Modelling Impact of Prevention Strategies on Cervical Cancer Incidence in SA
Prof Alex Welte
- The DST/NRF Centre of Excellence in Epidemiological Modelling and Analysis (SACEMA), University of Stellenbosch
Title of the project
Modelling the impact of prevention strategies on cervical cancer incidence in South Africa.
In recent years, mathematical models are increasingly applied to cervical cancer research.
Globally, but particularly important for the South African context, existing models share the
limitation that the well-established interaction between HIV and HPV is not dynamically
simulated, and therefore the need exists for a sophisticated model that fully incorporates the
effect of HIV on HPV incidence and cervical disease progression. This study proposes to 1)
extend an individual-based HIV/STI model to include HPV and its progression to cervical cancer,
calibrate the model to South African HPV and HIV prevalence data and validate the model with
cervical cancer incidence data and 2) evaluate the potential impact of various prevention
methods on cervical cancer incidence.
HPV vaccine efficacy trials have shown high levels of vaccine effectiveness against new
detection of HPV – more than 95% in the most “naïve” groups (women aged 15-25 with no
evidence of previous HPV infection) to as low as 50% in women aged 25 or older who could have
evidence of previous infection (DNA or seropositive). In the discussion sections of the papers
describing the RCTs, authors state that this differential effectiveness could be attributed to
possible reactivation of latent HPV infections – i.e. infections with viral loads below detectable
thresholds. However, models that estimate the long term impact of HPV vaccination on HPV
prevalence or cervical cancer incidence have not considered reactivation of latent infection. We
show that models that include viral latency in the natural history structure of HPV fit better to
HPV prevalence data, produce vaccine effectiveness estimates similar to those in RCTs and
predict a slower decline in HPV prevalence than models that do not include this stage.
Using individual-level data from the Western Cape Department of Health, we can estimate
coverage of cervical cancer screening in an unbiased way. In 2018/19, coverage among the
general population of women aged 30 and older accessing public health services is 57%, well
below the national target of 70%. Coverage of screening for the HIV positive population has been
stable at 50% for the last decade, but coverage among women on ART is higher, at 56%, than
among women HIV positive but not on ART (34%). For the last 5 years, 50% of HIV negative
women who are referred to colposcopy services have no evidence of attending such services.
This fraction is higher, around 60% for HIV positive women. Overall, age-standardised cervical
cancer incidence was 58 per 100,000 women in 2014, much higher than the national estimate of
22 per 100,000 reported by the National Cancer Registry, which only reports on cancer diagnosed
by pathology. Cervical cancer incidence among adult women on ART was 380 per 100,000
women, very similar to the estimate of 450 per 100,000 reported in the IeDEA data.
- Estimated impact of human papillomavirus vaccines on infection burden: the effect of structural assumptions (2019)
- Are associations between HIV and human papillomavirus transmission due to behavioural confounding or biological effects? (2018)