Materials and methods
Search strategy
Two authors conducted the search and study review independently. We searched PubMed, Embase, Cochrane Library, CVID, CBM, CNKI and Web of Science up to 1 November 2019. The search terms used were: PD-1/PD-L1 inhibitors or Pembrolizumab or Keytruda or Nivolumab or Opdivo or Durvalumab or MEDI4736 or Atezolizumab or checkpoint inhibitors and head and neck cancer or HNC or head and neck squamous cell carcinoma or head and neck neoplasms or HNSCC. We also searched ‘clinicaltrials.gov’ with the search term Head and Neck Squamous Cell Carcinoma, and we also searched for Squamous cell carcinoma of head and neck, Squamous cell carcinoma of the head and neck, head and neck cancer, head and neck neoplasms or HNSCC and PD-1/PD-L1 inhibitors or Pembrolizumab or Keytruda or Nivolumab or Opdivo or Durvalumab or MEDI4736 or Atezolizumab or checkpoint inhibitors.
Study selection
Randomized controlled trials (RCT) were used, and the inclusion criteria for our meta-analysis were as follows: 1) single drug treatment with PD-1/PD-L1 inhibitors; 2) the papers should have exact data whether in the time domain or frequency domain; 3) the groups being compared were treated with chemotherapy, targeted therapy, a combination of two immunosuppressive inhibitors or other drug combinations. Exclusion criteria for the meta-analysis were as follows: 1) non-randomized trials; 2) studies that were not published in English or Chinese; 3) studies that did not report the end points of overall response rate (ORR), overall survival (OS), and progression-free survival (PFS).
Data extraction and quality assessment
Two authors independently examined all potentially relevant studies identified through the search. We read all of the papers to determine whether they met our inclusion criteria and decided whether they should be included. Any disagreements were resolved by consensus. Data were extracted by one researcher and supervised by another researcher and included: 1) author information; 2) publication year; 3) number of patients in the study; 4) treatment plan; 5) study design; 6) the main results.
Data synthesis and statistical analysis
We used Review Manager software (version 5.0; Cochrane Collaboration, London, UK) as the tool for calculating the outcomes. The relative risk (RR) and 95% confidence interval (CI) were used to compare the outcomes between the different groups. The chi-squared test was used to test heterogeneity. Cochran’s Q test was applied to assess the heterogeneity of the different studies. The inconsistency of each study was measured with the I2 statistic. When I2 ≥ 50%, a random effects model was used to take account of clinical heterogeneity in the data, otherwise a fixed effects model was used. Sensitivity analysis was used to analyze the effect of poor quality studies, and publication bias was analysed using an inverted funnel chart test.