The present evidence-based analysis was conducted following the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analysis) 2020 statement (32) and was prospectively registered in the PROSPERO (CRD42021228512). The PRISMA 2020 checklist is shown in Supplementary Table S1 . We performed a systematic literature search using PubMed, Embase, and Web of Science up to August 2021 for studies that compared the efficacy and/or safety between off-clamp and on-clamp RAPN for renal tumors and published in English. We searched the databases using the following terms: “robot-assisted”, “robotic-assisted”, “robot”, “robotic”, “partial nephrectomy”, “nephron sparing surgery”, “clamp”, “clamping”, “off-clamp”, and “on-clamp”. The detailed search strategy is presented in Supplementary Table S2 . In addition, the reference lists of all eligible studies were manually reviewed. Two investigators searched and evaluated the included studies independently. Any disagreement in literature search was resolved by consensus.

Studies were included if they met the following criteria: (1) the study design was randomized controlled, cohort, or case–control; (2) studies were conducted in adults with renal tumors; (3) studies comparing off-clamp RAPN with on-clamp RAPN; (4) at least one perioperative (operating time, estimated blood loss (EBL), conversion rate, length of stay (LOS), complication rate, and transfusion rate), renal functional (postoperative estimated glomerular filtration rate (eGFR) decrease and serum creatine increase during follow-up), or oncological (positive surgical margins rate and recurrence rate) outcome was evaluated; and (5) sufficient data to calculate odds ratio (OR) or weighted mean difference (WMD).

We excluded reviews, letters, editorial comments, case reports, conference abstracts, pediatric articles, unpublished articles, and non-English articles. We defined the off-clamp RAPN as the RAPN performed without any hilar clamping procedure, and the on-clamp RAPN was identified as clamping the main renal artery during the entire procedure. Thus, studies that focused on selective-clamp RAPN, super-selective clamp RAPN, and early-unclamping RAPN were also excluded.

Data extraction was performed by two investigators independently. Any disagreement was resolved by the third investigator to make a final decision. We extracted the following data from included studies: first author, publication year, study period, country of study, study design, sample size, age, body mass index (BMI), American Society of Anesthesiologist (ASA) score, preoperative eGFR, preoperative serum creatinine, tumor size, R.E.N.A.L. score, follow-up time, operating time, EBL, conversion rate, LOS, complication rate, transfusion rate, long-term (postoperative 6 months or longer) % decrease in eGFR, positive surgical margin rate and recurrence rate. When continuous variables in the study were reported as median with range or interquartile range, we calculated the mean ± standard deviation through the validated mathematical method (33, 34). When data were missing or not reported in the study, we contacted the corresponding authors to obtain completed data if available.

The Newcastle–Ottawa Scale (NOS) was used for evaluating the quality of included studies (35), and studies with seven to nine points were regarded as high quality (36). In addition, we assessed the level of evidence for each study according to the Oxford Centre for Evidence-Based Medicine Levels of Evidence Working Group (37). Two investigators independently evaluated the quality and level of evidence for eligible studies, and any discrepancy was resolved through discussion.

Evidence synthesis was performed in Review Manager 5.3 version (Cochrane Collaboration, Oxford, UK). The WMD and OR were applied for the comparison of continuous and dichotomous variables, respectively. All metrics were reported with 95% confidential intervals (CIs). The heterogeneity in studies was assessed through the chi-squared (χ²) test (Cochran’s Q) and inconsistency index (I ²) (38). χ² p value < 0.05 or I ² > 50% were considered as significant heterogeneity. A random-effect model was used to estimate the combined WMD or OR when significant heterogeneity was detected (χ² p value < 0.05 or I ² > 50%). Otherwise, the fixed-effect model was applied. In addition, we performed one-way sensitivity analyses to evaluate the effect of included studies on the combined results for outcomes with significant heterogeneity. Publication bias was evaluated visually by creating funnel plots via Review Manager 5.3 version (Cochrane Collaboration, Oxford, UK), as well as by conducting Egger’s regression tests (39) using Stata 12.0 version (Stata Corp, College Station, TX, USA) for outcomes with 10 or more included studies. p value < 0.05 was considered as statistically significant publication bias.

The flowchart of the systematic search and selection process is presented in Figure 1 . A total of 1,736 relevant articles in PubMed (n = 283), Embase (n = 974), and Web of Science (n = 479) were yielded through systematic literature search. After removing duplicate papers, 968 titles and abstracts were reviewed. Finally, 21 full-text articles involving 4,493 patients (1,274 off-clamp versus 3,219 on-clamp) were included for the pooled analysis (9–29). Of these articles, 7 were prospective cohort studies (9, 10, 12, 13, 16, 18, 25, 27), 10 were retrospective cohort studies (11, 14, 15, 17, 19–21, 23, 24, 29), and 3 were prospective randomized studies (22, 26, 28). Table 1 shows the characteristics, level of evidence, and quality score of each included study. The median (range) quality score was 7 (5–8), and 15 studies were identified as high quality (9, 11, 12, 14, 16–24, 28, 29). The details of quality assessment of all eligible studies are presented in Supplementary Table S3 .

There were no significant differences among the two groups in terms of age (WMD: -0.16; 95% CI: -0.73, 0.41; p = 0.58), gender (male/total, OR: 0.99; 95% CI: 0.85, 1.15; p = 0.87), BMI (WMD: 0.19; 95% CI: -0.18, 0.56; p = 0.32), ASA score (WMD: 0.02; 95% CI: -0.08, 0.13; p = 0.66), preoperative eGFR (WMD: 0.89; 95% CI: -0.33, 2.11; p = 0.15), and preoperative serum creatine (WMD: -0.01; 95% CI: -0.06, 0.03; p = 0.60). However, the two groups were significantly different in baseline characteristics in terms of R.E.N.A.L. score (WMD: -0.55; 95% CI: -0.93, -0.17; p = 0.004) and tumor size (WMD: -0.37; 95% CI: -0.67, -0.08; p = 0.01) ( Table 2 ).

Data of operating time were synthesized from 17 studies including 2,636 patients (833 off-clamp versus 1,803 on-clamp) (9–16, 18–24, 26, 27). Pooled analysis revealed a significant shorter operating time in the off-clamp group (WMD: -18.93; 95% CI: -33.87, -4.00; p = 0.01) with a significant heterogeneity (I ² = 96%, p < 0.00001) ( Figure 2A ). A visual assessment of the funnel plot indicated the presence of slight publication bias ( Figure 3A ). However, Egger’s test was not statistically significant (p = 0.737).

Analysis of EBL was conducted in 16 studies with 3,338 patients (835 off-clamp versus 2503 on-clamp) (9, 11–16, 18–22, 24, 26, 27, 29). Pooled analysis detected a significantly higher EBL in the off-clamp group (WMD: 20.27; 95% CI: 6.11, 34.44; p = 0.005) with a statistically significant heterogeneity (I ² = 79%, p < 0.00001) ( Figure 2B ). Funnel plots revealed a slight publication bias ( Figure 3B ) while no statistically significant publication bias was detected through Egger’s test (p = 0.061).

Seven studies involving 577 patients (206 off-clamp versus 371 on-clamp) were included in the analysis (14, 15, 17, 21, 22, 24, 25). Pooled results demonstrated that the rate of conversion to radical or open surgery was similar between the two groups (OR: 1.57; 95% CI: 0.54, 4.60; p = 0.41), and no significant heterogeneity was observed (I ² = 0%, p = 0.88) ( Figure 2C ).

Data of the complication rate (including intraoperative and/or postoperative complications) were available in 19 studies with a total of 3,847 patients (1,061 off-clamp versus 2,786 on-clamp) (9–11, 13–27, 29). Pooled analysis revealed a significantly lower rate of complication in the off-clamp group compared with the on-clamp group (OR: 0.72; 95% CI: 0.57, 0.92; p = 0.009) ( Figure 2E ). No significant heterogeneity (I ² = 0%, p = 0.68) and statistical (Egger’s test, p = 0.067) or visual ( Figure 3C ) evidence of publication bias were detected.

There were 16 articles that reported the data of transfusion rate between the two groups, including 1,706 patients (597 off-clamp versus 1,109 on-clamp) (9–18, 20–22, 24–26). Evidence synthesis observed a similar transfusion rate in the two groups (OR: 0.89; 95% CI: 0.55, 1.45; p = 0.64) without significant heterogeneity (I ² = 0%, p = 0.72) ( Figure 2F ) and statistical (Egger’s test, p = 0.368) or visual ( Figure 3D ) evidence of publication bias.

Eighteen studies with 3,664 patients (1,000 off-clamp versus 2,664 on-clamp) were included in the analysis for positive surgical margin rate (9–11, 13–15, 17–27, 29). Pooled analysis indicated that the off-clamp group had a significantly lower positive surgical margin rate (OR: 0.54; 95% CI: 0.36, 0.79; p = 0.002) ( Figure 4A ). No significant heterogeneity (I ² = 0%, p = 0.92) and statistical (Egger’s test, p = 0.946) or visual ( Figure 3E ) evidence of publication bias were observed.

Data of recurrence rate were obtained from nine studies with 2,462 patients (576 off-clamp versus 1,886 on-clamp) (12, 13, 15, 17, 20, 22, 24, 25, 29). No significant difference was observed between the two groups for recurrence rate (OR: 0.89; 95% CI: 0.52, 1.54; p = 0.69), and no significant heterogeneity (I ² = 0%, p = 0.87) was detected ( Figure 4B ).

Ten articles were included in the analysis for long-term % decrease in eGFR, involving 1,417 patients (566 off-clamp versus 851 on-clamp) (11–13, 18–22, 24, 28). Evidence synthesis showed that the off-clamp group had a significantly lower long-term % decrease in eGFR (WMD: -3.17; 95% CI: -5.81, -0.54; p = 0.02) with a significant heterogeneity (I ² = 81%, p < 0.00001) ( Figure 5 ). Both funnel plot ( Figure 3F ) and Egger’s test (p = 0.423) did not detect publication bias.

We conducted one-way sensitivity analyses for comparison of operating time, EBL, LOS, and long-term % decrease in eGFR to evaluate the influence of each individual study on the combined WMD through removing the individual study one by one. Sensitivity analyses revealed that the new combined WMD remained constant after exclusion of any individual study for operating time ( Figure 6A ), EBL ( Figure 6B ), LOS ( Figure 6C ), and long-term % decrease in eGFR ( Figure 6D ). However, when we excluded the data reported by Antonelli et al. in 2021 (28), the heterogeneity for the long-term % decrease in eGFR disappeared (I ² = 21%, p = 0.26), suggesting that this study accounts for most of the heterogeneity.