Optimal Genetic Screening for Cystic Fibrosis

Cystic fibrosis (CF) is a life-threatening genetic disorder. Early treatment of CF-positive newborns can extend life span, improve quality of life, and reduce healthcare expenditures. As a result, newborns are screened for CF throughout the United States. Genetic testing is costly; therefore, CF screening processes start with a relatively inexpensive but not highly accurate biomarker test. Newborns with elevated biomarker levels are further screened via genetic testing for a panel of variants (types of mutations), selected from among hundreds of CF-causing variants, and newborns with mutations detected are referred for diagnostic testing, which corrects any false-positive screening results. Conversely, a false negative represents a missed CF diagnosis and delayed treatment. Therefore, an important decision is which CF-causing variants to include in the genetic testing panel so as to reduce the probability of a false negative under a testing budget that limits the number of variants in the panel. We develop novel deterministic and robust optimization models and identify key structural properties of optimal genetic testing panels. These properties lead to efficient, exact algorithms and key insights. Our case study underscores the value of our optimization-based approaches for CF newborn screening compared with current practices. Our findings have important implications for public policy.