DYNAMIC CIRCADIAN-ADJUSTED DPYD PHENOTYPING FOR FLUOROPYRIMIDINE THERAPY: AN IN SILICO COST-EFFECTIVENESS ANALYSIS FOR PRECISION ONCOLOGY
Author(s)
Jashuva T, PharmD1, Manoj Kumar Mudigubba, MPH, PharmD, PhD2.
1Department of Pharmacy Practice, Raghavendra Institute of Pharmaceutical Education & Research (RIPER), Anantapur, India, 2Department of Pharmacy Practice, Raghavendra Institute of Pharmaceutical Education and research, Anantapur, India.
1Department of Pharmacy Practice, Raghavendra Institute of Pharmaceutical Education & Research (RIPER), Anantapur, India, 2Department of Pharmacy Practice, Raghavendra Institute of Pharmaceutical Education and research, Anantapur, India.
OBJECTIVES: CPIC-standard DPYD germline genotyping assigns static metabolizer phenotypes, ignoring circadian variation in dihydropyrimidine dehydrogenase (DPD) enzymatic activity. This in silico study estimates the pharmacoeconomic significance of circadian-adjusted DPYD phenotyping on metabolizer reclassification and evaluates its cost-effectiveness from a European payer perspective.
METHODS: A three-arm cascade state-transition model compared empiric dosing (Arm A), static CPIC germline DPYD genotyping of four Level A variants (Arm B), and circadian-adjusted dynamic phenotyping (Arm C). Germline variant prevalence and activity scores were sourced from population pharmacogenomic databases; TCGA-COAD/READ (n=382) contributed clinical outcomes only, with germline pharmacogenomics, tumour genomics, and gene expression as distinct inputs. Circadian DPYD oscillation was established via colocalization of Genotype-Tissue Expression (GTEx v8) hepatic expression quantitative trait loci (eQTL) with CircaDB rhythmicity scores (PP4=0.83). DPD enzyme activity was modelled through sinusoidal expression-to-activity conversion, adjusting CPIC activity scores hourly across Zeitgeber time (ZT0-ZT23); patients falling below the normal/intermediate metabolizer (NM/IM) threshold were reclassified. Health states: full-dose treatment, dose reduction, Grade 3-4 toxicity, and death. Five-year horizon; 3.5% discounting; willingness-to-pay €30,000/QALY; electronic decision support (≤€50/patient; no additional laboratory testing). Uncertainty: PSA (10,000 iterations), one-way deterministic sensitivity, and threshold analysis.
RESULTS: Hepatic DPYD eQTL colocalization (PP4=0.83) suggested circadian-phase-dependent expression with a 2.4-fold amplitude oscillation. Circadian phenotyping predicted 8-12 NM-to-IM reclassifications per 100 patients, estimating 1.7-2.4 Grade 3-4 toxicity events prevented and €3,900-5,600 in hospitalization costs averted versus Arm B. ICER of Arm C versus Arm B was approximately €15,278/QALY. PSA supported cost-effectiveness at €30,000/QALY in 89.2% of iterations (dominant in 41.3%). Threshold analysis indicated cost-effectiveness when reclassification exceeded 4 per 100 patients.
CONCLUSIONS: These in silico findings suggest circadian-informed DPYD phenotyping may improve healthcare resource utilization relative to static testing. The estimated ICER supports prospective validation and health technology assessment of precision oncology programmes. Decision support without additional laboratory requirements may reduce adoption barriers for value-based reimbursement.
METHODS: A three-arm cascade state-transition model compared empiric dosing (Arm A), static CPIC germline DPYD genotyping of four Level A variants (Arm B), and circadian-adjusted dynamic phenotyping (Arm C). Germline variant prevalence and activity scores were sourced from population pharmacogenomic databases; TCGA-COAD/READ (n=382) contributed clinical outcomes only, with germline pharmacogenomics, tumour genomics, and gene expression as distinct inputs. Circadian DPYD oscillation was established via colocalization of Genotype-Tissue Expression (GTEx v8) hepatic expression quantitative trait loci (eQTL) with CircaDB rhythmicity scores (PP4=0.83). DPD enzyme activity was modelled through sinusoidal expression-to-activity conversion, adjusting CPIC activity scores hourly across Zeitgeber time (ZT0-ZT23); patients falling below the normal/intermediate metabolizer (NM/IM) threshold were reclassified. Health states: full-dose treatment, dose reduction, Grade 3-4 toxicity, and death. Five-year horizon; 3.5% discounting; willingness-to-pay €30,000/QALY; electronic decision support (≤€50/patient; no additional laboratory testing). Uncertainty: PSA (10,000 iterations), one-way deterministic sensitivity, and threshold analysis.
RESULTS: Hepatic DPYD eQTL colocalization (PP4=0.83) suggested circadian-phase-dependent expression with a 2.4-fold amplitude oscillation. Circadian phenotyping predicted 8-12 NM-to-IM reclassifications per 100 patients, estimating 1.7-2.4 Grade 3-4 toxicity events prevented and €3,900-5,600 in hospitalization costs averted versus Arm B. ICER of Arm C versus Arm B was approximately €15,278/QALY. PSA supported cost-effectiveness at €30,000/QALY in 89.2% of iterations (dominant in 41.3%). Threshold analysis indicated cost-effectiveness when reclassification exceeded 4 per 100 patients.
CONCLUSIONS: These in silico findings suggest circadian-informed DPYD phenotyping may improve healthcare resource utilization relative to static testing. The estimated ICER supports prospective validation and health technology assessment of precision oncology programmes. Decision support without additional laboratory requirements may reduce adoption barriers for value-based reimbursement.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
MSR199
Topic
Economic Evaluation, Health Technology Assessment, Methodological & Statistical Research
Disease
Gastrointestinal Disorders, Genetic, Regenerative & Curative Therapies, No Additional Disease & Conditions/Specialized Treatment Areas, Oncology, Personalized & Precision Medicine