Impact of screening test performance and cost on mortality reduction and cost-effectiveness of multimodal ovarian cancer screening

Abstract

Ongoing ovarian cancer screening trials are investigating the efficacy of a two-step screening strategy using currently available blood and imaging tests [CA125 and transvaginal sonography (TVS)]. Concurrently, efforts to develop new biomarkers and imaging tests seek to improve screening performance beyond its current limits. This study estimates the mortality reduction, years of life saved, and cost-effectiveness achievable by annual multimodal screening using increasing CA125 to select women for TVS, and predicts improvements achievable by replacing currently available screening tests with hypothetical counterparts with better performance characteristics. An existing stochastic microsimulation model is refined and used to screen a virtual cohort of 1 million women from ages 45 to 85 years. Each woman is assigned a detailed disease course and screening results timeline. The preclinical behavior of CA125 and TVS is simulated using empirical data derived from clinical trials. Simulations in which the disease incidence and performance characteristics of the screening tests are independently varied are conducted to evaluate the impact of these factors on overall screening performance and costs. Our results show that when applied to women at average risk, annual screening using increasing CA125 to select women for TVS achieves modest mortality reduction (~13%) and meets currently accepted cost-effectiveness guidelines. Screening outcomes are relatively insensitive to second-line test performance and costs. Identification of a first-line test that does substantially better than CA125 and has similar costs is required for screening to reduce ovarian mortality by at least 25% and be reasonably cost-effective.

Figure 1. Ovarian cancer case sensitivity function for PEB CA125 and HM by time prior to clinical diagnosis*

*The CA125 sensitivity function is derived by applying the PEB algorithm to serial CA125 levels measured in pre-clinical samples obtained from Carotene and Retinol Efficacy Trial Participants (see Supplemental Methods Sec 3). The baseline 5% positive test result corresponding to 95% specificity is applied at times remote from diagnosis when the function falls below this threshold. The sensitivity function for HM was derived by doubling both the sensitivity and lead-time of CA125. The 95% CI for each piece of the CA125 sensitivity function is estimated using the exact binomial method.

Figure 2. Cost-effectiveness of screening using hypothetical screening tests by screening test cost*

*Hypothetical screening test costs are presented as fold cost relative to the cost of CA125 ($31) and TVS ($111). $210 and $750 correspond to the base-case assumptions for the cost associated with HM and HI respectively.

Figure 3. Cost-effectiveness of ovarian cancer screening strategies in populations at increased risk*

*We evaluated the cost-effectiveness of all four bi-modal screening strategies when applied to populations with increased ovarian cancer incidence rates of 2X, 4X, and 8X relative to the general population.

Figure 4. Impact of screening frequency on mortality reduction and cost-effectiveness of ovarian cancer screening strategies*

*Mortality reduction and cost-effectiveness for screening intervals of 6, 18, and 24 months are shown in addition to the base case assumption of 12 months, for each of the four screening scenarios. All other parameters were held constant.