(S050) Cost-Effectiveness of Neurocognitive Preservation for Whole-Brain Radiotherapy

April 30, 2015

Neurocognitive-preservation strategies for WBRT are not cost-effective from a societal perspective, though they may be effective for patients who are cognitively intact at baseline and have a long expected survival.

Ben Y. Durkee, MD, PhD, Joseph Sanford, MD, Anna Oh, PhD, Daniel Slate, BS, Brandon Turner, BS, Erqi L. Pollom, MD, Iris C. Gibbs, MD, Scott G. Soltys, MD; Department of Radiation, Department of Radiation Oncology, Stanford University; Department of Nursing, University of California, San Francisco; Stanford University Graduate School of Business; Stanford University School of Medicine

BACKGROUND: Whole-brain radiotherapy (WBRT) is the standard of care for nonsurgical intracranial metastatic disease. Patients receiving WBRT are at risk for neurocognitive degeneration, which is weighed against the neurocognitive detriment of tumor progression. Recent results from RTOG 0614 and RTOG 0933 have shown cognitive benefit with memantine (WBRT+M) or hippocampal avoidance (HA-WBRT) [Li, JCO 2007; Brown, Neuro Onc 2013; Gondi, JCO 2014].

Neurocognitive compromise is associated with lower quality of life (QoL). Utility scores for these health states are well described and reproducible. The cost of these cognitive preservation strategies must be weighed against the gains in QoL during the patients’ final months.

METHODS: We created a decision tree with an integrated three-state Markov model. Treatment strategies included best supportive care (BSC), WBRT, WBRT+M, and HA-WBRT. Health states were cognitively intact, cognitively impaired, and dead. Cycle-specific probabilities for cognitive states were derived from PCI-P120-9801, RTOG 0614, and RTOG 0933. Survival data were derived from recursive partitioning analysis (RPA) of three RTOG trials [Gaspar, IJROBP 1997].

Utility scores for the base case of a patient with metastatic cancer were derived from the available literature. The effect of cognitive detriment was extrapolated from patients with mild dementia, who scored similarly on a Hopkins Verbal Learning Test. The model was run using each RPA class (I/II/III) as the base case scenario.

The analysis was done from a societal perspective, with a single payer system. Threshold for cost-effectiveness was set at a willingness to pay (WTP) of $100,000 per quality-adjusted life-year (QALY). Costs were derived from the Medicare Physician Fee Schedule. We performed multiway sensitivity analyses to address uncertainties in cost, utility scores, efficacy of intervention, life expectancy, and distribution of baseline cognitive states.

RESULTS: BSC is the dominant strategy for the base case scenario. No strategy for neurocognitive preservation is cost-effective at a WTP of $100,000/QALY. Neurocognitive-preservation strategies become cost-effective in the theoretical cohort of patients with perfect baseline cognition and long expected survival (> 14 mo for base case distribution; > 10 mo for perfect baseline cognition). The model was sensitive to assumptions about the initial distribution of patients’ cognitive states and cost. It was minimally sensitive to utility scores and efficacy of the intervention.

CONCLUSION: Neurocognitive-preservation strategies for WBRT are not cost-effective from a societal perspective, though they may be effective for patients who are cognitively intact at baseline and have a long expected survival.

Proceedings of the 97th Annual Meeting of the American Radium Society - americanradiumsociety.org