Analysis of Within-Subject Variation of Caffeine Metabolism When Used to Determine Cytochrome P4501A2 and N-Acetyltransferase-2

The following comments pertain exclusively to the determination of cytochrome CYP1A2 and not to that of N-acetyltransferase 2. With the help of excellent laboratory work and a logically designed experimental procedure, the authors show the level of reproducibility of two caffeine metabolic ratios by repeat investigations in a number of subjects. Unfortunately, this effort is not very useful because the main concern is the interpretation of these metabolic ratios in terms of CYP1A2 activity. We are worried that the widespread adoption of the recommended methods in cancer research would lead to results that are not much better than half truths (see correlation coefficient below). This assessment also fits to the lack of good reproducibility that is shown by McQuilkin et al. The basis of our concern has been published in specific terms by Tang et al. (1) in a paper not quoted here, and in general terms in a review paper by Kalow and Tang (2), chided by McQuilkin, et al. for not quoting detector sensitivities and other details. Briefly, the main trouble with the Butler methods tested by McQuilkin et al. is the urine-flowdependent excretion of caffeine so that the interpretation of Butler's ratio in metabolic terms introduces an error component. Second, the timing of urine collection (4-5 h) is not necessarily proper for all subjects. After population screening, the Butler methods led to claims for bior trimodal frequency distribution of CYP1A2 activity, whereas all other methods using caffeine for CYP1A2 determination lead to normal distribution curves within nonsmokers and within groups of heavy smokers, a serious discrepancy for cancer studies. Systemic caffeine clearance is the best known measurement of CYP1A2 activity in humans. The Butler test correlated with systemic caffeine clearance with a correlation coefficient of only r = 0.46 (1). In principle, CYP1A2 assays based on blood tests are best, saliva tests are good if properly done, and urine tests are potentially useful.