Estimating the effect of HIV on cervical cancer elimination in South Africa: Comparative modelling of the impact of vaccination and screening

Abstract

Background: In 2020, the World Health Organization (WHO) launched its initiative to eliminate cervical cancer as a public health problem. To inform global efforts for countries with high HIV and cervical cancer burden, we assessed the impact of human papillomavirus (HPV) vaccination and cervical cancer screening and treatment in South Africa, on cervical cancer and the potential for achieving elimination before 2120, considering faster HPV disease progression and higher cervical cancer risk among women living with HIV(WLHIV) and HIV interventions.

Methods: Three independent transmission-dynamic models simulating HIV and HPV infections and disease progression were used to predict the impact on cervical cancer incidence of three scenarios for all women: 1) girls' vaccination (9-14 years old), 2) girls' vaccination plus 1 lifetime cervical screen (at 35 years), and 3) girls' vaccination plus 2 lifetime cervical screens (at 35 and 45 years) and three enhanced scenarios for WLHIV: 4) vaccination of young WLHIV aged 15-24 years, 5) three-yearly cervical screening of WLHIV aged 15-49 years, or 6) both. Vaccination assumed 90% coverage and 100% lifetime protection with the nonavalent vaccine (against HPV-16/18/31/33/45/52/58). Cervical cancer screening assumed HPV testing with uptake increasing from 45% (2023), 70% (2030) to 90% (2045+). We also assumed that UNAIDS 90-90-90 HIV treatment and 70% male circumcision targets are reached by 2030. We examined three elimination thresholds: age-standardised cervical cancer incidence rates below 4 or 10 per 100,000 women-years, and >85% reduction in cervical cancer incidence rate. We conducted sensitivity analyses and presented the median age-standardised predictions of outcomes of the three models (minimum-maximum across models).

Findings: Girls' vaccination could reduce age-standardised cervical cancer incidence from a median of 47.6 (40.9-79.2) in 2020 to 4.5 (3.2-6.3) per 100,000 women-years by 2120, averting on average ∼4% and ∼46% of age-standardised cumulative cervical cancer cases over 25 and 100 years, respectively, compared to the basecase. Adding 2 lifetime screens helped achieve elimination over the century among all women (2120 cervical cancer incidence: 3.6 (1.9-3.6) per 100,000 women-years), but not among WLHIV (10.8 (5.3-11.6)), and averted more cumulative cancer cases overall (∼45% over 25 years and ∼61% over 100 years compared to basecase) than girls' vaccination alone. Adding three-yearly cervical screening among WLHIV (to girls' vaccination and 2 lifetime cervical screens) further reduced age-standardised cervical cancer incidence to 3.3 (1.8-3.6) per 100,000 women-years overall and to 5.2 (3.9-8.5) among WLHIV by 2120 and averted on average 12-13% additional cumulative cancer cases among all women and 21-24% among WLHIV than girls' vaccination and 2 lifetime cervical screens over 25 years or longer. Long-term vaccine protection and using the nonavalent vaccine was required for elimination.

Interpretation: High HPV vaccination coverage of girls and 2 lifetime cervical screens could eliminate cervical cancer among women overall in South Africa by the end of the century and substantially decrease cases among all women and WLHIV over the short and medium term. Cervical cancer elimination in WLHIV would likely require enhanced prevention strategies for WLHIV. Screening of WLHIV remains an important strategy to reduce incidence and alleviate disparities in cervical cancer burden between women with and without HIV, despite HIV interventions scale-up.

Funding: World Health Organization. National Cancer Institute, National Institutes of Health. MRC Centre for Global Infectious Disease Analysis, UK Medical Research Council. National Institute of Child Health and Human Development research. Cancer Association of South Africa. Canadian Institutes of Health Research and the Fonds de recherche du Québec - Santé research.

Keywords: Cervical cancer elimination; HIV; HPV vaccination; South Africa; cervical cancer screening.

Fig. 1

Temporal dynamics of cervical cancer incidence after HPV vaccination and cervical cancer screening for all women - among all women: Age-standardised cervical cancer incidence rate per 100,000 women-years among all women after the ramp-up of girls' vaccination only (Sc1), girls' vaccination and 1 lifetime cervical screen (Sc2), and girls' vaccination and 2 lifetime cervical screens (Sc3). Panels show median predictions from the A) Det_HIV-HPV and B) MicroCOSM-HPV models for South Africa and the C) DRIVE model for KZN province. Vaccine coverage = 90%, Vaccine efficacy = 100% against HPV-16/18/31/33/45/52/58, Vaccine duration = Lifetime; Screening = HPV testing, Screening uptake = 45% (2023–2029), 70% (2030–2044), 90% (2045+). Treatment efficacy = 100%, Lost to follow-up = 10%. Scenarios are as described in Table 1.

Fig. 2

Relative impact of HPV vaccination and cervical cancer screening for all women – among all women: Relative decrease in median age standardised cervical cancer incidence among all women over time compared to basecase after the ramp-up of girls' vaccination only (Sc1), girls' vaccination and 1 lifetime cervical screen (Sc2), and girls' vaccination and 2 lifetime cervical screens (Sc3). Panels show predictions from the A) Det_HIV-HPV and B) MicroCOSM-HPV models for South Africa and the C) DRIVE model for KZN province. The vertical dotted lines in panels indicate the time when cancer incidence is reduced by 50%. Vaccine coverage = 90%, Vaccine efficacy = 100% against HPV-16/18/31/33/45/52/58, Vaccine duration = Lifetime; Screening = HPV testing, Screening uptake = 45% (2023–2029), 70% (2030–2044), 90% (2045+). Treatment efficacy = 100%, Lost to follow-up = 10%. Scenarios are described in Table 1.

Fig. 3

Temporal dynamics of cervical cancer incidence after HPV vaccination and cervical cancer screening for all women - by HIV and treatment status: Age standardised cervical cancer incidence per 100,000 women-years among women stratified by HIV and treatment status after HPV girls' vaccination only (Sc1) (A,B,C) and girls' vaccination and 2 lifetime cervical screens (Sc3) (D,E,F). Panels show median predictions from the Det_HIV-HPV (A, D) and MicroCOSM-HPV (B,E) models of South Africa and the DRIVE model of KZN (C,F). Vaccine coverage = 90%, Vaccine efficacy = 100% against HPV-16/18/31/33/45/52/58, Vaccine duration = Lifetime; Screening = HPV testing, Screening uptake = 45% (2023–2029), 70% (2030–2044), 90% (2045+). Treatment efficacy = 100%, Lost to follow-up = 10%. Scenarios are described in Table 1.

Fig. 4

Incremental impact of enhanced HPV vaccination and cervical cancer screening scenarios for WLHIV – among all women and among WLHIV: Relative decrease in median age-standardised cervical cancer incidence over time compared to basecase. Panels A and D compare girls' vaccination only (Sc1) and girls' vaccination + vaccination of young WLHIV (Sc4). Panels B and E compare girls' vaccination +2 lifetime cervical screens for all women (Sc3) and Sc3+3 yearly screening of WLHIV (Sc5). Panels C and F compare girls' vaccination +2 lifetime cervical screens for all women (Sc3) and scenario 3+vaccination and 3-yearly screening of young WLHIV(Sc6). Panels show predictions from each model among all women (A–C) and among WLHIV (D–F). The vertical dotted lines indicate when cancer incidence is reduced by 50%. Vaccine coverage = 90%, Vaccine efficacy = 100% against HPV-16/18/31/33/45/52/58, Vaccine duration = Lifetime; Screening = HPV testing, Screening uptake = 45% (2023–2029), 70% (2030–2044), 90% (2045+). Treatment efficacy = 100%, Lost to follow-up = 10%. Scenarios are described in Table 1.

Fig. 5

Cumulative fraction of cervical cancer cases averted after HPV vaccination and cervical cancer screening – among all women and among WLHIV: Age-standardized fraction of cumulative cervical cancer cases averted over time since 2020 following girls' vaccination only (Sc1), girls' vaccination + 1 lifetime cervical screen (Sc2), girls' vaccination + 2 lifetime cervical screen (Sc3) in panels A–B, F-G, and Sc1 + vaccination of young WLHIV (Sc4), Sc3+ 3-yearly screening of WLHIV (Sc4), or Sc3 + vaccination and 3-yearly screening of WLHIV (sc6) in panels C–E, H-J. Results are presented among all women (A–E) and among WLHIV (F–J). The relevant scenarios are compared to basecase (A, C, F, H), girls' vaccination alone (B, D, G, I), and girls' vaccination plus 2 lifetime cervical screens (E, J). The numbers represent the median of the three models. The error bars represent the minimum and maximum of the three models. Vaccine coverage = 90%, Vaccine efficacy = 100% against HPV-16/18/31/33/45/52/58, Vaccine duration = Lifetime; Screening = HPV testing, Screening uptake = 45% (2023–2029), 70% (2030–2044), 90% (2045+). Treatment efficacy = 100%, Lost to follow-up = 10%. Scenarios are as described in Table 1.

Fig. 5

Cumulative fraction of cervical cancer cases averted after HPV vaccination and cervical cancer screening – among all women and among WLHIV: Age-standardized fraction of cumulative cervical cancer cases averted over time since 2020 following girls' vaccination only (Sc1), girls' vaccination + 1 lifetime cervical screen (Sc2), girls' vaccination + 2 lifetime cervical screen (Sc3) in panels A–B, F-G, and Sc1 + vaccination of young WLHIV (Sc4), Sc3+ 3-yearly screening of WLHIV (Sc4), or Sc3 + vaccination and 3-yearly screening of WLHIV (sc6) in panels C–E, H-J. Results are presented among all women (A–E) and among WLHIV (F–J). The relevant scenarios are compared to basecase (A, C, F, H), girls' vaccination alone (B, D, G, I), and girls' vaccination plus 2 lifetime cervical screens (E, J). The numbers represent the median of the three models. The error bars represent the minimum and maximum of the three models. Vaccine coverage = 90%, Vaccine efficacy = 100% against HPV-16/18/31/33/45/52/58, Vaccine duration = Lifetime; Screening = HPV testing, Screening uptake = 45% (2023–2029), 70% (2030–2044), 90% (2045+). Treatment efficacy = 100%, Lost to follow-up = 10%. Scenarios are as described in Table 1.