Temporal variability of caffeine and caffeine metabolite concentrations in spot urine samples in 50 North Carolina adults over a six-week period (ICoFF2019)
On this page:
Objectives: Studies of caffeine as a risk factor in human health often rely upon dietary intake data to assess exposure. Variability associated with major dietary sources (e.g. coffee) and capturing intake from emerging sources (e.g. energy drinks) presents challenges. Spot urine concentrations of caffeine and select metabolites are associated with caffeine intake and may be suitable as intake biomarkers, but a better understanding of the temporal variability of these measurements is needed. The US EPA’s Pilot Study to Estimate Human Exposures to Pyrethroids Using an Exposure Reconstruction Approach (Ex-R Study) investigated the exposure of 50 adults (19–50 years old) in North Carolina, USA to pyrethroid insecticides over a 6-week period in 2009–2011. Designed to assess the temporal variability of pesticide exposure biomarkers, the urine sampling of the Ex-R Study holds the potential for gaining an understanding of repeat measures performed on any urinary compound. Our objective was to describe the temporal concentration variability of urine caffeine and its metabolites over periods ranging from 24 h to 6 weeks by use of the Ex-R Study.
Methods: We measured spot urine concentrations of caffeine and 14 metabolites by use of LC-MS/MS in 2292 archived samples from the Ex-R Study. Spot urine samples collected over the 6-week study were classified as bedtime voids, first-morning voids, or individual voids obtained over a continuous 24-h period. We reported our results by concentration, specific gravity (SG) and creatinine (CR) adjusted concentrations, and excretion rate. In addition to descriptive statistics, we calculated between-person (bR ̂0.95) and within-person (wR ̂0.95) 95% fold ranges estimates, and intra-class correlation coefficients (ICC). We used ICCs to determine whether a single spot urine sample was sufficient to meet a specified degree of reliability (ICC ≥0.80) over a given timeframe.
Results: We were able to quantify 14 of the 15 target analytes in ≥99% of the urine samples (LODs: 3–100 nmol/L). We found that estimates of bR ̂0.95 generally exceeded those of wR ̂0.95 for all analytes across all concentration adjustments and timeframes examined. The impact of normalizing for SG, CR, or excretion rate appeared to be mixed; in most cases bR ̂0.95 increased across all analytes, whereas wR ̂0.95 decreased the most in later products of caffeine metabolism (e.g. methyl and dimethyl uric acids). We noted that ICCs were typically highest for samples collected over the shortest timeframe (i.e. individual voids in each of the 24-h collection periods), were higher for caffeine metabolites produced via paraxanthine or theophylline versus theobromine, and increased in cases where normalization decreased wR ̂0.95. ICCs were ≥0.80 for CR- and SG-adjusted theophylline, 1,3- and 1,7-dimethyluric acid, 1-methylxanthine, 1-methyluric acid, and 5-acetylamino-6-amino-3-methyluracil when calculated for individual voids across all six 24-h collection periods. The highest ICCs determined across the entire 6-week period were for CR adjusted 1-methyluric acid in bedtime voids (0.77) or 24-h (0.76) samples.
Conclusion: Based upon the temporal variability we observed in the Ex-R Study, it may be possible to reliably estimate (ICC ≥0.80) urine concentrations of several caffeine metabolites over a given 24-h period by use of a single spot urine sample and SG- or CR adjustment, with more samples needed for longer timeframes.