Literature Search
Initially, a sum of 570 articles were identified according to previous search strategy. Then, 328 duplications were excluded. After reviewing the title and abstract, 199 studies were removed for reasons. The full texts of the remaining articles were assessed, and 23 studies were further excluded. Eventually, we included 20 articles from the rest studies in total. A flow diagram detailing the selection process of studies was shown in Figure 1.
Study characteristics
A total of 20 studies were identified after comprehensive review and involved in the meta-analysis, which consisted of 2003 cases, including 944 anogenital SCC (246 cervical, 265 vulvar, 366 penile, 67 anal SCC cases) and 1059 oropharynx SCC patients[9-29]. All studies were retrospective and published between 2013 and 2019. Immunohistochemistry (IHC) was applied to assess the expression of PD-L1 protein, and PD-L1 positivity was defined as percent of positive tumor cells in overall cancer cells with membranous or cytoplasmic staining. The commonest cutoff value of PD-L1 positivity was 5% in the meta-analysis[10, 11, 13, 14, 18, 20, 21, 23, 27-29]. Furthermore, there were 3 studies combining percentage of positive cells with staining intensity as a new definition of PD-L1 positivity (H-score)[12, 17, 24]. In addition, the relationships between HPV status and PD-L1 were investigated in 8 studies of anogenital SCC[9-11, 13, 15, 17, 19, 20], covering cervical, anal, penile and vulvar cancer cases, and 8 oropharynx SCC studies[21-26, 28, 29]. Among these researches, HPV-DNA status were considered as the biomarker of HPV oncogenic activity and were used to assess their associations with PD-L1 expression in 3 anogenital and 1 oropharynx cancer studies[11, 15, 20, 23]. Detailed characteristics of included studies were summarized in Table 1 and Table 2.
PD-L1 expression and clinicopathological features
In anogenital SCC, advanced age, higher tumor grade and lymph node metastases were related to PD-L1 positive expression (OR=1.63, 95%CI :1.04-2.58; OR=2.49, 95% CI:1.39-4.46; OR=1.85, 95% CI:1.28-2.66) (Table 2), demonstrated by the meta-analysis. However, no correlation was found between gender, tumor stage, recurrence or distant metastases and PD-L1 expression (Table 2). Contrary to anogenital SCC, PD-L1 in oropharynx SCC was less frequent in older patients (OR=0.60, 95%CI:0.37-0.98) (Table 2). Moreover, higher tumor grade and lymph node metastases were associated with PD-L1 positive expression in oropharynx SCC (OR=3.40, 95%CI:1.81-6.40; OR=1.97, 95% CI:1.32-2.92) (Table 2). The pooled OR of 8 anogenital SCC studies showed that PD-L1 negative expression was connected with HPV positivity (OR=0.47,95%CI:0.31-0.71, p=0.000) (Figure 2a) with a fixed effects model, and there was low heterogeneity (I2=10.4%, p=0.349) (Figure 2a). In order to reduce heterogeneous variables among the anogenital SCC studies, subgroups of meta-analysis were performed based on cancer types, antibody catalogs and cut-off values of PD-L1 positivity. The results demonstrated that there were significant differences in the relationship between PD-L1 expression and HPV status in the groups of penile SCC, PD-L1 antibody (clone E1L3N) and cut-off value greater than or equal to 5% (Table S2).
However, PD-L1 expression was higher in HPV positive oropharynx SCC patients with moderate heterogeneity in a random effects model (OR=3.01, 95%CI: 1.78-5.09, p=0.00; I2=55.6%, P=0.027) (Figure 2b). Stratified analysis showed that associations between high PD-L1 level and HPV positivity was significantly different in the group of antibody (clone SP142) and cut-off value greater or less than 5% (Table S2).
PD-L1 expression and oncological prognosis
In general, anogenital SCC cases that were PD-L1 positive held a significantly declined OS compared with PD-L1 negative patients (HR=2.18, 95%CI:1.37-3.47, p=0.001; I2=0.0%, P=0.569) (Figure 3a). Subgroup analysis according to cancer types, antibody catalogs and cut-off values of PD-L1 positivity were conducted, and our meta-analysis demonstrated that the predictive value of PD-L1 expression for OS of vulvar SCC, antibody (clone 22C3) and cut-off value greater than or equal to 5% (Table S3). Our meta-analysis also suggested that PD-L1 positive penile SCC cases held a worse CSS (HR=2.45, 95%CI:1.30-4.65, p=0.006; I2=45.1%, P=0.162) (Figure 3b).
In contrast to anogenital SCC, PD-L1 expression in oropharynx SCC was also a predictive value of OS (HR=0.76, 95%CI:0.60-0.97, p=0.025; I2=0.0%, P=0.480) (Figure 3c). Subgroup analysis were also conducted in accordance with antibody catalogs and cut-off values of PD-L1 positivity, and our meta-analysis suggested that the predictive value of PD-L1 expression for OS (clone SP142) and cut-off value were less than 5% (Table S3). Meanwhile, PD-L1 expression was associated with a better DFS in oropharynx cancer patients (HR=0.50, 95%CI:0.33-0.75, p=0.001; I2=25.8%, P=0.260) with no substantial heterogeneity (Figure 3d).
Sensitivity and publication analysis
We again confirmed that there was low heterogeneity in studies in the present meta-analysis by sensitivity analysis. Moreover, no substantial asymmetry was identified by Egger and Begg tests in the light of the visual inspection of the shape. This indicted low publication bias and outcomes of meta-analysis was statistically robust.