Literature Review Results
Through all databases searches, 317 articles were identifified (Fig. 2). In the first round analysis, 102 abstracts repeated in different databases were excluded. All remaining abstracts were reviewed, and 4 randomized, controlled trials involving 2391 patients (CheckMate 017, CheckMate 057, CheckMate 078 and KEYNOTE-010 trials) fulfilled eligibility criteria and study quality requirements. These trials were randomized and followed intention-to-treat analysis for the primary endpoint (overall survival) and secondary endpoints (progression-free survival, overall response rate). All trials were centrally randomized.
Study Design Of Included Trials
All four trials involved adult patients with advanced NSCLC who had progression during or after first-line chemotherapy (platinum-based doublet chemotherapy mainly). In addtion, the CheckMate 017 trial only involved advanced Squamous NSCLC, the CheckMate 057 trial only involved advanced nonsquamous NSCLC, the CheckMate 078 trial involved Chinese population of patients predominantly, and the KEYNOTE-010 trial involved advanced NSCLC with PD-L1 expression on at least 1% of tumour cells. All selected trials were superiority trials comparing treatment with nivolumab or pembrolizumab versus docetaxel (Table 1). In the CheckMate 017, CheckMate 057 and CheckMate 078 trials, patients received nivolumab at 3 mg/kg every 2 weeks or docetaxel at 75 mg/m2 every 3 weeks. In the KEYNOTE-010 trial, patients received pembrolizumab at 2 mg/kg or 10 mg/kg every 3 weeks or docetaxel at 75 mg/m2 every 3 weeks. Among 2391 patients, 765 received nivolumab, 690 patients received pembrolizumab, and 936 patients were treated with docetaxel. The treatment continued until disease progression, intolerable toxic effects, physician decision, patient withdrawal, or other reasons.
Table 1
Main characteristics of the four studies included in the meta-analysis
| CheckMate 017 | CheckMate 057 | CheckMate 078 | KEYNOTE-010 |
Year of publication | 2015 | 2015 | 2019 | 2016 |
Object of study | Nivolumab vs. Docetaxel | Nivolumab vs. Docetaxel | Nivolumab vs. Docetaxel | Pembrolizumab vs. Docetaxel |
Type of study | Prospective phase III randomized tria | Prospective phase III randomized tria | Prospective phase III randomized tria | Prospective phase II/III randomized tria |
Primary endpoint | Overall survival | Overall survival | Overall survival | Overall survival |
Patients enrolled | Advanced Squamous NSCLC who had progression during or after first-line chemotherapy | Advanced Nonsquamous NSCLC who had progression during or within platinum-based doublet chemotherapy | Advanced NSCLC who had progression during or after platinum-based doublet chemotherapy | Advanced NSCLC who had progression after platinum-based chemotherapy, with PD-L1 expression on at least 1% of tumour cells |
Number of patients | 272 | 582 | 504 | 1033 |
Treatment arm | 3 mg/kg every 2 weeks | 3 mg/kg every 2 weeks | 3 mg/kg every 2 weeks | 2 mg/kg or 10 mg/kg every 3 weeks |
Control arm (Docetaxel) | 75 mg/m2 every 3 weeks | 75 mg/m2 every 3 weeks | 75 mg/m2 every 3 weeks | 75 mg/m2 every 3 weeks |
Assessment Of Risk Of Bias
Risk of bias analysis (Fig. 3) showed that two studies had high risk of blinding of participants and personnel bias, the other two trials were rated with low risk of bias.
Patients
Most demographic and baseline characteristics were similar between the trials (Table 2). The CheckMate 078 trial mostly involved Asian patients, whereas the other three trials involved White, Asian and Black patients. The majority of patients in the four trials were Eastern Cooperative Oncology Group (ECOG) performance status (PS) 1. In the treatment arm, the CheckMate 078 trial excluded all the patients with EGFR or ALK mutations, however, the proportion of EGFR and ALK mutation were unkonwn in CheckMate 017 trial, 15% and 4% in CheckMate 057 trial, 9% and 1% in KEYNOTE-010 trial. All the patients had at least 1 line of prior treatment.
Table 2
Baseline Caracteristics of the studies included
| CheckMate 017 | CheckMate 057 | CheckMate 078 | KEYNOTE-010 |
| Nivolumab (n = 135) | Docetaxel (n = 137) | Nivolumab (n = 292) | Docetaxel (n = 290) | Nivolumab (n = 338) | Docetaxel (n = 166) | Pembrolizumab (n = 690) | Docetaxel (n = 343) |
Average age (year) | 62 | 64 | 61 | 64 | 60 | 60 | 63 | 62 |
Range | 39–85 | 42–84 | 37–84 | 21–85 | 27–78 | 38–78 | 56–69 | 56–69 |
Sex |
Man | 111(82%) | 97(71%) | 151(52%) | 168(58%) | 263(78%) | 134(81%) | 426(62%) | 209(61%) |
Woman | 24(18%) | 40(29%) | 141(48%) | 122(42%) | 75(22%) | 32(19%) | 264(38%) | 134(39%) |
ECOG PS | | |
0 | 27(20%) | 37(27%) | 84(29%) | 95(33%) | 47(14%) | 21(13%) | 232(34%) | 116(34%) |
1 | 106(79%) | 100(73%) | 208(71%) | 194(67%) | 291(86%) | 144(87%) | 454(66%) | 224(65%) |
Race |
White | 122(90%) | 130(95%) | 267(91%) | 266(92%) | 30(9%) | 15(9%) | 496(72%) | 251 (73%) |
Asian | 4(3%) | 2(1%) | 9(3%) | 8(3%) | 308(91%) | 151(91%) | 145(21%) | 72 (21%) |
Black | 1(1%) | 2(1%) | 7(2%) | 9(3%) | 0(0%) | 0(0%) | 21(3%) | 7 (2%) |
Nonspecifified | 2(1%) | 1(1%) | 9(3%) | 7(2%) | 0(0%) | 0(0%) | 10(1%) | 2 (1%) |
Disease stage |
IIIB | 29(21%) | 24(18%) | 20(7%) | 24(8%) | NA | NA | NA | NA |
IV | 105(78%) | 112(82%) | 272(93%) | 266(92%) | NA | NA | NA | NA |
Smoking status |
Former or current | 121(90%) | 129(94%) | 231(79%) | 227(78%) | 236(70%) | 118(71%) | 564(82%) | 269 (78%) |
Never | 10(7%) | 7(5%) | 58(20%) | 60(21%) | 102(30%) | 48(29%) | 123(18%) | 67 (20%) |
Unknown | 4(3%) | 1(1%) | 3(1%) | 3(1%) | 0(0%) | 0(0%) | 3(༜1%) | 7 (2%) |
Tumor histology |
Squamous | 135(100%) | 137(100%) | 0(0%) | 0(0%) | 133(39%) | 67(40%) | 156(23%) | 66 (19%) |
Non-squamous | 0(0%) | 0(0%) | 292(100%) | 290(100%) | 205(61%) | 99(60%) | 484(70%) | 240 (70%) |
Other | 0(0%) | 0(0%) | 0(0%) | 0(0%) | 0(0%) | 0(0%) | 15(2%) | 10 (3%) |
EGFR status |
Wild-type | NA | NA | 168 (58%) | 172 (59%) | 338(100%) | 166(100%) | 581(84%) | 294 (86%) |
Mutant | NA | NA | 44 (15%) | 38 (13%) | 0(0%) | 0(0%) | 60(9%) | 26 (8%) |
Unknown | NA | NA | 80 (27%) | 80 (28%) | 0(0%) | 0(0%) | 49(7%) | 23 (7%) |
ALK translocation |
No | NA | NA | 113 (39%) | 130 (45%) | 338(100%) | 166(100%) | 612(89%) | 310 (90%) |
Yes | NA | NA | 13 (4%) | 8 (3%) | 0(0%) | 0(0%) | 6(1%) | 2 (1%) |
Unkown | NA | NA | 166 (57%) | 152 (52%) | 0(0%) | 0(0%) | 72(10%) | 31 (9%) |
Efficacy
1 patient (1/135), 4 patients (4/292) and 1 patient (1/338) in the CheckMate 017, CheckMate 057 and CheckMate 078 trials had complete response respectively. There was no report of complete response in the KEYNOTE-010 trial.
Firstly, we performed a traditional pairwise meta-analysis of studies that directly compared diferent treatment modalities. In an analysis of overall survival in the direct meta-analysis, nivolumab and pembrolizumab both showed survival benefits compared to docetaxel (nivolumab: HR 0.68; 95% CI 0.59–0.78; pembrolizumab HR 0.66; 95% CI 0.57–0.76) (Fig. 4). In the direct pairwise meta-analysis for progression-free survival, nivolumab showed survival benefit compared to docetaxel (HR 0.78; 95% CI 0.63–0.96), pembrolizumab also showed benefit when compared with docetaxel (HR 0.84; 95% CI 0.7–0.95) (Fig. 5). In the direct pairwise meta-analysis for overall response rate, nivolumab and pembrolizumab also showed efficacy compared to docetaxel (nivolumab: OR 2.5; 95% CI 1.41–4.46; pembrolizumab OR 2.17; 95% CI 1.57–3.00) (Fig. 6).
Subsequently, we performed a network meta-analysis of studies that indirectly compared nivolumab and pembrolizumab. Nivolumab showed no signifcant difference in overall survival, progression-free survival and overall response rate when compared to pembrolizumab (OS: HR 1.03; 95% CI 0.84–1.26; PFS: HR 0.95; 95% CI 0.80–1.14; ORR: HR 1.08; 95% CI 0.67–1.73). (Fig. 7).
Safety
For the safety analysis, 2299 patients were included. In the CheckMate 017, CheckMate 057, CheckMate 078 and KEYNOTE-010 trials, the toxicity profile of 12 patients, 22 patients, 11 patients, and 42 patients respectively were missing from the results. The most common treatment-related adverse events (AEs) were fatigue, decreased appetite, rash, asthenia, diarrhea and nausea. We also performed a traditional pairwise meta-analysis of studies that directly compared the toxicity. Indirect comparison showed nivolumab and pembrolizumab both have less all-grade toxicity (except the toxicity of rash) when compared with docetaxel (for any event toxicity, nivolumab: OR 0.30; 95% CI 0.22–0.39; pembrolizumab OR 0.42; 95% CI 0.33–0.55) (Fig. 8). For the grade 3–5 toxicity, nivolumab and pembrolizumab also demonstrated less toxicity (except the toxicity of rash) when compared with docetaxel (for any event at grade 3–5 toxicity, nivolumab: OR 0.10; 95% CI 0.07–0.13; pembrolizumab: OR 0.31; 95% CI 0.24–0.40) (Fig. 9).
In indirect comparison, nivolumab showed less all-grade toxicity when compared with pembrolizumab, especially in subgroup of nausea (OR 0.57; 95% CI 0.34–0.95) (Fig. 10), however, there was no significant difference (for any event toxicity, OR 0.71; 95% CI 0.49–1.04). For the grade 3–5 toxicity, nivolumab showed signifcant difference when compared to pembrolizumab (for any event at grade 3–5 toxicity, OR 0.32; 95% CI 0.21–0.49) (Fig. 11). In general, it seems that nivolumab is safer than pembrolizumab.