Eligible studies
20309 relevant clinical records were generated based on the search strategy. After screening and qualification assessment, a total of 13 qualified phase II/III RCTs[5–17], including 7918 patients with advanced or metastatic NSCLC, were selected to complete this meta-analysis. The detailed investigation and research selection process was shown in Fig. 1.
Study characteristics and quality
The basic characteristics of the selected studies were shown in Table 1. All the intervention groups in the selected studies used PD-1 inhibitor (n = 7), PD-L1 inhibitor (n = 5), or PD-1/L1 + CTLA-4 inhibitors (n = 2); the control group used chemotherapy alone, among which half of them used docetaxel. There is no RCTs of using CTLA-4 inhibitor alone in the studies involved. Patients included in this study had two histological types: squamous and non-squamous. Besides, there were 7[5, 6 7, 8, 10, 11, 16] studies with non-first-line treatment and 6[9, 12, 13, 14, 15, 17] studies with first-line treatment. In addition, clinical trials with high methodological quality were evaluated via using the Cochrane Collaboration Tool. The evaluation results were shown in Supplementary Table 1.
Table 1
Characteristics of included clinical trials in the meta-analysis.
Author, Year
|
Trial Phase
|
Median follow-up time
|
Histology
|
TPS
|
Line
|
Therapeutic regimen
|
Number of patients
|
Treatment
|
Control
|
Treatment
|
Control
|
Borghaei, H. 2015
|
3
|
NA
|
non-squamous
|
all
|
non-first
|
nivolumab
|
docetaxel
|
292
|
290
|
Brahmer, J. 2015
|
3
|
12 months
|
squamous
|
all
|
non-first
|
nivolumab
|
docetaxel
|
135
|
137
|
Herbst, Roy S. 2016
|
2/3
|
NA
|
squamous or non-squamous
|
positive
|
non-first
|
pembrolizumab
|
docetaxel
|
690
|
340
|
Fehrenbacher, Louis. 2016
|
2
|
NA
|
squamous or non-squamous
|
all
|
non-first
|
atezolizumab
|
docetaxel
|
144
|
133
|
Reck, M. 2016
|
3
|
11.2 months
|
squamous or non-squamous
|
≥ 50%
|
first
|
pembrolizumab
|
platinum-based chemotherapy
|
154
|
151
|
Rittmeyer, Achim. 2017
|
3
|
21 months
|
squamous or non-squamous
|
all
|
non-first
|
atezolizumab
|
docetaxel
|
425
|
425
|
Barlesi, Fabrice. 2018
|
3
|
NA
|
squamous or non-squamous
|
all
|
non-first
|
avelumab
|
docetaxel
|
396
|
396
|
Mok, Tony S. K. 2019
|
3
|
12.8 months
|
squamous or non-squamous
|
≥ 1%
|
first
|
pembrolizumab
|
chemotherapy
|
637
|
637
|
Hellmann, M. D. 2019
|
3
|
24 months
|
squamous or non-squamous
|
≥ 1%
|
first
|
nivolumab + ipilimumab
|
chemotherapy
|
396
|
397
|
Herbst, R. S. 2020
|
3
|
13.4 months
|
squamous or non-squamous
|
all
|
first
|
atezolizumab
|
chemotherapy
|
277
|
277
|
Rizvi, N. A. 2020
|
3
|
NA
|
squamous or non-squamous
|
≥ 25%
|
first
|
durvalumab
|
platinum-based chemotherapy
|
163
|
162
|
durvalumab + tremelimumab
|
platinum-based chemotherapy
|
163
|
162
|
Lu, S. 2021
|
3
|
25.9 months
|
squamous or non-squamous
|
all
|
non-first
|
nivolumab
|
docetaxel
|
338
|
166
|
Sezer, Ahmet. 2021
|
3
|
NA
|
squamous or non-squamous
|
≥ 50%
|
first
|
cemiplimab
|
platinum-doublet chemotherapy
|
283
|
280
|
NA: not available |
TPS: PD-1/L1 tumour proportion score |
Analysis of OS
OS was reported in all the included studies. Compared with chemotherapy alone, PD-1/L1 or CTLA-4 inhibitors alone or in combination (PD-1/L1 + CTLA-4) had been proved to be capable to prolong OS in patients with advanced or metastatic NSCLC [HR = 0.75, 95%CI (0.70–0.80), P < 0.001], as being detailly showed in Fig. 2. PD-1/L1 or CTLA-4 inhibitors alone or in combination compared with chemotherapy alone could improve OS of patients both in the first [HR = 0.77, 95%CI (0.70–0.84)] and non-first line treatment [HR = 0.74, 95%CI (0.67–0.81)] (Fig. 3). As for different targets, PD-1 inhibitors [HR = 0.70, 95%CI (0.64–0.78)], PD-L1 inhibitors [HR = 0.80, 95%CI (0.72–0.88)], and PD-1/L1 + CTLA-4 inhibitors [HR 0.81, 95%CI (0.68–0.95)] have all improved the OS, and it seemed that the pharmacological effect of PD-1 inhibitors was more noticeable (Fig. 4). Among all of the relevant drugs, nivolumab [HR = 0.71, 95%CI (0.62–0.81)], pembrolizumab [HR = 0.72, 95%CI (0.61–0.85)], and atezolizumab [HR = 0.76, 95%CI (0.66–0.86)] had been used most commonly and all of them could extend OS in patients with advanced or metastatic NCSLC, compared with chemotherapy alone (Fig. 5). In all the included studies, both male [HR = 0.71, 95%CI (0.63–0.81)] and female [HR 0.80, 95%CI (0.68–0.94)] could benefit from PD-1/L1 or CTLA-4 inhibitors alone or their combination (Fig. 6). Six studies collaboratively showed both patients < 65 years [HR = 0.74, 95%CI (0.66–0.82)] and patient ≥ 65 years [HR = 0.76, 95%CI (0.65–0.89)] could get a prolonged OS with PD-1/L1 or CTLA-4 inhibitors alone or their combination compared with chemotherapy alone (Fig. 7). PD-1/L1 and CTLA-4 inhibitors showed a greater improvement in OS than chemotherapy not only in patients with advanced or metastatic squamous NSCLC [HR = 0.76, 95%CI (0.67–0.87], but also in patients with advanced or metastatic non-squamous lung cancer [HR = 0.66, 95%CI (0.58–0.76)] (Fig. 8). In these included studies, with PD-1/L1 or CTLA-4 inhibitors alone or their combination, both group of patients, with TPS ≥ 0.01 [HR = 0.76, 95%CI (0.71–0.82)] and TPS < 0.01 [HR = 0.75, 95%CI (0.58–0.97] respectively, had a longer OS compared with patient group who were given chemotherapy only (Fig. 9).
Analysis of PFS
PFS was reported in 12 studies. Similar to OS, giving PD-1/L1 or CTLA-4 inhibitors alone or in combination resulted better in PFS than applying chemotherapy alone in patients with advanced or metastatic NSCLC [HR = 0.83, 95%CI (0.73–0.95), P < 0.001] (detailed data are shown in Supplementary Fig. 1). PD-1/L1 or CTLA-4 inhibitors alone or in combination was directly related to better clinical efficacy in both first-line [HR = 0.77, 95%CI (0.58–1.01)] and non-first line treatment [HR = 0.89, 95%CI (0.79–1.01)] (Supplementary Fig. 2). Among PD-1 [HR = 0.75, 95%CI (0.61–0.91)], PD-L1 [HR = 0.94, 95%CI (0.81–1.09)] and PD-1/L1 + CTLA-4 [HR = 1.05, 95%CI (0.72–1.53)] inhibitors, only PD-1 has showed a statistically significant improvement, comparing with chemotherapy alone, in PFS in patients with advanced or metastatic NSCLC (Supplementary Fig. 3). As for nivolumab [HR = 0.78, 95%CI (0.63–0.96)], pembrolizumab [HR = 0.79, 95%CI (0.57–1.11)], and atezolizumab [HR = 0.89, 95%CI (0.77–1.02)], only nivolumab has demonstrated a statistically significant improvement in PFS in advanced or metastatic NSCLC patients compared to chemotherapy alone (supplemental Fig. 4). In the study of PFS, only male patients [HR = 0.60, 95%CI (0.44–0.81)] who had received treatment with PD-1/L1 or CTLA-4 inhibitors alone or in their combination achieved better treatment outcomes than chemotherapy alone eventually. Female [HR = 1.06, 95%CI (0.72–1.57)] patients, on the other hand, have indicated no statistically significant improvement (Supplementary Fig. 5). When it comes to different age groups, PD-1/L1 or CTLA-4 inhibitors alone or their combination improved the efficacy of PFS both in patients < 65 years [HR = 0.72, 95%CI (0.54–0.95] and ≥ 65 years [HR = 0.74, 95%CI (0.52–1.05)]. However, the difference was not statistically significant in the ≥ 65 years age group (Supplementary Fig. 6). As for histological types, both patients with squamous [HR = 0.61, 95%CI (0.48–0.77)] and patients with non-squamous [HR = 0.82, 95%CI (0.64–1.04] had a longer survival if they were given PD-1/L1 or CTLA-4 inhibitors alone or in their combination comparing to those who had received chemotherapy alone. However, only patients with squamous subtype had statistically significant improvement in treatment outcomes (Supplementary Fig. 7). In these included studies, comparing pharmacological treatment with PD-1/L1 or CTLA-4 inhibitors alone or in combination with chemotherapy alone, there was a better prognosis in terms of PFS in patients with TPS ≥ 0.01 [HR = 0.82, 95%CI (0.65–1.03)] and patients with TPS < 0.01 [HR = 0.66, 95%CI (0.43–1.01], though the difference is not statistically significant (Supplementary Fig. 8).
Analysis of AEs
Any grade or 3–5 grade AEs were extracted from 4184 patients in the trial group and 3506 patients in the control group. A total of 5794 patients who were classified in all 5 grades of AEs are covered in this meta-analysis. Since a large number of AEs types were reported, the most common AEs for analysis were selected and showed detailly in Table 2. Compared with controls, treatment with PD-1/PD-L1 or CTLA-4 inhibitors alone or in combination was associated with fewer AEs in any grades [RR = 0.77, 95%CI (0.70–0.84)] and AEs in 3–5 grade [RR = 0.41, 95%CI (0.31–0.54)]. This statistical difference was also observed in most common AEs, including neutropenia [any grade RR = 0.05; 3–5 grade RR = 0.03], fatigue [any grade RR = 0.55; 3–5 grade RR = 0.51], nausea [any grade RR = 0.30; 3–5 grade RR = 0.32] and decreased appetite [any grade RR = 0.53; 3–5 grade RR = 0.48]. Meanwhile, this statistical difference was not investigated in some other AEs, such as diarrhea, alopecia, and constipation. Besides, the incidence of rash in any grade of AEs [RR = 1.69, 95%CI (1.12–2.55)] did not have any advantage over the conventional chemotherapy, but there was no statistical difference in 3–5 grade RR rash [RR = 2.24, 95%CI (0.92–5.43)].
Table 2
Subgroup analysis of the adverse events (AEs).
Immunotherapy vs. chemotherapy
|
No. of
studies
|
RR
|
95%CI
|
Heterogeneity
|
(I2) (%)
|
Any grade adverse events
|
11
|
0.77
|
0.70–0.84
|
93
|
Any grade fatigue
|
13
|
0.55
|
0.43–0.64
|
59
|
Any grade nausea
|
12
|
0.3
|
0.21–0.43
|
91
|
Any grade decreased appetite
|
13
|
0.53
|
0.42–0.67
|
79
|
Any grade asthenia
|
10
|
0.58
|
0.46–0.73
|
62
|
Any grade diarrhea
|
11
|
0.71
|
0.49–1.04
|
89
|
Any grade anemia
|
11
|
0.16
|
0.10–0.24
|
88
|
Any grade alopecia
|
10
|
0.02
|
0.01–1.03
|
0
|
Any grade neutropenia
|
12
|
0.05
|
0.03–0.07
|
38
|
Any grade febrile neutropenia
|
6
|
0.02
|
0.01–0.07
|
0
|
Any grade leukopenia
|
6
|
0.14
|
0.08–0.27
|
38
|
Any grade rash
|
9
|
1.69
|
1.12–2.55
|
74
|
Any grade vomiting
|
6
|
0.28
|
0.13–0.61
|
92
|
Any grade constipation
|
7
|
0.28
|
0.13–0.61
|
92
|
Any grade thrombocytopenia
|
6
|
0.08
|
0.05–0.14
|
4
|
Any grade 3 or higher adverse events
|
11
|
0.41
|
0.31–0.54
|
93
|
Grade 3 or higher fatigue
|
13
|
0.51
|
0.35–0.75
|
26
|
Grade 3 or higher nausea
|
12
|
0.32
|
0.17–0.60
|
2.8
|
Grade 3 or higher decreased appetite
|
13
|
0.48
|
0.29–0.80
|
0
|
Grade 3 or higher asthenia
|
10
|
0.28
|
0.17–0.48
|
0
|
Grade 3 or higher diarrhea
|
11
|
0.75
|
0.32–1.78
|
63
|
Grade 3 or higher anemia
|
11
|
0.12
|
0.06–0.21
|
60
|
Grade 3 or higher alopecia
|
10
|
0.16
|
0.04–0.61
|
0
|
Grade 3 or higher neutropenia
|
12
|
0.03
|
0.02–0.05
|
0
|
Grade 3 or higher febrile neutropenia
|
6
|
0.02
|
0.01–0.07
|
0
|
Grade 3 or higher leukopenia
|
6
|
0.06
|
0.02–0.15
|
0
|
Grade 3 or higher rash
|
9
|
2.24
|
0.92–5.43
|
0
|
Grade 3 or higher vomiting
|
6
|
0.23
|
0.08–0.65
|
0
|
Grade 3 or higher constipation
|
7
|
0.98
|
0.29–3.29
|
0
|
Grade 3 or higher thrombocytopenia
|
6
|
0.04
|
0.02–0.12
|
0
|
RR: risk ratio; CI: confidence interval. |
Publication bias and sensitivity analysis
When it comes to OS, no significant publication bias was elucidated via Egger’s test (P = 0.126, Supplementary Fig. 9), and sensitivity analysis indicated that the result was insensitive to the included studies (Supplementary Fig. 10). As for PFS, Egger's test showed that there was no statistically significant publication bias (P = 0.1, Supplementary Fig. 11), and sensitivity analysis indicated that the result was insensitive to included studies (Supplementary Fig. 12).