Literature search results

A total of 518 relevant articles were identified. Eleven studies were selected based on the inclusion and exclusion criteria [16,17,18,19,20,21,22,23,24,25,26]. See Fig. 1 for an illustration of the literature screening process. The publication years ranged from 2009 to 2023. All selected studies were two-arm trials involving seven different treatment regimens, which included Pembrolizumab + TN, Camrelizumab + TN, Camrelizumab + TP, Pembrolizumab + PC, Pembrolizumab + TP, Camrelizumab + TP, Sintilimab + TP, and Toripalimab + TP.

Fig. 1

Literature search flow chart

Essential characteristics included in the study

Refer to Table 1 for details on the authors, publication dates, and number of cases across the 11 studies included in the study characteristics table.

Table 1 Basic information included in literatureLiterature quality evaluation

The overall quality of the 11 included literatures was fair; see Fig. 2 for details.

Fig. 2Results of network meta-analysisNetwork graph

The network diagram presented in Fig. 3 depicts the relationships among PCR, MPR, DCR, ORR, R0 resection rate, adverse reaction rate, and surgical complications.

Fig. 3Pairwise comparison

For pCR, the regimens of Camrelizumab + TN [OR = 13.93, 95% CI (1.49, 129.78)], Camrelizumab + TP [OR = 6.50, 95% CI (3.47, 12.17)], and Sintilimab + TP [OR = 7.02, 95% CI (1.44, 34.25)] were superior to chemotherapy alone. No significant differences were observed among the remaining treatment groups. In terms of MPR, Camrelizumab + TP [OR = 4.62, 95% CI (2.74, 7.77)] demonstrated superiority over chemotherapy alone. Similarly, Camrelizumab + TP [OR = 3.61, 95% CI (1.34, 9.74)] was superior in ORR. Pembrolizumab + PC [OR = 2.37, 95% CI (1.37, 4.11)], Pembrolizumab + TP [OR = 3.93, 95% CI (1.58, 9.74)], and Sintilimab + TP [OR = 4.56, 95% CI (1.60, 13.06)] were superior to chemotherapy alone in achieving R0 resection rates. Camrelizumab + TP [OR = 5.79, 95% CI (1.07, 31.16)], Pembrolizumab + PC [OR = 2.93, 95% CI (1.40, 6.15)], and Sintilimab + TP [OR = 5.25, 95% CI (1.27, 21.66)] also showed superiority over chemotherapy alone. However, no significant differences were observed between the groups in terms of myelosuppression, vomiting, and postoperative cardiac complications. Camrelizumab + TP [OR = 10.15, 95% CI (2.70, 38.15)] was significantly more likely than chemotherapy alone to cause a rash after neoadjuvant therapy. Additionally, the combination of Camrelizumab + TP [OR = 13.82, 95% CI (2.48, 76.94)] was found to be significantly more effective than Toripalimab + TP. See Tables 2, 3, 4, 5, 6, 7, 8, 9, and 10 for detailed comparisons.

Table 6 R0 resection ratesCumulative sort probability results

In terms of clinical efficacy, the SUCRA curve results indicated the following probability order for achieving pCR: Camrelizumab + TN (0.809), Sintilimab + TP (0.631), Camrelizumab + TP (0.609), Pembrolizumab + TN (0.437), and Chemotherapy (0.014). These results suggest that Camrelizumab + TN had the highest efficacy in improving the rate of pCR. In the MPR study, Camrelizumab + TP (0.976) showed greater efficacy than Pembrolizumab + TN (0.423) and Chemotherapy (0.101), indicating the highest efficacy of Camrelizumab + TP in improving the rate of MPR. For DCR, the efficacy of different treatments was ranked as Sintilimab + TP (0.817) > Pembrolizumab + TP (0.707) > Camrelizumab + TP (0.647) > Pembrolizumab + PC (0.245) > Chemotherapy (0.084), indicating that Sintilimab + TP had the highest efficacy in improving the DCR. In the ORR analysis, Sintilimab + TP (0.841) had a higher efficacy compared to Pembrolizumab + TP (0.784), Pembrolizumab + PC (0.509), Camrelizumab + TP (0.336), and Chemotherapy (0.03), suggesting that Sintilimab + TP had the highest efficacy in improving the ORR. Regarding the R0 resection rate, the order from highest to lowest was Camrelizumab + TP (0.771), Sintilimab + TP (0.754), Pembrolizumab + PC (0.558), Pembrolizumab + TN (0.416), Camrelizumab + TN (0.368), and Chemotherapy (0.132), suggesting that Camrelizumab + TP has the greatest advantage in improving the R0 resection rate.

The SUCRA curves also displayed the associated probabilities of adverse reactions and postoperative complications. For myelosuppression, the ranking from highest to lowest possibility was Pembrolizumab + TN (0.861), Toripalimab + TP (0.721), Sintilimab + TP (0.621), Chemotherapy (0.459), Camrelizumab + TP (0.311), Pembrolizumab + PC (0.298), and Pembrolizumab + TP (0.228), indicating that Pembrolizumab + TN has the highest likelihood of causing myelosuppression, while Pembrolizumab + TP has the lowest likelihood. For rash initiation, the order from highest to lowest probability was Pembrolizumab + TN (0.835), Camrelizumab + TP (0.761), Sintilimab + TP (0.61), Chemotherapy (0.204), and Toripalimab + TP (0.09), suggesting that Pembrolizumab + TN has the highest probability of rash initiation, while Toripalimab + TP has the lowest probability. For emesis, the order of probability from highest to lowest was Toripalimab + TP (0.655), Sintilimab + TP (0.644), Pembrolizumab + TN (0.626), Chemotherapy (0.395), Camrelizumab + TP (0.355), and Pembrolizumab + PC (0.326), indicating that Toripalimab + TP has the highest probability of emesis, while Pembrolizumab + PC has the lowest probability. For the incidence of postoperative cardiac events, the order from highest to lowest probability was Chemotherapy (0.565), Camrelizumab + TP (0.545), Camrelizumab + TN (0.509), and Sintilimab + TP (0.381), suggesting that traditional chemotherapy alone has the highest likelihood of causing postoperative cardiac adverse events. See Fig. 4 for details.

Fig. 4

Cumulative sort probability results

Inconsistency test

No consistency test was conducted, as none of the nine outcome measures in this study formed closed loops.

Publication bias test

We assessed publication bias across outcome measures using correction-comparison funnel plots. The plots demonstrated good symmetry, with no significant asymmetries or abnormal distributions observed. This suggested that our findings possess high statistical reliability and were not significantly affected by publication bias. Refer to Fig. 5 for visual representation.

Fig. 5