A total of 100 patients were included in the analysis. The median age of all study participants was 68 years (range, 32-82). Seventy-four patients (74.0%) were male. The most common histologic subtype was SCLC (n = 90, 90%), followed by LCNEC (n = 6, 60%) and combined histology (n = 4, 4%). At baseline, 73 patients (73.0%) had ECOG PS of 0 or 1, whereas 25 (25.0%) and 2 (2.0%) had ECOG PS of 2 and 3, respectively. Additionally, 22 patients (22.0%) had brain metastases at presentation, including 8 patients (19.5%) in the atezolizumab group and 13 patients (22.0%) in the durvalumab group. Forty-one patients (41.0%) received atezolizumab and 59 patients (59.0%) received durvalumab; aside from histological subtype, baseline characteristics did not differ between the two subgroups (Supplementary Table 1). Patient characteristics are summarized in Table 1.
Efficacy of first-line chemoimmunotherapy
At data cutoff, the median follow-up was 14.4 months (95% CI, 10.6-21.4 months). Tumor response was evaluable in 93 patients (93.0%). The objective response rate (ORR) was 48.4% in the entire study population; the ORR was 58.5% in the atezolizumab group and 40.4% in the durvalumab group (p = 0.082). Seventy-two patients (72.0%) experienced disease progression or death, including 33 (80.5%) in the atezolizumab group and 39 patients (66.1%) in the durvalumab group. The median rwPFS was 7.2 months (95% CI, 6.7-12.0 months) in the entire study population (Figure 1A); rwPFS was similar between the atezolizumab group (median, 7.8 months; 95% CI, 6.4-14.9 months) and the durvalumab group (median, 7.2 months; 95% CI, 6.2-11.5 months; p = 0.18).
Fifty-two patients (52.0%) had died at the time of data cutoff, including 27 patients (65.9%) in the atezolizumab group and 25 patients (42.3%) in the durvalumab group. The median rwOS was 14.4 months (95% CI, 9.4-18.6 months) in the entire study population (Figure 1B); rwOS was similar between the atezolizumab group (median, 16.0 months; 95% CI, 11.3-31.4 months) and the durvalumab group (median, 12.7 months; 95% CI, 7.2 months - not reached; p = 0.20).
ECOG PS (³2 versus 0-1)
In the entire study cohort, rwOS was significantly shorter in patients with ECOG PS score ³2 (median, 6.6 months; 95% CI, 5.6 months-not reached) than in patients with an ECOG PS score <2 (median, 16.0 months; 95% CI, 12.7 months-not reached; Figure 2A). The HR for death was 2.57 (95% CI 1.43-4.60; p = 0.002).
Brain metastases
In the entire study cohort, rwOS was similar in patients with brain metastases (median, 7.3 months; 95% CI, 6.5-33.7 months) and patients without brain metastases (median, 14.6 months; 95% CI, 10.9-18.6 months; Figure 2B). The HR for death was 1.33 (95% CI, 0.71-2.47; p = 0.4).
Elderly patients (³70 versus <70)
In the entire study cohort, rwOS was similar in patients younger than 70 years (median, 11.5 months; 95% CI, 7.7-18.6 months) and patients 70 years or older (median, 14.9 months; 95% CI, 10.9 months-not reached; Figure 2C). The HR for death was 0.90 (95% CI, 0.51-1.59; p = 0.7).
Biomarker analysis
Baseline blood cell counts did not show any clear or consistent association with rwPFS or rwOS (Supplementary Table 2). However, white blood cell count, neutrophil count, NLR, and dNLR correlated with both rwPFS and rwOS at disease progression (padj < 0.05). Only the EPSILoN score was significantly associated with rwPFS but all four prognostic scores (RMH, GRIm, LIPI, EPSILoN) were significantly associated with rwOS (padj < 0.05; Supplementary Table 3).
Multivariable Cox Proportional Hazards Model
In the multivariate setting, including five variables (smoking status, BMI, LDH, NLR, and number of metastatic sites), the number of metastatic were associated with an increased HR for death (HR, 1.47; 95% CI, 1.11-1.94; p=0.007; Table 2). Interestingly, BMI was associated with a decreased HR for death (HR, 0.89; 95% CI, 0.81-0.97; p=0.008). Notably, there appears to be a sharp increase in the HR with more than three metastatic sites (Supplementary Figure 1). Due to the relatively small sample size, the variables show wide confidence intervals, most of which include zero.
Subsequent therapy
Forty-one patients (41.0%) received second-line treatment after progression on first-line chemoimmunotherapy. Out of 33 patients with available tumor response data, one patient (3.0%) had a partial response and four patients (12.1%) had stable disease. The median rwPFS2 reached 2.3 months (95% CI, 1.9 months-not reached).