Premium
Use of Spot Urine Caffeine and Caffeine Metabolite Concentrations for Distinguishing a Recommended Upper Limit of Caffeine Intake
Author(s) -
Rybak Michael E.,
Sternberg Maya R.,
Pao ChingI
Publication year - 2017
Publication title -
the faseb journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.709
H-Index - 277
eISSN - 1530-6860
pISSN - 0892-6638
DOI - 10.1096/fasebj.31.1_supplement.646.8
Subject(s) - caffeine , receiver operating characteristic , theophylline , paraxanthine , urine , metabolite , population , area under the curve , cutoff , chemistry , analyte , medicine , metabolism , chromatography , cyp1a2 , physics , environmental health , cytochrome p450 , quantum mechanics
Urine excretion of caffeine and select caffeine metabolites is common in the US population and associated with caffeine intake. In particular, caffeine metabolites resulting from metabolism via paraxanthine and/or theophylline show promise as potential caffeine intake biomarkers. The 2015–2020 Dietary Guidelines for Americans state that up to 400 mg/day of caffeine can be incorporated into a healthy eating style. In this study we performed receiver operating characteristic (ROC) analysis of unweighted data from the NHANES 2009–2010 to determine optimal cutoffs for spot urine caffeine and 14 caffeine metabolites that would maximize the correct classification of individuals with 24‐h caffeine intake >400 mg/d vs. ≤400 mg/d (from diet and supplements). We limited our study sample to subjects ≥12 y of age and excluded individuals taking prescription medications known to interfere with caffeine metabolism. For each analyte we calculated an unweighted ROC estimate for the optimal cutoff concentration, as well as estimates of sensitivity, specificity, area under the ROC curve (AUC), and an overall rate of correct classification. We calculated these estimates for the total study population, as well as stratified by age, sex and race‐ethnicity. Using our ROC estimates for optimal cutoff concentrations, we observed the highest overall rate of correct classification (72.2%) with 1,3‐dimethyluric acid (13U) (cutoff: 12.4 μmol/L; sensitivity: 69.7%; specificity: 72.2%). In addition to 13U, we also saw correct classification rates >70% for 5‐acetylamino‐6‐amino‐3‐methyluracil (AAMU) (cutoff: 98.4 μmol/L; sensitivity: 69.0%; specificity: 71.4%) and 1‐methyluric acid (1U) (cutoff: 98.1 μmol/L; sensitivity: 71.7%; specificity: 70.8%). The greatest AUC was observed for 1U (0.776; 95% CI: 0.740–0.811), with similarly high AUCs observed for AAMU (0.763; 95% CI: 0.726–0.800), 13U (0.776; 95% CI: 0.726–0.799), and 1‐methylxanthine (1X) (0.768; 95% CI: 0.733–0.804) which had correct classification rate of 68.5% (cutoff: 45.3 μmol/L; sensitivity: 0.724; specificity: 0.682). Theobromine (cutoff: 16.3 μmol/L; sensitivity: 69.7%; specificity: 49.4%) had the lowest overall rate of correct classification (50.9%), as well as the lowest AUC (0.603; 95% CI: 0.563–0.642). Depending on the analyte, we noted that the ROC estimate for optimal cutoff concentration could vary by age, sex, or race‐ethnicity. We believe that our study is the first report of using ROC analysis to determine cutoff values for spot urine caffeine and caffeine metabolite concentrations to classify individuals based on their caffeine intake, and that this approach holds promise as a means of identifying individuals whose recent caffeine intake may have exceeded a recommended upper limit of 400 mg/day.