The cost-effectiveness of screening lung cancer patients for targeted drug sensitivity markers

BACKGROUND: New oncology drugs are being developed in conjunction with companion diagnostics with approval restricting their use to certain biomarker-positive subgroups. We examined the impact of different predictive biomarker screening techniques and population enrichment criteria on the cost-effectiveness of targeted drugs in lung cancer, using ALK and crizotinib to build the initial model. METHODS: Health economic modeling of cost per Quality Adjusted Life Year was based on literature review and expert opinion. The modeled population represented advanced non-small cell lung cancer (NSCLC), eligible for predictive biomarker screening with prescribing restricted to biomarker-positive patients. RESULTS: For assays costing $1400 per person, cost per quality-adjusted life year (QALY) gained for ALK screening all advanced NSCLC, excluding treatment cost, is $106 707. This falls to $4756 when only a highly enriched population is screened (increasing biomarker frequency from 1.6 to 35.9%). However, the same enrichment involves missing 56% patients who segregate within the unscreened group. Cheaper screening tests that miss some true positives can be more cost-effective if proportional reductions in cost exceed proportion of subjects missed. Generic modeling of idealised screening assays, including treatment cost, reveals a dominant effect of screening cost per person at low biomarker frequencies. Cost-effectiveness of <$100 000 per QALY gained is not achievable at biomarker frequencies <5% (with drug costs $1-5000 per month and screening costs $600-1400 per person). INTERPRETATION: Cost-effectiveness of oncology drugs whose prescribing is restricted to biomarker-positive subgroups should address the cost of detecting marker-positive patients. The cost of screening dominates at low frequencies and strategies to improve cost-effectiveness based on the assay cost, drug cost and the group screened should be considered in these scenarios. British Journal of Cancer (2012) 106, 1100-1106. doi:10.1038/bjc.2012.60 www.bjcancer.com Published online 28 February 2012 (C) 2012 Cancer Research UK