Of 6,451 patients diagnosed with advanced NSCLC, 85 (12.2%) received NGS testing (Table 1). The median age at diagnosis was 61 years (range, 19–90 years), 45.2% of patients had a history of smoking, and 41.6% of patients were female. In this cohort, 95.8% of patients received EGFR testing and 95.3% received ALK testing (in <5% of the remaining samples, the testing status of both genes was unknown). There was a significant association between immunotherapy in the first four lines of treatment and NGS testing.
All survival analyses were performed using the Kaplan-Meier method. For the entire cohort, the median OS was 22.6 months and the 5-year survival rate was 22.89%. There was a significant difference in survival between patients who received targeted therapy and those who did not in the unadjusted survival curves (HR, 0.52 [95% CI, 0.49–0.56], log-rank P < 0.0001, Figure 1A) or among the 2177 matched pairs with well-matched characteristics and balance in the propensity score analysis (HR, 0.56 [95% CI, 0.52–0.60], log-rank P < 0.0001, Fig. 1B).
Survival of patients with NGS testing and routine testing
OS was compared between patients who underwent NGS testing and those who underwent routine testing. In the unadjusted survival curves, there was an absolute difference between the two cohorts (HR, 0.87 [95% CI, 0.78–0.96], log-rank P = 0.0097, Fig. 2A), especially in the 12- and 24-month mortality rates. In the propensity score analysis, there was no obvious OS difference between the two cohorts (760 matched patients, HR, 0.95 [95% CI, 0.82–1.11], log-rank P = 0.53, Fig. 2B). Patients who received NGS had a significantly higher 1-year survival rate than that of patients with routine testing (HR, 0.66 [95% CI, 0.53–0.83], log-rank P = 0.0004), as well as a higher rate of patients receiving targeted therapy (68.0% vs. 53.6%, P < 0.0001, Fig. 3A/B) and patients receiving off-label drugs (28.10% vs. 11.40%, P < 0.0001, Fig. 3C/D). The NGS testing cohort had a greater number of different types of off-label drugs.
Survival of patients with uncommon mutations
In NSCLC, common mutations include EGFR 19del and L858R mutations, and ALK and ROS1 fusion. Here, we defined mutations other than these common mutations as uncommon mutations. In patients with uncommon mutations, OS was compared between patients who received NGS testing and those who underwent routine testing. There was an absolute difference between the two cohorts (HR, 0.79 (95% CI, 0.69–0.91), log-rank P = 0.0038, Fig. 4A) in the unadjusted survival curves, especially in the 12-month mortality (HR, 0.58 [95% CI, 0.48–0.70], log-rank P < 0.0001). After using the propensity score analysis, the 12-month mortality between the two cohorts in 314 matched pairs was significantly different (HR, 0.72 [95% CI, 0.54–0.97], log-rank P = 0.03) (Supplementary Fig. 1), although the OS of these patients was not significantly different (HR, 0.86 [95% CI, 0.69–1.08], log-rank P = 0.19, Fig. 4B).
Next, the patients were divided into four groups: patients with uncommon mutations who received NGS, those who received routine testing, those who received targeted therapies, and those who did not receive targeted therapies. The results of unadjusted survival curves showed that the median OS was longer in the two groups with targeted therapies, especially in the groups with NGS testing (Supplementary Fig. 2). Among uncommon mutation patients with targeted therapies, propensity score analysis showed that the 12- and 24-month mortality was significantly higher in the NGS testing group than in the routine testing group (Supplementary Fig. 2). This suggested that NGS was a better detection method for patients with uncommon mutations. This could be associated with the fact that the rate of off-label treatments was significantly higher in the NGS group than in the routine testing group (45.4% vs. 9.6%, P < 0.0001, Supplementary Fig. 3), and indicated that NGS testing resulted in a more objective and precise clinical therapy for uncommon mutation patients.
All patients received first-line treatment, and 3,517 (54.52%) and 1,326 (20.56%) received second- and third-line treatments, respectively. Patients receiving NGS testing were more likely to receive targeted therapy (51.7%) as first-line treatment than routine testing patients (34.6%, P < 0.0001), whereas the opposite result was obtained for second-line treatment (19.8% for NGS testing vs. 25.1% for routine testing, P = 0.012, Table 2) and for third-line treatment (12.9% vs. 15.4%, P = 0.333). The rate of chemotherapy ± anti-VEGF treatment as first-line treatment was lower in the NGS testing group than in the routine testing group (39.5% vs. 62.5%, P < 0.0001), whereas it was similar between the two cohorts in the second- and third-line treatments. In the first-, second- and third-line therapy groups, patients receiving NGS testing were more likely to receive immunotherapy (1st-line: 3.6% vs. 1.0%, 2nd-line: 8.3% vs. 2.8%, 3rd-line: 4.6% vs.1.8%) or participate in clinical trials (1st-line: 5.2% vs. 1.9%, 2nd-line: 5.7% vs. 3.0%, 3rd-line: 2.5% vs.1.7%) than those receiving routine testing (Table 2). Among patients without EGFR mutations or ALK fusions, the percentage receiving targeted therapies, immunotherapy, and clinical trials was higher in the NGS testing group than in the routine testing group. Correspondingly, the percentage receiving chemotherapy ± anti-VEGF treatment showed the opposite trend (Table 3).