External validation of the Annual Recurrence Risk Model (ARRM) for tailored surveillance strategy in patients with cervical cancer

Objective The Annual Recurrence Risk Model (ARRM), developed by the SCCAN consortium and endorsed by the European Society of Gynecological Oncology, predicts the annual risk of cervical cancer recurrence. However, it lacks an external validation, which we aimed to address in the current retrospective study. Methods We included patients with pathology confirmed T1a-T2b cervical cancers who underwent radical surgery at the European Institute of Oncology, Milan, from 01/2010-12/2022. Through the ARRM risk calculator, patients were assigned a score from 0 to 100 points, which allowed classification into five risk groups (0, 1-25, 26-50, 51-75, 76-100 points). Differences in five-year disease-free survival were evaluated through log-rank tests with pairwise comparisons. Annual risk of recurrence was calculated using conditional survival analysis. Results Overall, 411 patients with cervical cancers were included: 0 (0.0%) scored 0 points, 149 (36.3%) 1-25 points, 224 (54.5%) 26-50 points, 37 (9.0%) 51-75 points, 1 (0.2%) 76-100 points. The patient from 76-100 points was excluded from further analyses. Five-year disease-free survival rates were 96.3% [95%CI: 90.0-98.6], 85.7% [95%CI: 80.1%-89.9%], and 66.6% [95%CI: 47.3%-80.2%] in groups 1-25, 26-50, and 51-75 points, respectively (p<0.01). Compared to 26-50 and 51-75 points, the annual risk of recurrence was lower in the 1-25 points group, at around 1% from year 1 to 5. Conclusion The ARRM tool confirmed its validity in stratifying cervical cancer into groups with significantly different disease-free survival rates in an independent large population from a tertiary center. The annual risk of recurrence should be carefully considered when tailoring follow-up, always taking into account patient's perspective.