Patients’ characteristics
In this study, 39 patients were enrolled and 34 patients were included in the retrospective analysis. The enrollment process is shown in Figure 1. Two patients were excluded, due to one 19del false positive and one T790M mutation at baseline. Thirty-seven patients were successfully tested for plasma with a success rate of 100% (37/37), while three patients failed to perform tissue re-biopsy due to few tumor cells by pathological examination. The characteristics of initial and secondary biopsy of enrolled patients were summarized in Table 1 and 2. Thirty-four patients’ clinical characteristics for retrospective analysis were shown in Table 3. Pathology revealed only one adenosquamous cell carcinoma, the others were adenocarcinoma (97.06%, 33/34). And the common EGFR activating alterations were largely represented (67.65%, 23/34, 19del; 32.35%, 11/34, 21L858R). During first-line treatment, median PFS resulted 13 months. After progression, 70.59% (24/34) of patients received third-generation EGFR-TKI treatment, and median PFS resulted 6 months. The first patient started third-generation EGFR-TKI treatment on October 24, 2017 and the last one assumed the first dose on August 17, 2018. The data cut-off for this analysis was April 25, 2019.
Genetic alterations in ctDNA and tissue of enrolled patients
Of the 37 patients who had plasma sent for ctDNA NGS, 34 (91.89%) also had tissue sent for solid tumor NGS. Out of 34 patients with samples sent for both ctDNA and tissue NGS, 34 had shared at least 1 alteration identified by both tissue NGS and ctDNA analysis. In Figure 2A, the most frequent alterations detected by ctDNA NGS are displayed. The three most frequent ctDNA alterations involved the following genes: TP53 (67.57%, 25/37), followed by KRAS (8.11%, 3/37) and amplification of c-Met (5.41%, 2/37) (Figure 2B). The most frequent alterations detected by tissue NGS involved the following genes: TP53 (52.94%, 18/34), amplification of c-Met (11.76%, 4/34) and PIK3CA (11.76%, 4/34) (Figure 2C&D). In total, 83.78% (31/37) of patients harbored concomitant mutations, 70.27% (26/37) by ctDNA and 70.59% (24/34) by tissue NGS. Patients with a history of smoking (85.71% [6/7] vs 66.67% [18/27]) was found in tissue with a higher incidence of concomitant mutations, but not in ctDNA (37.5% [3/8] vs 79.31% [23/29]).
A total of 23 cases of EGFR T790M mutation in plasmas and tissues were detected after first-line EGFR-TKI treatment. The total positive rate was 62.16% (23/37). The positive rate of tissue samples to detect EGFR T790M mutation was 64.71% (22/34), and the positive rate of plasma samples to detect EGFR T790M mutation was 51.35% (19/37). One of the patients had a positive mutation in plasma, while the tissue was negative. This patient was treated with Osimertinib in the follow-up treatment. The best efficacy was evaluated as PR, and PFS was 7.0 months. Therefore, we combined the efficacy and test results to consider the tissue test results as false negative, suggesting that tissue and ctDNA assay provided complementary results. The remaining 18 patients with positive blood tests were consistent with those having positive tissue tests (Figure 3A). Therefore, the consistency rate between tissue and plasma for EGFR T790M mutation was 78.26% (18/23). In addition, EGFR T790M mutation in plasma samples were tested and compared with ddPCR and NGS as illustrated by the Venn diagrams. Tissues were detected and compared by ARMS-PCR, Cobas® ARMS-PCR and NGS (Figure 3B&C). The consistency rate in plasma and tissue for different platforms were 84.21% (16/19) and 78.26% (18/23), respectively.
Treatment outcome
Of the 34 patients in the retrospective analysis, 24 received third-generation EGFR-TKI treatment and 10 received others after progression with first-line EGFR-TKI treatment. Twenty-four (70.59%, 24/34) of patients exhibited prominent tumor shrinkage during treatment. Among them, 22 received third-generation EGFR-TKI treatment, revealing the strength of third-generation EGFR-TKIs and radio/chemotherapy for patients with resistance to first-line treatment (Figure 4). Thirteen partial responses were observed and all occurred in T790M-positive patients (contained one who was detected only by ctDNA assay and one was not detected by the tissue NGS but was detected on other platforms), the objective response rate (ORR) was 54.17% as shown in Table 4 (p=0.003). Comparing with third-generation EGFR-TKI treatment, radio/chemotherapy(others) significantly showed shorter PFS(p<0.001, median survival, 4.0 months, ratio, 0.25[95% CI, 0.19-0.85 months] vs 10.0 months, ratio, 4[95%CI, 1.17-5.34]; HR, 4.49[95%CI, 1.36-14.76])and OS (p=0.058, HR, 4.72[95%CI, 0.61-36.57]). Similarly, T790M-negative patients significantly showed shorter PFS than patients with T790M mutation (p=0.010, median survival, 5.0 months, ratio, 0.5[95% CI, 0.23-1.08] vs 10.0 months, ratio, 2.5[95%CI, 0.92-4.33]; HR, 2.37[95%CI, 0.92-6.09]), while OS was not statistically significant (Table 4). There was significant difference in the response rate between the T790M-positive and T790M-negative patients (57.14% vs 7.69%, respectively; p= 0.004, Chi-squared test). Furthermore, radio/chemotherapy (others) was significantly associated with poor PFS and OS in the multivariate analysis (p=0.073 and p=0.049; age, sex, history of smoking, type of treatment, EGFR status, EGFR T790M mutation, and concomitant mutations were entered into the multivariable Cox proportional hazards regression model) (Table 4). Interestingly, concomitant genetic alterations were significantly associated with a poor PFS for patients receiving third-generation EGFR-TKIs with T790M mutation as shown in Figure 5 (p=0.0374, median survival, 13.0 months, ratio,1.37 [95% CI, 0.39-4.76] vs 9.5 months, ratio, 0.73[95%CI, 0.21-2.54]; HR, 0.33 [95%CI, 0.12-0.87]).
Resistance mechanisms to third-generation EGFR-TKI treatment
After third-generation EGFR-TKI treatment, 22 patients presented PD and two patients presented SD. The analysis of putative mechanisms of resistance in these patients showed that the activation of known by-pass signaling pathways was observed in seven EGFR T790M-positive patients with PD, including three with PIK3CA mutation, two with MET amplification, one with PTEN deletion and one with STK11 mutation, excepting one with KRAS activating mutation in EGFR T790M-negative patients. Nine progressive patients presented other concomitant genetic alterations, including TP53, RB1, NOTCH1, FANCA, CTNNB1, and BRCA1. Notably, both patients without disease progression had co-mutations in the TP53. Moreover, in five patients (30%) (#1, #3, #6, #7, #40), no mechanisms of resistance to third-generation EGFR-TKIs were detected (Table 5).