HSPH-Cervical (Harvard) Harvard T.H. Chan School of Public Health

Model Overview

The Harvard School of Public Health (HSPH) modeling group uses two distinct models: 1) a dynamic compartmental model of human papillomavirus (HPV) sexual transmission between males and females; and 2) a static, individual-based model of HPV-induced cervical carcinogenesis. The models can be used as stand-alone models or can be linked to allow for the inclusion of direct and indirect benefits from vaccination, potential synergies between vaccination and screening strategies, and vaccine cross-protection of non-16/18 types.

HPV Transmission

Transmission of HPV infections in males and females is modeled using an agent-based simulation of sexual mixing. The agents are characterized by attributes of sex, age, and sexual activity (none, low, medium, high).¹ Every month, females and males form heterosexual partnerships, and transmission of type-specific HPV can dynamically occur as a function of sexual behavior patterns in the population, prevalence of HPV in the population (generated each cycle in the model), and female-to-male or male-to-female transmission probabilities of HPV per susceptible-infected partnership. Following clearance of HPV, individuals develop partial type- and sex-specific natural immunity, reducing future risk of that same type of infection. Deaths and births occur monthly using age- and sex-specific background mortality and fertility rates, respectively. HPV types are stratified by the nine types in the nonavalent (9v) vaccine (i.e., 16, 18, 31, 33, 45, 52, 58, 6, 11), and acquisition and clearance of each HPV type occurs independently.

Cervical Carcinogenesis

The static model is a microsimulation (i.e., individual-based) model of HPV-induced cervical carcinogenesis, in which individual women representative of a single birth cohort enter the model at an early age (e.g., 9 years) and are followed over their lifetimes.^1-4 Each woman undergoes monthly transitions between health states that describe underlying true health, including HPV infection, pre-cancer (i.e., cervical intraepithelial neoplasia [CIN]2, CIN3), and invasive cancer (i.e., local, regional, distant). CIN2 and CIN3 are modeled as non-sequential precancerous health states with distinct probabilities of progression to cancer, whereas CIN1 is interpreted as a microscopic manifestation of acute HPV infection and is therefore incorporated into the HPV-infected state.⁵ States are further stratified into oncogenic HPV types 16, 18, 31, 33, 45, 52, and 58, each considered separately; other pooled oncogenic types; and pooled non-oncogenic types. Transition probabilities can vary by age, HPV type, duration of infection or lesion status, and a woman’s history of prior HPV infection and CIN treatment. Cancer detection can occur through symptoms or screening. Each month, all women are subjected to background mortality and hysterectomy, as well as excess mortality from cervical cancer.

Vaccination

The dynamic model is used to project the effects of HPV vaccination in reducing HPV infections of the 9v types over time (16, 18, 31, 33, 45, 52, 58, 6, and 11), capturing both direct and indirect benefits. These estimates of reductions in type-specific HPV incidence are applied to the corresponding input parameters of HPV incidence in the static cervical model to account for the impact of vaccination on cervical cancer outcomes in the context of complex screening strategies. The model also captures health and cost outcomes related to non-cervical HPV-related cancers (i.e., vulvar, vaginal, penile, anal, and oropharyngeal), genital warts, and juvenile-onset recurrent respiratory papillomatosis (JORRP).

Screening, Diagnosis, and Treatment of Pre-cancer

Screening assumptions in the static model can vary by screening modality, start age, stop age, frequency, coverage, triage testing, and compliance to recommended follow-up. Tests for primary screening and triage include cytology (conventional, liquid-based), HPV DNA testing (pooled or genotyping), as well as cytology and HPV co-testing. Management of screen-positive women can vary by age, follow-up test, time to follow-up test(s), and number of negative follow-up tests required to return to routine screening.

Cancer Treatment and Survival

Cancer staging (i.e., local, regional, distant) and progression is modeled, accounting for symptomatic detection and the possibility of downstaging at diagnosis due to screening. In addition to background mortality, women with cervical cancer are subject to excess mortality, based on five-year survival estimates depending on cancer stage, age, and time since diagnosis according to Surveillance, Epidemiology, and End Results (SEER) data.

Calibration and Validation

A likelihood-based approach is used to calibrate highly uncertain parameters to fit corresponding empirical data (e.g., HPV prevalence, HPV type distribution in cancer). In the static model, parameters include: 1) age- and type-specific incidence of HPV infection; 2) natural immunity following type-specific HPV infection; and 3) progression rates of HPV to CIN2 and CIN3 and of CIN2 and CIN3 to invasive cancer. Baseline values for each of the uncertain parameters are randomly selected from a pre-determined plausible range, creating a unique natural history parameter set. Goodness of fit is ascertained by calculating the likelihood of model-projected outcomes from each parameter set against corresponding calibration targets (e.g., age- and type-specific prevalence of HPV). Parameter calibration of the dynamic transmission model followed a similar approach for the following uncertain model parameter categories: 1) sex-specific transmission probabilities of HPV per infected-susceptible partnership for each of the nine individual HPV infection types; and 2) natural immunity following type-specific HPV infection. For model validation, model-projected outcomes of cervical cancer incidence rates by age in the absence of any intervention are compared against those reported historically in SEER cancer registries prior to widespread Pap smear screening, as well as current SEER incidence and mortality rates with screening operational in the model.

References

1. Kim JJ, Simms KT, Killen J, et al. Human papillomavirus vaccination for adults aged 30 to 45 years in the United States: A cost-effectiveness analysis. PLoS Med. 2021 Mar 11;18(3):e1003534. [Abstract]
2. Burger EA, de Kok IMCM, Groene E, Killen J, Canfell K, Kulasingam S, et al. Estimating the Natural History of Cervical Carcinogenesis Using Simulation Models: A CISNET Comparative Analysis. J Natl Cancer Inst. 2020 Sep 1;112(9):955-963. [Abstract]
3. Burger EA, Kim JJ, Sy S, Castle PE. Age of Acquiring Causal Human Papillomavirus (HPV) Infections: Leveraging Simulation Models to Explore the Natural History of HPV-induced Cervical Cancer. Clin Infect Dis. 2017 Sep 15;65(6):893-899. [Abstract]
4. Simms KT, Yuill S, Killen J, Smith MA, Kulasingam S, de Kok IMCM, et al. Historical and projected hysterectomy rates in the USA: Implications for future observed cervical cancer rates and evaluating prevention interventions. Gynecol Oncol. 2020 Sep;158(3):710-718. [Abstract]
5. Spencer JC, Burger EA, Campos NG, Regan MC, Sy S, Kim JJ. Adapting a model of cervical carcinogenesis to self-identified Black women to evaluate racial disparities in the United States. J Natl Cancer Inst Monogr. 2023 Nov 8;2023(62):188-195. [Abstract]
6. de Kok IMCM, Burger EA, Naber SK, Canfell K, Killen J, Simms K, et al. The Impact of Different Screening Model Structures on Cervical Cancer Incidence and Mortality Predictions: The Maximum Clinical Incidence Reduction (MCLIR) Methodology. Med Decis Making. 2020 May;40(4):474-482. [Abstract]
7. Kim JJ, Goldie SJ. Health and economic implications of HPV vaccination in the United States. N Engl J Med. 2008;359(8):821-32. [Abstract]
8. Kim JJ, Goldie SJ. Cost effectiveness analysis of including boys in a human papillomavirus vaccination programme in the United States. BMJ. 2009;339:b3884. [Abstract]
9. Kim JJ, Kuntz KM, Stout NK, Mahmud S, Villa LL, Franco EL, et al. Multiparameter calibration of a natural history model of cervical cancer. Am J Epidemiol. 2007;166(2):137-50. [Abstract]
10. Campos NG, Burger EA, Sy S, Sharma M, Schiffman M, Rodriguez AC, et al. An updated natural history model of cervical cancer: derivation of model parameters. Am J Epidemiol. 2014;180(5):545-55. [Abstract]
11. Katki HA, Kinney WK, Fetterman B, Lorey T, Poitras NE, Cheung L, et al. Cervical cancer risk for women undergoing concurrent testing for human papillomavirus and cervical cytology: a population based study in routine clinical practice. Lancet Oncol. 2011;12(7):663-72. [Abstract]
12. Katki HA, Schiffman M, Castle PE, Fetterman B, Poitras NE, Lorey T, et al. Five-year risks of CIN 3+ and cervical cancer among women with HPV testing of ASC-US Pap results. J Low Genit Tract Dis. 2013;17(5 Suppl 1):S36-42. [Abstract]
13. Wheeler CM, Hunt WC, Cuzick J, Langsfeld E, Robertson M, Castle PE, et al. The influence of type specific human papillomavirus infections on the detection of cervical precancer and cancer: A population-based study of opportunistic cervical screening in the United States. Int J Cancer. 2014;135(3):624-34. [Abstract]