Oncologic outcomes of intersphincteric resection versus abdominoperineal resection for lower rectal cancer: a systematic review and meta-analysis.

Background: The efficacy of intersphincteric resection (ISR) surgery for patients with lower rectal cancer remains unclear compared to abdominoperineal resection (APR). The aim of this study is to compare the oncologic outcomes for lower rectal cancer patients after ISR and APR through a systematic review and meta-analysis.

Materials and methods: A systematic electronic search of the Cochrane Library, PubMed, EMBASE, and MEDLINE was performed through January 12, 2022. The primary outcomes included 5-year disease-free survival (5y-DFS) and 5-year overall survival. Secondary outcomes included circumferential resection margin involvement, local recurrence, perioperative outcomes, and other long-term outcomes. The pooled odds ratios, mean difference, or hazard ratios (HRs) of each outcome measurement and their 95% CIs were calculated.

Results: A total of 20 nonrandomized controlled studies were included in the qualitative analysis, with 1217 patients who underwent ISR and 1135 patients who underwent APR. There was no significant difference in 5y-DFS (HR: 0.84, 95% CI: 0.55-1.29; P =0.43) and 5-year overall survival (HR: 0.93, 95% CI: 0.60-1.46; P =0.76) between the two groups. Using the results of five studies that reported matched T stage and tumor distance, we performed another pooled analysis. Compared to APR, the ISR group had equal 5y-DFS (HR: 0.76, 95% CI: 0.45-1.30; P =0.31) and 5y-LRFS (local recurrence-free survival) (HR: 0.72, 95% CI: 0.29-1.78; P =0.48). Meanwhile, ISR had equivalent local control as well as perioperative outcomes while significantly reducing the operative time (mean difference: -24.89, 95% CI: -45.21 to -4.57; P =0.02) compared to APR.

Conclusions: Our results show that the long-term survival and safety of patients is not affected by ISR surgery, although this result needs to be carefully considered and requires further study due to the risk of bias and limited data.