Results
Study selection and characteristics
We obtained 862 articles after searching through databases, and 759 articles remained after removing duplicates. We excluded 601 articles according to the titles and abstracts, then got 158 full-text articles, and finally included five RCTs and nine single-arm trials in this review (Figure 1 )(15, 21-34).
Three of five RCTs used PD‐1/PD‐L1 immune inhibitors alone, and one combined with conventional therapy. Experimental groups used both individual and combination therapies In one of the RCTs(22). Six of nine single-arm trials used PD‐1/PD‐L1 immunotherapy alone, and three combined with conventional therapy.
Our study included five phase-I clinical trials, one phase I/II trials, four phase II trials, and four phase-III trials. This review included 2,911 patients with HNC, in which control groups contained 959 patients. Experimental groups contained 1350 patients in the 5 RCTs, in which 1028 patients were treated with PD‐1/PD‐L1 immune inhibitors alone, and 322 patients were treated with PD‐1/PD‐L1 immunotherapy combining conventional therapy; while 602 patients in the nine single-arm trials, in which 504 participants were treated with PD‐1/PD‐L1 inhibitors alone and 98 participants were treated with PD‐1/PD‐L1 inhibitors with conventional therapy. The anti-PD‐1/PD‐L1 mAbs used in the experimental groups included nivolumab (n = 1), pembrolizumab (n = 6), durvalumab (n = 4), avelumab (n = 2), and atezolizumab (n = 1). The characteristics of studies and primary outcomes were presented inTable 1 and Supplementary File 2 , respectively.
OS, PFS, and PD-L1 related OS
Available data of overall survival were extracted from 3 (21, 23, 24)of 5 RCTs and represented experimental groups significantly improves overall survival (OR = 0.63, 95% CI: 0.49‐0.82, I² = 35%, P = 0.0004)(Figure 2a) while the data from 2 (21, 23)of 5 RCTs showed that there was no statistical difference in progression (OR = 0.69, 95% CI: 0.48‐1.01, I² = 0%, P = 0.05)(Figure 2b) . Notably, the risk of death was lower in the experimental group than in the control group in participants with HPV negative (OR = 0.54, 95% CI: 0.36‐0.81, I² = 0%, P = 0.003) and smoke current/former (OR = 0.64, 95% CI: 0.45‐0.91, I² = 2%, P =0.01),whereas the patients with other factors (e.g., female, ECOG=0, ECOG≥1, smoke never, and HPV positive) did not show statistically significant differences in overall survival (Supplementary File 3 ).
PD-L1 status was characterized by PD-L1 combined positive score (CPS) calculated by the percentage of tumor cells, lymphocytes, and macrophages in the amount of tumor cells (22, 23). PD-L1 was positive when PD-L1 CPS ≥1, otherwise PD-L1 was negative. In PD-L1 positive patients, experimental groups survived longer than control groups (OR = 0.75, 95% CI: 0.65‐0.85, I² = 0%, P < 0.0001) (Figure 2c ) while no statistical difference was found between the two groups in PD-L1 negative patients (OR = 0.96, 95% CI: 0.71‐1.30, I² = 0%, P = 0.79) (Figure 2d ).
HPV related OS, ORR and DCR
HPV+ patients had a more favorable OS than HPV- patients in PD-1/PD-L1 immune inhibitors alone or with conventional therapy (OR = 0.56, 95% CI: 0.44‐0.71, I² = 39%, P <0.00001) (Figure 3a ). ORR (OR = 1.66, 95%CI: 1.01‐2.73, I² = 0%, P = 0.05) (Figure 3b ) and DCR (OR = 0.92, 95%CI: 0.55‐1.56, I² = 29%, P = 0.77) (Figure 3c ) showed no significant differences between HPV positive patients and HPV negative patients.
Treatment-related AEs
the common types of treatment-related AEs
The most common type of any grade treatment-related AEs in PD-1 inhibitors groups was fatigue (18.76%, 95% CI (%): 16.35- 21.16), while grade≥3 AEs was anemia (2.07%, 95% CI (%): 1.20- 2.95). In PD-L1 inhibitors groups, the most common type of any grade AEs was also fatigue (11.22%, 95% CI (%): 8.48- 13.97) (Online Resource 2), while grade≥3 AEs was Gamma-glutamyl transferase (GGT) increased (1.18%, 95% CI (%): 0.24- 2.12). The most common type of any grade AEs in PD-L1 inhibitors with conventional therapy groups was called mucositis oral (96.08%, 95% CI (%): 90.75- 100.00), which was the most common grade≥3 AEs in the groups as well (52.94%, 95% CI (%): 39.24- 66.64). All the common types of treatment-related AEs are given inSupplementary File 4 .
Any grade AEs and grade≥3 AEs
The occurrences of any grade AEs in participants treated with PD-1/PD-L1 immunotherapy alone were much lower than standard of care (OR = 0.34, 95% CI: 0.26‐0.44, I² = 0%, P <0.00001) (Figure 4a ). The occurrences of grade≥3 AEs in PD-1/PD-L1 immunotherapy alone groups were much lower than standard of care groups (OR = 0.30, 95% CI: 0.22‐0.40, I² = 0%, P < 0.00001) (Figure 4b ).
Individual analyses of treatment-related AEs
In the PD-1/PD-L immunotherapy alone groups, the risk of some categories of any grade AEs (fatigue, nausea, decreased appetite, anemia, asthenia, vomit, stomatitis, weight loss, mucosal inflammation, alopecia, neutropenia, dermatitis acneiform, leukopenia, and neuropathy peripheral) were lower than control groups, which revealed PD-1/PD-L1 inhibitors might induce less harm in the digestive system and the blood system than standard of care. However, the incidence of hypothyroidism was higher in the experimental groups (OR = 26.88, 95% CI: 7.54‐95.86, I² = 0%, P<0.00001). The incidence of any grade of diarrhea had no significant statistical differences between experimental and control groups (Figure 5 ). Some types of grade≥3 AEs (e.g., nausea, rash, decreased appetite, anemia, asthenia, stomatitis, mucosal inflammation, neutropenia, neutrophil count decreased, and leukopenia) occurred less in experimental groups. In contrast, no statistically significant differences were observed in grade≥3 AEs (e.g., fatigue, and diarrhea) between the two groups (Figure 6 ).
In the only using PD-1 inhibitors experimental groups, the incidence of some categories of any grade AEs (e.g., fatigue, anemia, asthenia, vomit, stomatitis, weight loss, mucosal inflammation, alopecia, neutropenia, and nausea) were lower than control groups. In contrast, the incidence of any grade of diarrhea in PD-1 inhibitors experimental groups had no significant statistical differences compared to control groups (Supplementary File 5 ). There were fewer grade ≥3 AEs (e.g., anemia, asthenia, stomatitis, mucosal inflammation, neutropenia, neutrophil count decreased, rash, and nausea) in the only using PD-1 inhibitors experimental groups except for fatigue and diarrhea (Supplementary File 6 ). In contrast, the incidence of grade ≥3 of fatigue in PD-1 inhibitors experimental groups had no significant statistical differences compared to control groups.