onset of 1g/2g tki treatment: EGFR exon 19 deletion, WGD deletion

Loss and TP53 changes (Figure 4a). Using the Bayesian approach, individual patients can be divided into different groups according to their pretreatment molecular genotype, and the odds of EGFR t790m resistance are very different (Figure 4b). For example, in non-WGD tumors, the probability of t790m resistance is between 87.2% and 97.9%, which may mean that sequential treatment of first-generation/second-generation to third-generation EGFR tki may be a viable clinical strategy for these patients (11% of patients in our cohort). The predictive power of these characteristics needs to be further validated in a larger cohort. However, these results illustrate how data-driven treatment algorithms are derived from real-world evidence and may help define the best sorting strategy for individual patients.

Our study has conducted a comprehensive and comprehensive analysis of the genomic and transcriptional maps of EGFR    TKI resistance for the first time. It is worth noting that our data show that genomic plasticity has been underestimated to some extent so far. Although there are reports of organized transformation in 1% to 3% of patients after tKI resistance, we found that in t790m? tumors, adenocarcinoma markers (Napsin-A and TTF-1) are generally lost, while non-tru subtypes (PI and PP) are significantly enriched. Although there is a lack of paired base

Line samples are a limitation of our study, but comparisons with treatment of non-small cell lung cancers with pure EGFR mutations show that loss of adenocarcinoma lineage markers, especially in T790M? disease, may represent early dedifferentiation events caused by chronic EGFR. TKI exposure. More significant genomic changes in T790M? disease include TP53 mutation (86% vs 50%), 3q amplification (57% vs 13%) and Met alterations (19% vs 3%), which further lead to T790M? plasticity and drug resistance.

Of particular interest in clinical interest is the immunothermal subgroup in the T790m? cohort, which represents a significantly shorter TTP in a group of patients, characterized by high Gep scores, overexpression of PD-L1, and an immunosuppressive microenvironment rich in chemokine. Consistent with our findings, retrospective analysis shows the relationship between high expression of PD-L1 and low response rates and PFS, suggesting that "inflammatory" TME mediates primary resistance to EGFR    TKI. Recently, inhibition of EGFR signaling has been found to deplete Treg and increase ifnγ

Signaling, supporting the link between “inflammatory” TMEs, is believed to be an adaptive change that may weaken the response to targeted therapy. Our data further suggest that inflammatory TMEs may occur in primary or secondary resistance and are variably composed of CD8T cells (tumor antigen-specific and/or bystander), Treg and MDSC. Finally, high expression of Iodo    1 was observed, especially in T790M? immunothermal tumors, accompanied by overexpression of kynurenine (kynu; Figure 2B), suggesting that the Iodo pathway is in dimensional

Treg activation and play a role in the immunosuppressive environment in tumor subpopulation. Recently, through single-cell RNA-seq for a range of oncogene-driven non-small cell lung cancer tumors, Maynard and colleagues have also emphasized the importance of IO pathway, immune microenvironment and alveolar regenerating cell characteristics in targeted therapy. Our data extend these observations to illustrate that treatment-induced adaptive cell status may be affected by genomic alterations and suggests EGFR-dependent and lineage plasticity in cancer cells, immune cell populations and chemokine

There are complex interactions between sexes. To better elucidate the role of these immune mediators, prospective studies are underway. Immune checkpoint inhibition, including combined anti-PD-1 and anti-CTLA-4 treatments, showed a significant lack of efficacy in clinical trials in TK-resistant settings. In addition to efforts to evaluate immunosuppressive targets of the adenosine axis, such as adenosine 2A receptor (A2AR), CD39 and CD73, reasonable targets in future clinical trials may include IODO, MDSSC depletion strategies such as bevacizumab or selective inhibition of PI3K-γ.
Chapter completed!