Patients and clinical characteristics
A total of 117 patients were retrospectively analyzed in this retrospective study. Patients received first-generation EGFR-TKI or plus chemotherapy. Among them, 50 patients in combination group (T+C) received chemotherapy plus first-generation EGFR-TKI, while the other 67 patients in the monotherapy group (T) received EGFR-TKI alone. The patients’ clinical characteristics were summarized in Table 1, including age, sex, histology, ECOG PS, smoking history, EGFR mutation, brain metastasis, and stage. There was no significant difference in clinical characteristics between the two groups.
After two cycles / two months of treatment, the response rate was evaluated. Of the 50 patients in the combination group, 1 achieved complete response (CR), 38 achieved partial response (PR), 10 achieved stable disease (SD), 1 achieved PD, resulting in an ORR of 78.00% and DCR of 98.00%. Of the 67 patients in the EGFR-TKI monotherapy group, 43 achieved PR, 22 achieved SD, 2 achieved PD, and no one achieved CR, resulting in an ORR of 64.18% and DCR of 97.01% (Table S1). As shown in Table S1, the ORR was slightly higher in the combination group than in the EGFR-TKI monotherapy group, but there was no statistically significant difference (p = 0.108).
As of March 2020, 79 patients (67.52%) had reached the endpoint of disease progression or death, and the median follow-up time was 26.27 months. The median PFS (mPFS) was 19.00 months (95% CI, 14.674-23.326) in the combination group and 11.70 months (95% CI, 10.807-12.593) in the EGFR-TKI monotherapy group, and the difference was statistically significant (p = 0.000) (Fig 1a).
The median OS (mOS) was not reached in the combination group and no difference in OS was identified at the time of this analysis (NA vs. 38.50 months, p = 0.586, Fig 1b). 1-year OS rate was 94.87% (37/39) in the combination group and 96.82% (61/63) in the TKI monotherapy group (p = 0.562). 2-year OS rate was 92.86% (13/14) in the combination group and 81.25% (39/48) in the TKI monotherapy group (p = 0.303) (Fig S1). As the overall survival data was not sufficient enough, further analysis was not performed.
EGFR mutation site analysis
In all patients, approximately 57.26% (n=67) had an exon 19 deletion (19 del), while 37.61% (n=44) had an exon 21 L858R mutation (21 L858R) in the EGFR gene. For the patients with EGFR exon 19 deletion, the mPFS was 20.93 months (95% CI, 7.927-33.933) in the combination group and 11.87 months (95% CI, 10.424-13.316) in the EGFR-TKI monotherapy group (p = 0.004) (Fig 2a). The mPFS for patients harboring L858R point mutation was 18.07 months (95% CI, 13.212-22.928) in the combination group and 11.17 months (95% CI, 9.992-12.3482) in the EGFR-TKI monotherapy group (p = 0.021) (Fig 2b).
EGFR mutation abundance analysis
In this study, abundance data was available in 51 patients, including 23 were in the combination group and 28 were in the EGFR-TKI monotherapy group. We explored the relationship between the EGFR mutation abundance and the efficacy of EGFR TKI combination with or without chemotherapy. The cutoff value of mutation abundance was set as 4.9% for exon 19 deletion and 9.5% for exon 21 L858R . The cutoff value of ctDNA abundance from plasma was set as 2% for exon 19 deletion and 5% for exon 21 L858R . Of all 51 patients, 39 patients harbored high abundance EGFR mutation and 12 had low abundance mutation. Among patients with high EGFR mutation abundance, the mPFS was 19.00 (95% CI, 15.3442-22.656) months in the combination group and 10.93 (95% CI, 9.687-12.179) months in the EGFR-TKI monotherapy group (p = 0.008) (Fig 2c). However, among patients with low EGFR mutation abundance, mPFS was 11.83 months (95% CI, 9.013-14.647) vs. 10.57 months (95% CI, 5.181-15.959), and the difference was not significant (p = 0.541)(Fig 2d).
Subgroup analysis was conducted to screen out the intended population. The Cox regression model was used to calculate hazard ratios. Fig 3 showed that the results of the subgroup analysis were basically consistent with the above-mentioned result. Most patients may obtain clinical benefits from the regimen of TKI combined with chemotherapy. But for patients with brain metastasis, T+C did not show a significant advantage and the risk of progression was 1.2 times higher in the combination group than that in the TKI monotherapy group (p =0.686; HR = 1.2; 95% CI, 0.3-1.8). Significant difference was also not deserved in subgroups of none or single EGFR co-mutation and multiple (≥2) co-mutation (p = 0.276, HR = 0.39, 95% CI (0.07-2.1) in none or single co-mutation subgroup and p = 0.154, HR = 0.49, 95% CI, (0.18-1.3) in multiple co-mutation subgroup).
Post-progression detected by NGS
Considering the impact of drug resistance on treatment, re-biopsy, and NGS-based testing were made. T790M was detected in 59.09% (26/44) of patients in the TKI monotherapy group and 57.14% (8/14) in the combination group, and the results suggested that there was no statistical difference in the frequency of T790M mutation (p = 0.898, Fig 4c). Acquired resistance also involved Her2 amplification, Met amplification, ALK fusion, and Myc amplification (Fig 4a,b). Of the 117 patients, 34 patients obtained T790M mutation after progression, and 27 patients received third-generation EGFR-TKIs, of which 7 were in the combination group, and 20 were in the EGFR-TKI monotherapy group.
The details about the AEs were shown in Table 2. Skin rash was the most common (64.00% in the combination group and 70.15% in the EGFR-TKI monotherapy group) (p = 0.804), followed by elevated liver enzymes (62.00% in the combination group and 50.74% in the EGFR-TKI monotherapy group, p = 0.228). The other AEs observed in the TKI monotherapy group included diarrhea (35.82%), mucositis (13,43%), constipation (10.45%), and nausea/vomiting (10.45%). Meanwhile, hematologic toxicities were more common in the combination group, such as leukopenia/neutropenia (44.00% vs. 7.46%, p =0.000), anemia (38.00% vs. 8.96%, p =0.000), thrombocytopenia (34.00% vs. 8.96%, p =0.001). The patients in the combination group were more likely to develop AEs, most of which showed the severity of grade 1 or 2. No drug-related interstitial lung disease or deaths were observed.