Assessing the Impact of Methodological Assumptions on Estimated Age of Disease-Causal Human Papillomavirus Infection in US Females

Alhaji Cherif, Xuedan You, J.r. Cook, Miriam Reuschenbach, Ya-Ting Chen, Craig S. Roberts, Ruifeng XuOpen 2025 · DOI

TL;DR

This study adapted a simulation model to estimate the age at which women in the US acquire HPV infections that lead to disease. They found that different assumptions about how long it takes for an infection to progress to CIN2+ (a precancerous lesion) had minimal impact on the estimated age of infection, highlighting the broad effectiveness of HPV vaccination.

What Problem Does This Paper Address?

Human Papillomavirus (HPV) is a major cause of cervical cancer and its precursor, CIN2+. While HPV vaccination is widely available, understanding the natural history of HPV infection, particularly the age at which disease-causing infections are acquired, is crucial for public health strategies. Previous studies have reported varying median ages for HPV infection, and the impact of methodological assumptions, such as the time lag between infection and disease onset and limitations in clinical trial data, on these estimates has not been fully explored. This paper aims to address this gap by investigating how these assumptions influence the estimated age distribution of disease-causal HPV infection in US females.

Methods

The researchers adapted a discrete event simulation model, previously used to estimate HPV infection and CIN2+ progression, to track 1000 women through three health stages: screening, HPV infection, and CIN2+ onset/diagnosis. The model simulates the age of causal HPV infection, CIN2+ onset, and CIN2+ diagnosis. To estimate the age of causal infection, the cumulative age distribution curve of CIN2+ diagnosis was shifted backward by an "offset" representing the time from HPV infection to CIN2+ diagnosis.

They specifically investigated two methodological factors:

  1. Time-to-CIN2+ Onset Distributions: They compared gamma and exponential distributions for the time between infection and CIN2+ onset to see how this affected the estimated age of causal infection and CIN2+ diagnosis.
  2. Censoring of Clinical Trial Data: They assessed the impact of censoring in the FUTURE I trial (which followed participants for 3 years) by exploring different assumptions about the proportion of patients with persistent infections or CIN1 who might progress to CIN2+ after the trial's end (ranging from 0% to 20%).

The model was calibrated using data from vaccine trials, screening registries, and age-specific CIN2+ diagnoses.

Datasets

The study utilized data from:

  • VIVIANE bivalent (2vHPV) and FUTURE I quadrivalent (4vHPV) HPV vaccine trials: Used to estimate the time from infection to CIN2+ onset.
  • New Mexico HPV Pap Registry (2008): Provided screening frequency data.
  • Surveillance registries: Provided age-specific CIN2+ diagnoses.

No new datasets were introduced; the study focused on re-analyzing existing data with different methodological assumptions.

Key Insights and Ideas

  • Robustness of Estimates: The study's core finding is that variations in key methodological assumptions had a surprisingly minimal impact on the estimated age distribution of disease-causal HPV infection. This suggests that the underlying estimates are relatively robust.
  • Importance of Vaccination: The minimal impact of these assumptions underscores the importance of expanded HPV vaccination efforts, as the benefits are likely to be substantial regardless of precise timing estimates.
  • Mid-Adult Risk: A significant proportion of causal HPV infections (around 41%) were estimated to be acquired after age 27, highlighting that mid-adult women remain at risk and may not have complete natural immunity.
  • Model Adaptability: The adapted discrete event simulation model proved useful for exploring different assumptions about disease natural history, demonstrating its value for public health modeling.

Key Results

  • The estimated median age for acquiring causal HPV infection was 24.9 years (95% CI: 23.8–25.9).
  • Approximately 41.4% (95% CI: 37.6–45.2) of causal HPV infections were acquired after the age of 27.
  • Using either gamma or exponential distributions for the time from infection to CIN2+ onset resulted in no statistically significant difference in the estimated median time from infection to CIN2+ diagnosis or the median age of disease-causing HPV infection.
  • The impact of censoring in clinical trial data (varying the proportion of persistent infections progressing to CIN2+ from 0% to 20%) had only a marginal and not statistically significant effect on the estimated median age at disease diagnosis and causal infection.

Limitations and Future Work

The authors acknowledge several limitations:

  • Model Simplification: The model does not differentiate between specific HPV genotypes or lesion types, which have varying oncogenic potentials and progression timelines.
  • Geographic/Demographic Variations: Findings might be influenced by variations in CIN2+ diagnosis sensitivity and screening frequency across different locations and populations.

Future research directions suggested include:

  • Genotype-Specific Trajectories: Exploring genotype-specific progression timelines to provide more nuanced insights.
  • Stratification: Stratifying the age distribution of CIN2+ diagnosis and time from infection to CIN2+ onset by genotypes and lesion types.

Summary updated

Abstract

We adapted a previous discrete event simulation model to assess the impact of varying time distributions from disease-causal human papillomavirus (HPV) infection to CIN2+ onset on the estimated age distribution of disease-casual infection in the United States. Differences in these assumptions had minimal effect on estimates, underscoring the importance of expanded HPV vaccination.
BibTeX
@inproceedings{cherif2025assessing,
  author = {Alhaji Cherif and Xuedan You and J.R. Cook and Miriam Reuschenbach and Ya-Ting Chen and Craig S. Roberts and Ruifeng Xu},
  title = {{Assessing the Impact of Methodological Assumptions on Estimated Age of Disease-Causal Human Papillomavirus Infection in US Females}},
  booktitle = {Open},
  year = {2025},
  doi = {10.1093/ofid/ofaf548},
}
Figure