Characteristics of frequently used cervical cancer models
| Model | Model structure | Major outcomes | Challenges | Core assumptions |
|---|---|---|---|---|
| MGH-Cervical (MGH)* | Dynamic compartmental model with 3 integrated components: (1) HPV and HIV transmission, (2) HIV progression (CD4/viral load) with ART, and (3) HPV-induced precancer and cancer development | Assessing population-level health outcomes with vs. without HPV vaccination to gauge vaccination effects in an HIV-affected population | Highly specialized to HIV co-endemic scenarios; requiring detailed HIV epidemiological data and is calibrated to a specific region, thus potentially limiting generalizability | Women living with HIV are assumed to have a higher risk of HPV acquisition, lower risk of clearance, and more rapid progression associated with CD4 cell count, compared with women without HIV. Women clearing HPV are considered to develop low-level natural immunity, whereas men clearing HPV are regarded as not developing natural immunity. Cancer progression is modeled in 3 stages (local, regional, and distant), with HIV status influencing both progression and mortality. |
| HSPH-Cervical (Harvard)* | Hybrid model linking a dynamic HPV transmission component (compartmental, simulating 9 HPV types) with a static individual-based cervical carcinogenesis model | Evaluating the effectiveness, feasibility, and cost-effectiveness of various screening and vaccination strategies | Does not model non-cervical HPV-associated cancers; does not explicitly model HIV | CIN1 is not modeled as an explicit state; instead, persistent HPV infection was assumed to represent CIN1. HPV infection is modeled to be able to progress directly to CIN2 or CIN3 (non-sequential progression). After invasive cancer develops, no regression is allowed, and women remain in stage-specific survival states. Multiple HPV infections are modeled independently, and type-specific immunity is considered, allowing reinfection. Cross-protection for bivalent and quadrivalent vaccines is explicitly modeled. The model is also expanded to simulate other HPV-associated cancers. |
| UMN-Cervical (Minnesota) | Comprising 2 linked models: a dynamic individual-based HPV transmission model and a cohort-based cervical disease progression model (HPV infection through CIN and cancer). | Quantifying the population-level effects of vaccination and screening | Using grouped HPV type categories (beyond HPV-16/18, pooling other high-risk types) in modeling assumptions | Sexual behavior is calibrated according to data from the U.S. NSFG. HPV infection can skip stages, progressing directly to any CIN state or regressing in a non-sequential manner. Cancer was modeled according to FIGO stages I–IV, then mapped to SEER categories. Cervical cancer survivors are assumed not to be at risk of recurrence. The baseline scenario posits perfect vaccine efficacy with lifelong immunity. Waning or vaccine failures are examined only in sensitivity analyses. |
| STDSIM-MISCAN-cervix (Erasmus) | Integrating findings of MISCAN-CERVIX (static micro-simulation model) and STDSIM (a stochastic microsimulation model for the transmission of HPV) | Projecting cervical cancer incidence and mortality (by age, stage, and year), and evaluating long-term effectiveness and cost-effectiveness of screening and vaccination strategies | Core model is static and does not explicitly simulate HPV transmission or herd immunity (requires coupling with a separate HPV transmission model for such analyses) | CIN1/2 lesions are allowed to arise without HPV infection, but are assumed not to progress to cancer. The model uses detailed FIGO staging (I–III) with sequential progression to clinical cancer and death. The duration of HPV infection and CIN lesions is considered equal across HPV types, and transitions are modeled with fixed duration distributions. |
| Policy1-Cervix (UYSD)* | Comprehensive model combining a dynamic HPV transmission/vaccination component with detailed natural history (HPV infection through CIN1–3 and cancer) and screening modules | Evaluating effectiveness and cost-effectiveness of various screening and vaccination strategies across populations | Requiring substantial locale-specific data and calibration; model accuracy depends on incorporation of local screening behaviors and demographic patterns | HPV types are grouped as 16, 18, High-5 (31/33/45/52/58), and other high-risk categories. Infections across these groups are modeled as independent and could occur simultaneously. CIN states allows “two-step” transitions, whereby CIN1 can directly progress to CIN3 or regress to normal. Women treated for precancer are assumed to continue to have elevated lifetime risk and to require additional follow-up. |
| HPV-ADVISE* | Individual-based dynamic transmission model simulating HPV infection, progression, and vaccination; core groups included in LMIC version | Estimating population-level effects, efficiency, and cost-effectiveness of HPV vaccination strategies under resource constraints; supporting WHO global recommendations | Requiring detailed behavioral and sexual activity data for accurate calibration; assumptions regarding core groups might limit generalizability to settings lacking such data | Multiple concurrent infections are allowed, with no cross-type interactions. The model also includs other HPV-associated cancers. Stochastic pair formation is incorporated, with a dedicated MSM sub-model in which vaccinating boys has both direct and indirect effects on MSM, whereas vaccinating only girls has no effects on MSM. |
| OncoSim-Cervical | Microsimulation model combining HPV transmission dynamics with Monte Carlo simulation of cervical cancer natural history; calibrated to Canadian registry and screening data | Projecting cervical cancer incidence and elimination timelines; assessing effects of HPV-based screening, vaccination, and follow-up interventions; supporting regional policy planning | Highly tailored to the Canadian context, such that parameters reflect local healthcare structure and behaviors; limited generalizability to low-resource or demographically different settings | The model includs only 6 grouped HPV categories (6, 11, 16, 18, other high-risk, and other low-risk types). Vaccination is modeled as a binary state (vaccinated or unvaccinated), with no dose-response or partial-course effects. The baseline scenario posits no waning of immunity and full protection for targeted types. HPV testing is assumed to be 100% sensitive and specific for infection. |
↵*HSPH-Cervical, Policy1-Cervix, and HPV-ADVISE models are used by the Cervical Cancer Elimination Modelling Consortium (CCEMC) for projections in 78 low- and middle-income countries, and have been applied in countries including India, Vietnam, Uganda, Nigeria, Papua New Guinea, and South Africa. The MGH-Cervical model has also been applied in South Africa and Kenya. These applications show that context-specific adaptation of models can provide actionable, country-tailored elimination pathways. HPV, human papillomavirus; HIV, human immunodeficiency virus; ART, antiretroviral therapy; CD4, CD4+ T lymphocytes; CIN, cervical intraepithelial neoplasia; FIGO, International Federation of Gynecology and Obstetrics; NSFG, National Survey of Family Growth.