Assessing the Risk of Cervical Cancer Development when Introducing New Prophylactic Management Schemes at Country Level Using a Markov Model: The Case Applied to France

Objective: to assess the absolute and relative risk (RR) of developing cervical cancer (CC) at individual and population level following different age specific prevention strategies (vaccination and/or screening) in France. Method: because CC is a very long lasting disease with different identifiable disease stages a lifetime Markov process model with an annual cycle length was selected as model structure to replicate the natural history of oncogenic Human Papilloma Virus (onc-HPV) infection to cancer using 9 specific stages: No-HPV; onc-HPV; Cervical Intraepithelial Neoplasia (CIN1 and CIN23); persistent CIN23; CC; cured CC; CC death; and overall death. A screening module (organised and/or opportunistic), was integrated into the natural history model structure to indicates early detection and treatment of CIN and cancer lesions therefore reducing the natural progression of the disease. A vaccination module was also added to the model to integrate the reduction of age-specific HPV infection rate based on the latest clinical trial results of a bi-valent HPV vaccine. All other data used as model input came from literature further validated by a clinicians and epidemiologists expert group. The model follows a cohort of 11 years old young girls over life-time each exposed or not to the vaccine and/or the screening program in place (every 3 years from 20 to 65 years of age). Base case analysis at the population level, assumes a 100% vaccine coverage and 60% following the screening program. Lifetime risk assessment of developing CC selecting different ages of vaccination with or without screening programs are computed by dividing the cumulative incident number of CC over life-time by the number of subjects alive at the selected age. Results: at the individual level the model predicts a lifetime absolute CC risk at the age of 11 of 1.80% without screening and vaccination and 0.07% with both vaccination and screening (RR-reduction of 25.7). At the age of 30 these risks were respectively: 1.75% and 0.09% (RR-reduction of 19.4). At the age of 50 we have 0.98% and 0.08% (RR-reduction of 12.3). At the population level, given the vaccination coverage, the risk differs however significantly for screening with vaccination strategies. At the age of 11 for the base case the population risk was 0.19% (RR-reduction of 9.4); at the age of 30 0.44% (RR-reduction of 3.9) and at the age of 50 0.32% (RR-reduction of 3.1). Conclusion: lifetime CC risk impact of implementing vaccination will be dependent on the current screening program put into place: the younger the vaccination age and the lower the screening program pressure, the higher the vaccination benefit. However at any age-group selected -from 11 to 50 years- a dramatic risk reduction is seen when combining screening with vaccination. A lifetime Markov process model is a very helpful tool to assess the different risk estimates of cc prevention strategies.