A risk-stratified follow-up strategy based on residual cervical biopsy after conization for HSIL: a randomized comparative study with an exploratory cost analysis

High-grade squamous intraepithelial lesion (HSIL) is a common cervical precancerous lesion, and cervical conization is its standard treatment. Postoperative residual lesions and recurrence remain major clinical challenges. This study evaluated whether intraoperative multi-point biopsy of the residual cervix during conization can support a risk-stratified follow-up strategy for individualized postoperative management of patients with HSIL. In this prospective, randomized, parallel-group comparative study with an exploratory cost comparison, 200 patients with HSIL who underwent cold knife conization between June 2022 and May 2024 were randomly assigned in a 1:1 ratio to the study group ( n = 100; conization plus intraoperative multi-point biopsy of the residual cervix) or the control group ( n = 100; conization alone). Except that double-negative patients (negative surgical margin and negative residual cervical pathology) in the study group received simplified low-risk surveillance, all patients received standard surveillance. The primary endpoint was the lesion residual rate within 6 months after surgery; secondary endpoints were the 2-year recurrence rate, the missed diagnosis rate, follow-up adherence, and healthcare resource use and direct medical costs. All analyses were performed on an intention-to-treat basis. This study was not prospectively registered. The lesion residual rate (20.0% vs. 16.0%; risk difference + 4.0%, 95% CI − 6.7 to + 14.7), the 2-year recurrence rate among patients without initial residual disease (15.0% [12/80] vs. 13.1% [11/84]; risk difference + 1.9%, 95% CI − 8.9 to + 12.9) and the missed diagnosis rate (1.0% vs. 1.0%) did not differ significantly between the groups. With 100 patients per group, the study was powered only to detect absolute differences of approximately 17% points or larger in these low-frequency safety outcomes and therefore could not demonstrate non-inferiority. Among margin-negative patients, the double-negative subgroup (M⁻R⁻) had the most favourable prognosis (6-month HPV clearance 96.8%; lesion residual and 2-year recurrence rates both 1.6%), whereas the very small residual-positive subgroup (M⁻R⁺, n = 5) showed poor outcomes; these subgroup analyses are exploratory. Protocol-based follow-up adherence was higher in the M⁻R⁻ subgroup than in the control group (91.9% vs. 66.0%; risk difference + 25.9%, 95% CI + 13.2 to + 36.7). The study group had fewer per-patient follow-up visits (mean difference − 2.1, 95% CI − 2.3 to − 1.9) and lower mean annual direct medical costs (mean difference − 1134 CNY, 95% CI − 1253 to − 1015). Intraoperative multi-point biopsy of the residual cervix is technically feasible and identified a small subgroup of margin-negative patients with occult residual lesions. In this exploratory, single-centre study, a risk-stratified follow-up strategy was associated with higher protocol-based adherence among low-risk patients and lower healthcare resource use, without a statistically significant difference in residual or recurrent disease. Because the study was not powered to assess non-inferiority, and the low-risk classification relies on a post-randomization measurement that is not available in the control group, these findings are hypothesis-generating and cannot currently justify changes to established postoperative surveillance recommendations; they require external validation before routine individualized follow-up can be considered. This study was not prospectively registered in a public trial registry this is acknowledged in the Limitations.