HPV Co-infections Associated with Genital Warts: Machine Learning Approach

Genital warts are the most common clinical manifestation of human papillomavirus (HPV) infection. Although benign, genital warts are highly infectious and can cause psychosocial distress and physical discomfort. HPV6 and HPV11 are the most two common HPV types in genital warts, however, the impact of HPV co-infections is still unclear. The objective of this study was to evaluate HPV co-infections associated with condyloma, the most common type of genital warts. A total of 1806 men within the HPV Infection in Men (HIM) Study who had > 2 visits with a valid biopsy sample and > 21 months follow-up for those did not develop condyloma were included. Logistic regression was used to evaluate HPV types at baseline associated with 2-year cumulative incidence of condyloma. The pairwise HPV co-infection patterns among 37 HPV types were evaluated using the Multivariate Adaptive Regression Splines (MARS), a machine learning approach. Fourteen types of HPV were significantly associated with condyloma incidence; the top five HPVs were HPV 6 (OR= 17.0, p=2x10^-32), HPV11 (OR=8.3, p=2x10^-5), HPV51 (OR= 2.8, p=0.0003), HPV42 (OR=6.4, p=0.0008) and HPV55 (OR=3.4, p=0.0009).  Five pairs of HPV co-infections were significantly associated with condyloma incidence, including HPV6+HPV42 (OR=16.7 of any positive vs. both negative, p=3x10^-35), HPV6+HPV16 (OR=8.7, p=3x10^-27), HPV6+HPV54 (OR=10.4, p=3x10^-29), HPV11+HPV51 (OR=3.8,p=1x10^-7), and HPV55+HPV68 (OR=3.4, p=9x10^-5). Our findings suggest that HPV co-infection with certain HPV types may be more strongly associated with genital condyloma than individual HPV infections.