A Dynamic Prediction Study of Persistent Nine-valent Human Papillomavirus Infection Among HIV-negative Men Who Have Sex With Men in Urumqi, China: a Longitudinal and Time-to-event-based Joint Model.

BACKGROUND

Persistent infection with 9-valent human papillomavirus (9v-HPV) is a major risk factor for HPV-related cancers among men who have sex with men (MSM). This study aimed to develop a dynamic prediction model to assess whether longitudinal sexual-behavior-related and exposure-related indicators could predict persistent 9vHPV infection among human immunodeficiency virus type 1 (HIV-1)-negative MSM.

METHOD: Between 2016 and 2023, HIV-negative MSM in Urumqi, China, were followed every 6 months. Baseline questionnaires and anal swabs were collected, and HPV genotyping was performed using an assay detecting 37 HPV genotypes. The primary outcome was persistent infection with 1 or more of the 9 genotypes covered by the 9-valent HPV vaccine. Persistent 9vHPV infection was defined as detection of at least one of these 9 genotypes on 2 or more consecutive visits after a negative baseline. Key indicators included same-sex partner number, anal intercourse frequency, condom use, commercial sex, club drug use, partner-seeking venues, and sexual role. A Bayesian joint model integrating longitudinal sexual-behavior-related indicators and time-to-event outcomes was developed. Predictive accuracy was assessed using the time-dependent area under the curve (AUC) with internal validation.

RESULTS

Among 704 MSM, factors including employment, commercial sex, receptive anal intercourse, and versatile roles were associated with increased persistence risk. Higher number of sexual partners (HR = 1.95) and more frequent anal intercourse (HR = 2.04) were strong predictors. The partner-based model demonstrated superior predictive accuracy (AUC = 0.727).

CONCLUSIONS

The Bayesian joint model provides accurate dynamic predictions of persistent 9v-HPV infection in MSM, supporting personalized prevention strategies.