3.1 Selection of studies and characteristics
Using the search terminology, we initially identified 3445 studies from our databases searching. Among those 3445 studies, 21 RCTs met our strict inclusion criteria (Supplementary Figure S1). All the 21 included trials were evaluated and compared side effects of ICIs therapies with other controlled treatments (e.g. chemotherapy, placebo, or ICI) in solid tumors representing data from a total of 11930 patients.
Among the studies (Table 1), two involved PD-L1 (Atezolizumab: 751 patients) [7, 8]; seven involved PD-1 (Nivolumab: 4 studies [6, 11, 25, 26], 997 patients; Pembrolizumab: 3 studies [9, 10, 27], 1305 patients); four involved CTLA-4 (Tremelimumab: 2 studies [12, 28], 705 patients; Ipilimumab: 2 studies [29, 30] 864 patients); and three studies compared combination therapy of PD-1 plus CTLA-4 inhibitors with ICI monotherapy (835 patients, Nivolumab + Iipilimumab or Pembrolizumab + Ipilimumab) [31-33]. Four studies compared combination therapy of chemotherapy plus ICIs with chemotherapy [4, 34-36]. Additionally, one study compared Pembrolizumab with Iipilimumab [37]. Eight studies had data from malignant melanoma(MM) patients[4, 25-29, 32, 33]; Six from non-small cell lung cancer (NSCLC) patients[6-9, 34, 36]; Other 6 studies from other cancers including small cell lung cancer (SCLC)[31, 35], mesothelioma[12],urothelial cancer[10], prostate cancer[30], and head-neck squamous cell carcinoma[11].
Cochrane risk of bias tool was used to measure the quality of the included studies, and the results are manifest in Figure 1, Supplementary Table S2, and Supplementary Figure S2. All the included studies had described the details regarding blinding of outcome assessment and random sequence generation. However, some of them had described incomplete outcome data and allocation concealment. Some studies failed to mention blinding of participants, personnel, and selective reporting. Other indices of bias lacked specific description in all the included clinical studies.
3.2 Risk of overall ophthalmic treatment-related adverse events
Across 20 eligible studies, 135 cases of overall ophthalmic trAEs were observed among 7588 ICI-treated patients (Table 1). The predicted incidences of overall all-grade ophthalmic trAEs were 3.32% for chemotherapy, 4.32% for combination therapy of PD-1 plus CTLA-4, 1.08% for the PD-L1 inhibitor, 1.16% for the PD-1 inhibitor, and 1.2% for the CTLA-4 inhibitor (Table 2).
The OR of all-grade ophthalmic trAEs in patients with PD-1/PD-L1 inhibitors were 0.44 (95% CI: 0.23-0.83) and 0.18 (95% CI: 0.08-0.39) respectively, and it was statistically lower than chemotherapy (p<0.05). There was no significant difference between CTLA-4 inhibitor and control (chemotherapy or placebo) (OR, 1.28; 95%CI, 0.21-7.84; p>0.05). Patients with combination therapy (PD-1 plus CTLA-4 inhibitors or ICIs plus chemotherapy) have a higher risk of all-grade trAEs than CTLA-4/PD-1 inhibitor alone (OR 5.29, 95% CI: 1.59-17.57, p<0.01,) or chemotherapy (OR, 3.44; 95%CI, 1.72-6.88; p<0.001).
The risks of high-grade ophthalmic trAEs in PD-1/PD-L1/CTLA-4 inhibitors monotherapy or combination therapy had no significant difference than controls. (Figure 2)
3.3 Risk of immune-related ophthalmic adverse events
Across enrolled 14 studies, 53 cases of irAEs were observed among the 6001 ICI-treated patients (Table 1). The predicted incidences of all-grade ophthalmic irAEs were 0.8% for chemotherapy, 2.24% for combination therapy of PD-1 plus CTLA-4, 0.16% for the PD-L1 inhibitor, 0.94% for the PD-1 inhibitor, 0.53% for the CTLA-4 inhibitor. (Table 2)
Compared with PD-1 or CTLA-4 inhibitor monotherapy, patients with ICIs combination therapy (PD-1 plus CTLA-4 inhibitors) were significantly more likely to experience all-grade ophthalmic irAEs (OR, 3.67; 95%CI: 1.08-12.50; p<0.05), as well as patients with combination therapy (ICI plus chemotherapy) compared with chemotherapy (OR, 3.69; 95%CI: 1.32-10.32; p<0.05), but not concern high-grade ophthalmic irAEs.
Compared with chemotherapy or control, the risk of all-grade and high-grade ophthalmic irAEs in PD-1/PD-L1/ CTLA-4 inhibitors had no significant difference (p>0.05). Meanwhile, the risks of ophthalmic irAEs between PD-1 and CTLA-4 inhibitors were similar (p>0.05). (Figure 3)
3.4 Risk of non-specific treatment-related ophthalmic adverse events
Across enrolled 10 studies, NS-trAEs were observed among the 3957 ICI-treated patients (Table 1). The predicted incidences of ophthalmic NS-trAEs were shown in Table 2.
The risks of all-grade ophthalmic NS-trAEs inPD-1/ PD-L1inhibitors were significantly lower than chemotherapy (OR: 0.18; 95%CI: 0.04-0.72; p< 0. 05 and OR: 0.16; 95%CI: 0.07-0.36; p<0.001), but no concern high-grade. Meanwhile, the risks between PD-1 and CTLA-4 inhibitors were similar (p>0.05) (Figure 4).