Search tips
Search criteria 


Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
Cancer Epidemiol Biomarkers Prev. Author manuscript; available in PMC 2010 November 1.
Published in final edited form as:
PMCID: PMC2784188

CYP1A2, GSTM1, and GSTT1 polymorphisms and diet effects on CYP1A2 activity in a crossover feeding trial*


Cytochrome P-450 1A2 (CYP1A2) is a biotransformation enzyme that activates several procarcinogens. CYP1A2 is induced by cruciferous and inhibited by apiaceous vegetable intake. Using a randomized, cross-over feeding trial in humans, we investigated dose effects of cruciferous vegetables and effects of any interaction between cruciferous and apiaceous vegetables on CYP1A2 activity. We also investigated whether response varied by CYP1A2*1F, GSTM1, and GSTT1 genotypes (glutathione S-transferases that metabolize crucifer constituents) and whether CYP1A2 activity rebounds after apiaceous vegetables are removed from the diet. Participants (N = 73), recruited based on genotypes, consumed four diets for two weeks each: low-phytochemical diet (basal), basal plus single dose of cruciferous (1C), basal plus double dose of cruciferous (2C), and basal plus single dose of cruciferous and apiaceous vegetables (1C+A). CYP1A2 activity was determined by urine caffeine tests administered at baseline and the end of each feeding period. Compared with basal diet, the 1C diet increased CYP1A2 activity (P < 0.0001) and the 2C diet resulted in further increases (P < 0.0001) with men experiencing greater dose-response than women. The 1C+A diet decreased CYP1A2 activity compared to the 1C and 2C diets (P < 0.0001 for both). Although there was no overall effect of CYP1A2*1F or GSTM1-null/GSTT1-null genotypes or genotype-by-diet interactions, there were significant diet response differences within each genotype. Additionally, CYP1A2 activity recovered modestly one day after the removal of apiaceous vegetables. These results suggest complex interactions among dietary patterns, genetic variation, and modulation of biotransformation that may not be apparent in observational studies.

Keywords: Cruciferous, apiaceous, CYP1A2, GST, genotype


Cytochrome P-450 1A2 (CYP1A2) is a biotransformation enzyme that mediates the activation of several procarcinogens (13). Due to the major role of biotransformation enzymes in the metabolism of carcinogens and other xenobiotics, variation in biotransformation-enzyme activity may influence carcinogenesis. CYP1A2 activity in humans is influenced by genetic variation and environmental factors resulting in wide interindividual variability (reviewed in (4)). CYP1A2*1F (rs762551) is a polymorphism in intron 1 (−164A>C) that confers decreased inducibility (5), thus variation in CYP1A2 activity by CYP1A2*1F genotype may be most pronounced in the presence of inducers.

Cruciferous (i.e., broccoli-family) vegetable consumption increases CYP1A2 activity in humans (6, 7). Isothiocyanates and indoles (both metabolites of glucosinolates in cruciferous vegetables) have been evaluated for CYP1A2 induction in many test systems. The results have been inconsistent. However, in primary human-derived test systems, phenethyl isothiocyanate (PEITC) and the indole derivative 3,3′-diindolylmethane consistently induce CYP1A2 (8, 9), whereas sulforaphane, another isothiocyanate, has no effect (8).

Isothiocyanate metabolism involves conjugation with glutathione by glutathione-S-transferases M1 and T1 (GSTM1 and GSTT1) (10). In Caucasians, approximately 50% and 20% have null genotypes for GSTM1 and GSTT1, respectively, resulting in a lack of the enzymes; hence GSTM1-null and GSTT1-null individuals have been hypothesized to experience a longer exposure to glucosinolate metabolites due to slower metabolism (11, 12). Epidemiologic studies in Asia and Europe show that the chemopreventive association of cruciferous intake is more pronounced among individuals with the GSTM1/GSTT1 null genotypes (1214), but studies in the US show the opposite relationship (15, 16) or no relationship (17, 18). Recent feeding studies with broccoli indicate that individuals with the null genotypes had greater isothiocyanate excretion, supporting the US findings that GST-positive individuals have greater protection (19, 20) but not supporting the hypothesis that null genotype individuals have delayed clearance. Moreover, indole metabolism may involve glucuronidation (21), a process mediated by isothiocyanate- and indole-inducible UDP-glucuronosyl transferases, further suggesting a potential influence on CYP1A2 activity by GSTM1 and GSTT1 genotype.

Intake of apiaceous (i.e., carrot-family) vegetables decreases CYP1A2 activity in humans (6). Apiaceous vegetables are rich in furanocoumarins (22, 23) which inhibit human CYP1A2 in vitro and in vivo (2426). Furanocoumarins undergo extensive first-pass metabolism (27, 28); within 24-hours of ingestion, 95% of two predominant furanocoumarins are excreted as a series of glucuronic acid conjugates in the urine (28).

Our primary objectives were: 1) to test whether there is a dose-response effect of cruciferous vegetables on CYP1A2 activity; 2) to test whether concurrent consumption of cruciferous and apiaceous vegetables reduces CYP1A2 activity and 3) to establish the effect of CYP1A2*1F genotype on induction or inhibition of CYP1A2 activity in response to vegetable supplementation. Our secondary aims were to measure the combined effect of GSTM1 and GSTT1 genotypes on CYP1A2 activity and to test whether CYP1A2 activity rebounds after apiaceous vegetables are removed from a diet with concurrent exposure to CYP1A2 inducers.

Subjects and Methods


Participants were recruited from the Seattle area through media advertisements, flyers, and direct mailings. Subjects were 20–40 years old, healthy, and nonsmokers. Initial eligibility was determined by a self-administered screening survey; exclusion criteria were: history of gastrointestinal, hepatic, or renal disorders; pregnancy or lactation; allergies or intolerances to any food used in the feeding trial; weight change greater than 4.5 kg within the past year; major changes in eating habits within the past year; antibiotic use within the past 3 months; body weight greater than 150% of desirable; exercise patterns that result in major changes in diet; current use of prescription medication, including contraception; current use of over-the-counter medications; regular exposure to passive or second-hand smoke; occupational exposure to smoke or organic solvents; dislike of any food in the trial that would preclude participation; alcohol intake greater than 2 drinks/day (2 drinks is equivalent to 720 mL beer, 240 mL wine, or 90 mL spirits); no interest in participating in a controlled feeding trial. Final eligibility was based on CYP1A2*1F, GSTM1, and GSTT1 genotypes. The goal was to recruit participants into a 3×3 design according to the three groups of CYP1A2*1F genotypes (A/A, A/C, and C/C) and three groups of GST genotypes (GSTM1+/GSTT1+, GSTM1-null/GSTT1+, and GSTM1-null/GSTT1-null), with an equal number of men and women in each group. We randomized 73 individuals: 14 A/A men, 17 A/A women, 15 A/C men, 15 A/C women, 6 C/C men, and 6 C/C women. Of the 73, 14 men were GSTM1+/GSTT1+, 12 women were GSTM1+/GSTT1+, 16 men were GSTM1-null/GSTT1+, 15 women were GSTM1-null/GSTT1+, 5 men were GSTM1-null/GSTT1-null, 9 women were GSTM1-null/GSTT1-null, and 2 women were GSTM1+/GSTT1-null (recruited for their CYP1A2 genotype only). Nine initially randomized participants dropped out of the study after baseline data were collected but before completing one feeding period. New recruits were selected and randomized into the appropriate treatment orders to maintain the initial blocks. The study activities were approved by the Institutional Review Board at the Fred Hutchinson Cancer Research Center. Informed written consent was obtained from all participants prior to the start of the study.

Determination of genotypes

After initial eligibility was determined, participants were mailed buccal-cell-collection kits and genotypes were determined as buccal cells and consent for genotyping were returned. Buccal-cell DNA was extracted using the QIAamp DNA Blood Midi kit (Qiagen, Valencia, CA). The CYP1A2*1F polymorphism was analyzed using a PCR/RFLP method. At pH 8.3, a 20 μL reaction contained 100 ng genomic DNA, 10 mM Tris HCL, 1.25 mM MgCl2, 50 mM KCl, 200 μM dNTPs, 1.25 U Taq DNA polymerase, and 0.2 μM each primer (5′ CAA CCC TGC CAA TCT CAA GCA C 3′ and 5′ AGA AGC TCT GTG GCC GAG AAG G 3′). Thermal cycling included an initial denaturation at 95°C for 3 min, followed by 40 cycles of 94°C for 1 min, 62°C for 30 sec, 72°C for 1 min, and 1 cycle of 72°C for 5 min. The amplicon was digested with 5 Units Apa I for 2 hours and resolved by a 3% agarose gel. The wildtype A allele was cleaved into 708 and 212 bp fragments, whereas the variant C allele remained intact. GSTM1 and GSTT1 genotyping were conducted using PCR as previously described (29).

Study design

Before starting the study, participants completed a 4-day baseline study wherein they followed their habitual diet, kept 3-day food records (instructions provided by a dietitian), completed a 24-hour urine collection from day 3 to day 4, and came to the center on day 4 for the first urine caffeine test (a measure of CYP1A2 activity). After the baseline study, each participant received all 4 experimental diets using a crossover design in a randomly assigned Latin Square sequence. Randomization of participants was blocked on CYP1A2*1F-by-sex and GSTM1/GSTT1-by-sex. Each diet was consumed for 2 weeks; the menu was a 3-day dinner rotation with the same breakfast and lunch daily to minimize variability. Diet periods began on Sundays with dinner and ended on Saturdays (Day 14) with breakfast and measurement of CYP1A2 activity. There was a minimum 3-week wash-out period in between each 2-week feeding period; participants followed their habitual dietary patterns during the wash-out period.

Study diets

Participants received four controlled diets and were assessed for adherence as described previously (29). Briefly, the diets were: a basal diet low in phytochemicals (devoid of fruits and vegetables), basal plus ~7g/kg body weight of cruciferous vegetables (1C), basal plus ~14 g/kg body weight of cruciferous vegetables (2C), and basal plus ~7 g/kg body weight of cruciferous vegetables and ~4 g/kg body weight of apiaceous vegetables (1C+A). Vegetable doses were based on our previous data (6), but were modified to provide dosing by body weight. The cruciferous vegetables used were frozen broccoli and cauliflower, fresh daikon radish sprouts, and raw shredded cabbage (red and green). The apiaceous vegetables used were frozen carrots and fresh celery, dill, parsley, and parsnips. All frozen vegetables and other prepackaged foods were purchased in case lots to minimize variability. Fresh foods were purchased from the same local vendor. One week prior to and during each feeding period, subjects were asked to avoid the following: alcoholic beverages, prescription and over-the-counter medications, and nutritional supplements (including herbal products).

To test for rebound in CYP1A2 activity after vegetables were removed from the diet, 36 participants received 1 day of the basal diet added to the end of the 1C diet period and to the 1C+A diet period. Thus, participants were exposed to phytochemicals for 13 consecutive days followed by 1 day of foods low in phytochemicals with CYP1A2 activity measured on Day 14 and Day 15.

Energy needs for each participant were assessed by comparing the results of the Harris-Benedict equation (30), Mifflin - St. Jeor equation (31), and 3-day food records. Physical activity, using metabolic equivalent intensity levels (32), was considered during assessment of calorie needs and was based on a 1-week activity log that participants turned in on Day 4 of the baseline study. Body weight was monitored throughout each diet period; daily energy intakes were adjusted accordingly using basal diet foods in order to maintain weight.

Urine collection and analysis

CYP1A2 is the primary enzyme involved in the first two steps of caffeine metabolism; hence, CYP1A2 activity is typically measured by urine caffeine metabolite ratios (5, 33). Thus, on Day 7 and Day 14 (and on Day 15 of the 1C and the 1C+A diet periods for 36 individuals), participants completed a urine caffeine test for the measurement of CYP1A2 activity. Arriving at the center in the morning after an overnight 10-hour fast, they consumed 200 mg caffeine (NoDoz; Bristol-Meyers Squibb, NY) with 30 mL water. Participants ate a light breakfast (42–64 g ready-to-eat cereal, 240 mL milk, 122 g of a commercial fruit drink made with 10% fruit juice) and remained at the center for the next 5 hours, during which additional food and fluids were withheld. At hour 4, participants were asked to void and then at hour 5 to void again and collect their urine. The total volume and pH of each urine sample was measured. The pH was adjusted to 3.5 with 1 M HCl and 1.8 mL aliquots were stored at −70°C until time of analysis.

Acidified urine samples were thawed and centrifuged prior to analysis. Three hundred microliters ammonium sulfate was added to 500 μL of centrifuged urine containing 50 μg/mL of internal standard, 4-acetamidophenol (APAP). The samples were vortexed and sat for 1 hour at room temperature prior to extraction to assure complete dissolution of the ammonium sulfate. The urines were extracted with 2 × 3.0 mL chloroform: isopropanol, 9:1. After each addition of the extraction solvent, the samples were vortexed for 1 min and centrifuged for 15 minutes at 540 × g. The organic layers from both extractions were combined and dried under nitrogen in a 45°C water bath. The residue was reconstituted with 500 μL of 0.05% acetic acid, sonicated for 30 seconds, and vortexed. This solution was filtered through a 0.45 μm Whatman mini-Uni-prep filter device (Whatman Inc., Clinton, NJ) and placed directly into the sampling tray of the HPLC for analysis.

Samples were analyzed for the following caffeine metabolites using an Agilent Model 1100 series HPLC (Agilent Technologies, Wilmington, Del.): 5-acetylamino 6-formylamino 3-methyluric acid (AFMU), 7-methylxanthine (7X), 1-methyluric acid (1U), 3-methylxanthine (3X), 1-methylxanthine (1X), 1,3-dimethyluric acid (13U), 3,7-dimethylxanthine (37X), 1,7-dimethyluric acid (17U), 1,7-dimethylxanthine (17X), 1,3-dimethylxanthine (13X), 1,3,7-trimethylxanthine (137X, caffeine), with APAP as an internal standard. The purified standard materials were all purchased from Sigma-Aldrich (St Louis, MO) with the exception of AFMU which was purchased from Toronto Research Chemicals Inc. (North York, Ontario, Canada) and 17U which was purchased from Fluka Chemical Corp. (Ronkonkoma, NY). Urine samples were batched to include all samples for one study subject in the same HPLC run with quality-control samples included at the beginning and end of each batch. A seven-point calibration curve over a range of 0.25 to 50 mg/mL for each metabolite was analyzed with each batch of samples. Correlation coefficients for the standard curves were 0.99 or higher. Separation was done on a 5 μm Ultrasphere ODS C18 column 4.6×250mm (Beckman, Fullerton, CA) using a gradient elution. The mobile phase consisted of methanol (A), acetonitrile (B), and 0.05% acetic acid (D) with the following elution profile: 0–14 minutes 8% A, 92% D; 14–20 minutes ramp from 8% A to 19% A and 92% D to 81% D, hold for 3 minutes; 23–30 minutes 10% A, 8% B, 82% D; 30 to 36 minutes ramp from 8% B to 50% B and from 10% A to 50% A, hold for 5 minutes; 41–42 minutes return to starting conditions 8% A, 92% D and hold to the end of the run at 52 minutes. The flow rate was 1.1 mL per minute throughout the entire run. Caffeine and its metabolites were identified based on retention times of the primary standards, peak purity, and spectral analysis. Inter-assay precision (% CV) for the quality-control samples was as follows: AFMU 6.8%, 1U 15.6%, 1X 3.7%, 3X 12.2%, 7X 6.4%, 13U 15.1%, 13X 10.7%, 17X 3.3%, 17U 4.0%, 37X 8.1%, and 137X 4.6%. Using peak heights, metabolites were quantitated based on standard curves.

Several different caffeine metabolite ratios are cited in the literature for assessment of CYP1A2 activity (5, 6, 33, 34). We used the ratio (1U + 1X + AFMU)/17U because it is the least influenced by urine flow and most sensitive to changes in CYP1A2 activity (3537).

Additionally, participants collected their urine for 24 hours starting on Day 13 for assessment of urinary isothiocyanates (29, 38). Urinary isothiocyanate excretion was used as one method for confirming dietary compliance.

Statistical analysis

Differences in CYP1A2 activity were analyzed using a linear mixed regression model (39) (PROC MIXED in SAS, Release 9.0). Study participant was a random effect with the following as fixed effects in the model: diet, CYP1A2*1F genotype (A/A, A/C, or C/C), GSTM1 genotype (null/non-null), GSTT1 genotype (null/non-null), sex, period, and order. The dependent variable was the caffeine-metabolite ratio, which was log transformed to achieve normality in distribution. Our strategy for handling missing data was to: 1) analyze the available data using SAS Proc Mixed for unbalanced data; and then, 2) analyze the data using either random-coefficient selection models or random-coefficient pattern-mixture models depending on the drop-out mechanisms (reviewed in (40)). For testing of rebound in CYP1A2 activity, the dependent variable in the linear mixed regression model was the ratio of Day 15/Day 14 caffeine-metabolite ratio, which was log transformed to achieve normality in distribution. All data are presented as the log-back transformed least squares means and standard errors (SE) of basal diet, the ratio of vegetable diets to basal diet, and the ratio of vegetable diets to each other (log-back transformed the differences between the diets). There were no statistically significant differences between analyses with and without adjustment for race; hence, data are presented without adjustment for race. All analyses were 2-sided and P < 0.05 was considered statistically significant.


Prevalence of genotypes

Of those who met initial screening criteria and provided buccal cell samples for genotyping (n = 468), the overall frequency of the CYP1A2*1F allele in this population was 20.5%; prevalences of CYP1A2*1F homozygotes and heterozygotes were 10% and 41%, respectively. This is similar to the prevalence previously reported in a population of similar ethnic background (5, 41). The prevalence of the combined GSTM1-null/GSTT1-null genotypes was 11%; previously reported prevalence of the combined genotypes was 7% (among Caucasians (42)).

Of the 73 individuals who started the feeding study (Table 1), four dropped out after completing one feeding period, five after two periods, and three after three periods, resulting in 61 that completed all four feeding periods. Reasons for dropping out included: changes in employment, commitment burden of the study, diet intolerance, illness, and a move out of the area.

Table 1
Distribution of selected characteristics of study participants consuming controlled diets

Participants consumed ≥87% of the prescribed vegetables on each vegetable-supplemented diet; total urinary isothiocyanate concentrations (mean ± SD) further confirm adherence: 2.10 ± 2.5, 113.0 ± 61.7, 258.8 ± 170.6, and 106.0 ± 65.8 μmol/24 hours for basal, 1C, 2C, and 1C+A diets, respectively. The range of average daily intake of vegetables on the 1C diet was 319–662 g for men and 275–507 g for women. For the 2C diet, the range for men was 642–1329 g and 507–1011 g for women; for the 1C+A diet, 560–1065 g and 458–816 g for men and women, respectively. There were some statistically significant differences among diets for intakes of individual macronutrients (Table 2); however, these are relatively small differences as all diets provided similar total energy and proportions of macronutrients as a percentage of total energy (55% carbohydrate, 15% protein, 30% fat) because we adjusted the basal-diet components to accommodate the addition of vegetables. We made no attempt to adjust for fiber differences due to the addition of vegetables. Differing views exist on which urine caffeine metabolite ratio is the least influenced by urine flow and the most sensitive to changes in CYP1A2 activity with (1U + 1X + AFMU)/17U and (17X + 17U)/137X each proposed as the most suitable ratio 4–5 hours after caffeine administration (34, 35, 37, 43). Using the data from the 583 urine caffeine tests completed in this study, we confirmed that (1U + 1X + AFMU)/17U was the least correlated with urine flow in our study population (Pearson correlation = 0.09 (after log transformation of data); −0.50 for the ratio (17X + 17U)/137X). In addition, the direction of change in point estimates by diet were generally similar regardless of which ratio was used (data not shown).

Table 2
Averagea daily energy and macronutrient intake of 73 individuals consuming controlled diets

Diet and sex differences in CYP1A2 activity

Diet overall had statistically significant effects on CYP1A2 activity (Table 3). Compared to the basal diet, the 1C diet resulted in a 14% increase in CYP1A2 activity (P < 0.0001) and the 2C diet in a 27% increase in CYP1A2 activity. Compared to the 1C diet, the 2C diet resulted in an 11% further increase in CYP1A2 activity (P = 0.0006) and the 1C+A resulted in 13% decreased activity (P < 0.0001). The decrease in CYP1A2 activity was more pronounced when comparing the 1C+A diet with the 2C diet (22% reduction, P < 0.0001). There was no overall effect of sex, nor any differences by sex within each diet on CYP1A2 activity; however, when comparing effects of diet within sex, statistically significant diet effects were observed in one sex and not the other in some instances. Compared to the basal diet, the 1C diet induced CYP1A2 activity but men experienced only half the induction of women (Table 3). The 2C diet increased CYP1A2 activity in both men and women to the same extent compared to the basal diet, and the 1C+A diet had no effect in either sex (Table 3). Compared to the 1C diet, the 2C diet increased CYP1A2 activity further in men, but not in women, and the 1C+A diet resulted in inhibition of CYP1A2 activity in both sexes (Table 4).

Table 3
Effects of diet on CYP1A2 activity: the ratio between responses to basal and vegetable-supplemented diets
Table 4
Fold change in CYP1A2 activity: the ratio between responses to the vegetable diets

CYP1A2*1F genotype effects

There was no overall effect by CYP1A2*1F genotype and no genotype-by-diet interaction. However, when participants were stratified by CYP1A2*1F genotype, statistically significant diet effects were not always consistent across each CYP1A2*1F genotype. Compared to basal diet, the 1C diet resulted in 9% increased CYP1A2 activity in A/A individuals (P = 0.05), 13% increased activity in A/C (P = 0.008), and 20% increased activity in C/C (P = 0.008) (Table 3). Doubling the cruciferous vegetables dose resulted in a further increase in CYP1A2 activity, but only in those individuals with at least 1 A allele (Tables 3 and and4).4). Addition of apiaceous vegetables negated the inducing effect of cruciferous vegetables to a similar extent for all genotypes; CYP1A2 activity on the 1C+A diet was similar to activity on the basal diet (Table 3). Compared to the 1C diet, the 1C+A diet inhibited CYP1A2 activity in A/A (13% reduction, P = 0.0018) and C/C (16% reduction, P = 0.01) but not in A/C (8%, P = 0.528) (Table 4). Compared to the 2C diet, the 1C+A diet reduced CYP1A2 by 25%, 21%, and 18% in A/A, A/C, and C/C, respectively (P < 0.001 for each, Table 4).

GSTM1/GSTT1 genotype effects

There was no overall effect of GSTM1/GSTT1 genotype and there were no differences by genotype within each diet. However, there were two instances where the effect of diet was not consistent across all GSTM1/GSTT1 genotypes. Compared to the basal diet, the 1C diet modestly induced CYP1A2 activity by 10%, 12%, and 7% in GSTM1+/GSTT1+, GSTM1-null/GSTT1+, and GSTM1-null/GSTT1-null individuals, respectively, but reached statistical significance only for GSTM1+/GSTT1+ and GSTM1-null/GSTT1+ (P ≤ 0.02 for both, Table 3). Conversely, CYP1A2 induction by the 2C diet was statistically significant for all three genotypes: compared to the basal diet, the 2C diet resulted in a 25%, 24%, and 35% increase in CYP1A2 activity in GSTM1+/GSTT1+, GSTM1-null/GSTT1+, and GSTM1-null/GSTT1-null, respectively (P < 0.0001 for all three). Inhibition of CYP1A2 activity by the 1C+A diet compared to the 1C diet reached statistical significance in GSTM1-null/GSTT1+ and GSTM1-null/GSTT1-null only, but was statistically significant for all three genotypes when compared to the 2C diet (Table 4). The inhibition of CYP1A2 by the 1C+A diet relative to the 2C diet was the only instance where there was a statistically significant difference in response by genotype: inhibition was greater in GSTM1-null/GSTT1-null individuals than GSTM1+/GSTT1+ individuals (respectively, 33% reduction versus 18%, P = 0.02).

Rebound in CYP1A2 activity

The ratio of Day 15 CYP1A2 activity to Day 14 CYP1A2 activity was compared between the 1C diet and the 1C+A diet for 36 of the 73 participants. The ratio was higher for the 1C+A diet compared to the 1C diet (1.05 ± .01 and 1.01 ± .01, respectively, P = 0.025). CYP1A2*1F genotype did not modify the effect (P for interaction = 0.37).


This is the first investigation to our knowledge of the effects of combined intake of two botanical families with opposing effects on CYP1A2 activity. This is also the first comparison of effects on CYP1A2 activity from different doses of cruciferous vegetables. In a randomized, cross-over, controlled feeding study, we observed that combined intake of cruciferous and apiaceous vegetables reduced CYP1A2 activity relative to cruciferous intake alone. We also report a dose-response induction of CYP1A2 activity by cruciferous vegetable consumption in men but not in women. There were statistically significant diet differences by CYP1A2*1F and GSTM1/GSTT1 genotypes. Although CYP1A2 activity had not returned to baseline within 24 hrs after the removal of vegetables on the 1C+A diet, there was a statistically significant change indicating activity was recovering.

The ability of apiaceous vegetables to achieve a net inhibitory effect on CYP1A2 despite concurrent consumption of cruciferous vegetables implies a potential chemopreventive mechanism explained as follows: in addition to inducing CYP1A2, constituents of cruciferous vegetables also induce GST and other conjugating enzymes (44) which can detoxify many carcinogens, including reactive intermediates formed by CYP1A2. Because apiaceous vegetables do not induce GSTs (44) but inhibit CYP1A2, the combined intake of both botanical families may modulate biotransformation towards greater detoxification and less activation of procarcinogens. This combined effect warrants further investigation.

The finding that men experienced only half the induction of women of CYP1A2 on the 1C diet and that only men experienced further induction with the 2C diet compared to the 1C diet, suggests that men may require more cruciferous vegetables than women to induce CYP1A2 activity, given that activity on the basal diet was similar. All participants received vegetables based on the same dose per kilogram of body weight, to allow more effective investigation of sex differences. Limited data suggest that estrogen may induce expression of CYP1A2 (45); however, a review of pharmacokinetic studies using various CYP1A2 substrates opined that, although not conclusive, the current evidence suggests that men have higher CYP1A2 activity than women but more well-designed studies are needed that account for CYP1A2 inducers such as diet and smoking (46).

Although other studies have found that CYP1A2*1F has no effect on inducibility (47, 48), CYP1A2*1F has primarily been reported to alter inducibility, with C/Cs having lower inducibility than A/As (5, 49). We thus hypothesized that there would be no difference in CYP1A2 activity between genotypes on the basal diet, but those homozygous for CYP1A2*1F (C/C) would then have the lowest percent change in CYP1A2 activity in response to the inducing effect of cruciferous vegetables. However, our results were in the opposite direction for the 1C diet compared to basal (i.e., the greatest increase in activity was among those with the C/C genotype, the least among the A/A). Response to the 2C diet compared to 1C was as expected; only those with at least one A allele responded to the increased dose of cruciferous vegetables. Surprisingly, heterozygotes did not respond to the 1C+A diet compared to the 1C diet even though both homozygote groups (A/A and C/C) experienced a 13% and 16% reduction in activity, respectively, following the inclusion of apiaceous vegetables. It is unclear whether A/C could be resistant to the inhibiting effect of apiaceous vegetables or if this unexpected observation is due to noise in the data. Common polymorphisms in CYP1A2 have been found to be in linkage disequilibrium, thus polymorphisms other than CYP1A2*1F may influence results (50). Additionally, while a two-fold variation in CYP1A2 induction has been reported with the CYP1A2*1F genotype (5, 49), the variance in constitutive expression of CYP1A2 is reported in excess of 60-fold with recent evidence suggesting trans-acting factors may have a more significant role in the regulation of CYP1A2 expression and warrant further investigation (51). Another unexpected result was that, compared to the basal diet, CYP1A2 activity was statistically significantly increased in GSTM1-null/GSTT1-null individuals when they consumed the 2C diet but not the 1C diet. We hypothesized that having the GSTM1-null/GSTT1-null genotype would result in the largest induction of activity by both the 1C and 2C diets due to decreased metabolism of inducing agents (i.e., indoles and isothiocyanates). The reason for no significant effect on CYP1A2 activity in GSTM1-null/GSTT1-nulls on the 1C diet is unclear but one possible explanation could be alternative metabolic pathways for indoles and isothiocyanates (such as other GSTs (52)) that can compensate for the lack of GSTM1 and GSTT1 at lower levels of the compounds (1C diet) but not higher levels (2C diet). Moreover, other polymorphisms may exist in other biotransformation pathways that influence the effect of indoles and isothiocyanates, which may vary in prevalence between Asian and Caucasian populations and partially explain the inconsistent association between cruciferous vegetables, cancer, and GST status reported in the epidemiologic literature. Moyer et al. (53) reported two nonsynonymous single nucleotide polymorphisms in GSTM1 with varying prevalence in 100 Caucasian-Americans (CA) compared with 100 Han Chinese-Americans (HCA). The frequency of the Asn85Ser polymorphism was 6.8% in the CA sample yet not present in the HCA sample; the frequency of the Lys173Asn (rs1065411) polymorphism was 25% in the CA group but 61% in the HCA sample. Further investigation of the impact of other biotransformation polymorphisms on phytochemical disposition is warranted.

The gain in CYP1A2 activity 1 day after vegetables were removed from the 1C+A diet compared to the 1C diet was not as pronounced as we hypothesized based on the rapid metabolism of furanocoumarins (27, 28). However, furanocoumarins in apiaceous vegetables may not merely inhibit CYP1A2; several furanocoumarins are mechanism-based inactivators of other cytochrome P-450s (5456). With mechanism-based inactivation, the enzyme produces a reactive intermediate from the parent compound which then irreversibly binds to the enzyme, permanently inactivating it. The only way to overcome the loss of enzyme activity in vivo is synthesis of new enzyme. The net inhibitory effect of combined exposure to CYP1A2 inducers and inhibitors, along with the modest recovery in activity one day after phytochemical sources were removed, suggests that the rapidly metabolized furanocoumarins may be irreversibly inactivating the enzyme.

There are several strengths to our approach. First, participants were recruited based on their genotypes; this ensured adequate statistical power to conduct analyses by genotype. The crossover design provides considerable reduction in interindividual variation because each participant experiences all treatments or interventions; thus although the genotype for all other genes is unknown, there is no genetic variance between intervention groups. This enables a clearer investigation of the genes, specific genetic variances, and dietary components of interest. The detailed inclusion and exclusion criteria further minimized variability, allowing more direct investigation of the effect of diet by decreasing known non-dietary factors that influence CYP1A2 activity. Additionally, participants ate controlled diets, with most of the study vegetables served on site at dinner to enhance adherence. Because men typically weigh more than women, study vegetables were provided based on body weight to achieve uniform dosing and allow investigation of sex effects. Lastly, using phenotype measurements (urinary caffeine metabolism) as the endpoint overcomes issues related to making inferences from measurements of RNA or protein levels to the broader physiological state.

Nonetheless, there are a few limitations to this study. Participants were free living and could have consumed non-study foods that influence CYP1A2 activity without our knowledge. They also could have refrained from eating some of the study foods. However, participants were encouraged to report honestly any deviations from the tightly prescribed diets either verbally or on the daily food checklists and staff developed good rapport with participants to facilitate such disclosures. Second, it is possible, despite uniform dosing of vegetables, that exposure to the bioactive compounds during the four-year study was not uniform due to natural variation in phytochemical content in plant foods. Our efforts to address this included: purchasing frozen items in large lots from the same vendor, purchasing fresh items from the same vendor, and uniform storage and food preparation conditions. Moreover, variation in final exposure can still occur even if equal amounts of phytochemicals are consumed. This is due to interindividual differences in the metabolic disposition of the compounds. For example, interindividual variation in phytochemical-metabolizing gut microbiota can lead to variation in bioavailability (57). Although the crossover design can minimize the impact of this interindividual variation, it cannot influence the effect of potential changes in gut microbiota over time within the same individual. It should also be noted that, although we used a relatively quantitative and robust measure of CYP1A2 phenotype, the involvement of other metabolic and transport processes involved in both caffeine and phytochemical disposition may contribute additional variability to the data. Lastly, our small sample size limited power to undertake formal genotype-diet interaction estimates.

In summary, we report that combined consumption of cruciferous and apiaceous vegetables results in net CYP1A2 inhibition; the inhibitory effect of apiaceous vegetables is sufficient to overcome the inductive effect of crucifers. Increasing the dose of cruciferous vegetables leads to further increases in CYP1A2 activity. Differences between controlled diets were influenced by sex, CYP1A2*1F genotype, and GSTM1/GSTT1 genotype. Removal of apiaceous vegetables from the diet resulted in modest and incomplete, though statistically significant, recovery of CYP1A2 activity after 24 hrs. However, some of our a priori hypotheses, which were based on existing epidemiologic data, are not supported by our results. The interindividual variation related to sex, genetics, dietary patterns, etc. underscores the complexity and challenge in elucidating the relationship of dietary modulation of biotransformation and disease susceptibility in observational studies. More controlled feeding studies are warranted to further elucidate potential mechanisms for diet-disease associations detected in observational studies.


We thank Karen Noar, Kara Breymeyer, and their staff in the Human Nutrition lab for their dedicated work and JoAnn Prunty for her technical support.


*Grant support: This work was supported by US NIH grants R01CA070913, R01CA92288, and P30ES07033


1. Eaton DL, Gallagher EP, Bammler TK, Kunze KL. Role of cytochrome P4501A2 in chemical carcinogenesis: implications for human variability in expression and enzyme activity. Pharmacogenetics. 1995;5:259–74. [PubMed]
2. Boobis AR, Lynch AM, Murray S, et al. CYP1A2-catalyzed conversion of dietary heterocyclic amines to their proximate carcinogens is their major route of metabolism in humans. Cancer Res. 1994;54:89–94. [PubMed]
3. Guengerich FP, Shimada T, Yun CH, et al. Interactions of ingested food, beverage and tobacco components involving human cytochrome P4501A2, 2A6, 2E1 and 3A4 enzymes. Environ Health Perspect. 1994;102:49–53. [PMC free article] [PubMed]
4. Gunes A, Dahl ML. Variation in CYP1A2 activity and its clinical implications: influence of environmental factors and genetic polymorphisms. Pharmacogenomics. 2008;9:625–37. [PubMed]
5. Sachse C, Brockmöller J, Bauer S, Roots I. Functional significance of a C>A polymorphism in intron 1 of the cytochrome P450 CYP1A2 gene tested with caffeine. Br J Clin Pharmacol. 1999;47:445–9. [PMC free article] [PubMed]
6. Lampe JW, King IB, Li S, et al. Brassica vegetables increase and apiaceous vegetables decrease cytochrome P450 1A2 activity in humans: changes in caffeine metabolite ratios in response to controlled vegetable diets. Carcinogenesis. 2000;21:1157–62. [PubMed]
7. Kall MA, Vang O, Clausen J. Effects of dietary broccoli on human in vivo drug metabolising enzymes: evaluation of caffeine, estrone and chlozoxazone metabolism. Carcinogenesis. 1996;17:793–9. [PubMed]
8. Gross-Steinmeyer K, Stapleton PL, Liu F, et al. Phytochemical-induced changes in gene expression of carcinogen-metabolizing enzymes in cultured human primary hepatocytes. Xenobiotica. 2004;34:619–32. [PubMed]
9. Lake BG, Tredger JM, Renwick AB, Barton PT, Price RJ. 3,3′-Diindolylmethane induces CYP1A2 in cultured precision-cut human liver slices. Xenobiotica. 1998;28:803–11. [PubMed]
10. Keum YS, Jeong WS, Kong ANT. Chemoprevention by isothiocyanates and their underlying molecular signaling mechanisms. Mutat Res. 2004;555:191–202. [PubMed]
11. Slattery ML, Kampman E, Samowitz W, Caan BJ, Potter JD. Interplay between dietary inducers of GST and the GSTM1-genotype in colon cancer. Int J Cancer. 2000;87:728–33. [PubMed]
12. London SJ, Yuan JM, Chung FL, et al. Isothiocyanates, glutathione S-transferase M1 and T1 polymorphisms, and lung-cancer risk: a prospective study of men in Shanghai, China. Lancet. 2000;356:724–9. [PubMed]
13. Seow A, Yuan JM, Sun CL, Van Den Berg D, Lee HP, Yu MC. Dietary isothiocyanates, glutathione S-transferase polymorphisms and colorectal cancer risk in the Singapore Chinese Health Study. Carcinogenesis. 2002;23:2055–61. [PubMed]
14. Brennan P, Hsu CC, Moullan N, et al. Effect of cruciferous vegetables on lung cancer in patients stratified by genetic status: a mendelian randomisation approach. Lancet. 2005;366:1558–60. [PubMed]
15. Wang LI, Giovannucci EL, Hunter D, Neuberg D, Su L, Christiani DC. Dietary intake of Cruciferous vegetables, Glutathione S-transferase (GST) polymorphisms and lung cancer risk in a Caucasian population. Cancer Causes Control. 2004;15:977–85. [PubMed]
16. Joseph MA, Moysich KB, Freudenheim JL, et al. Cruciferous vegetables, genetic polymorphisms in glutathione S-transferases M1 and T1, and prostate cancer risk. Nutr Cancer. 2004;50:206–16. [PubMed]
17. Ambrosone CB, McCann SE, Freudenheim JL, Marshall JR, Zhang Y, Shields PG. Breast cancer risk in premenopausal women is inversely associated with consumption of broccoli, a source of isothiocyanates, but is not modified by GST genotype. J Nutr. 2004;134:1134–8. [PubMed]
18. Gaudet MM, Olshan AF, Poole C, Weissler MC, Watson M, Bell DA. Diet, GSTM1 and GSTT1 and head and neck cancer. Carcinogenesis. 2004;25:735–40. [PubMed]
19. Gasper AV, Al-Janobi A, Smith JA, et al. Glutathione S-transferase M1 polymorphism and metabolism of sulforaphane from standard and high-glucosinolate broccoli. Am J Clin Nutr. 2005;82:1283–91. [PubMed]
20. Steck SE, Gammon MD, Hebert JR, Wall DE, Zeisel SH. GSTM1, GSTT1, GSTP1, and GSTA1 polymorphisms and urinary isothiocyanate metabolites following broccoli consumption in humans. J Nutr. 2007;137:904–9. [PMC free article] [PubMed]
21. Dashwood RH, Uyetake L, Fong AT, Hendricks JD, Bailey GS. In vivo disposition of the natural anti-carcinogen indole-3-carbinol after po administration to rainbow trout. Food Chem Toxicol. 1989;27:385–92. [PubMed]
22. Lombaert GA, Siemens KH, Pellaers P, Mankotia M, Ng W. Furanocoumarins in celery and parsnips: method and multiyear Canadian survey. J AOAC Int. 2001;84:1135–43. [PubMed]
23. Beier RC. Natural pesticides and bioactive components in foods. Rev Environ Contam Toxicol. 1990;113:47–137. [PubMed]
24. Ono S, Hatanaka T, Hotta H, Satoh T, Gonzalez FJ, Tsutsui M. Specificity of substrate and inhibitor probes for cytochrome P450s: evaluation of in vitro metabolism using cDNA-expressed human P450s and human liver microsomes. Xenobiotica. 1996;26:681–93. [PubMed]
25. Mays DC, Camisa C, Cheney P, Pacula CM, Nawoot S, Gerber N. Methoxsalen is a potent inhibitor of the metabolism of caffeine in humans. Clin Pharmacol Ther. 1987;42:621–6. [PubMed]
26. Bendriss EK, Bechtel Y, Bendriss A, et al. Inhibition of caffeine metabolism by 5-methoxypsoralen in patients with psoriasis. Br J Clin Pharmacol. 1996;41:421–4. [PMC free article] [PubMed]
27. Kharasch ED, Hankins DC, Taraday JK. Single-dose methoxsalen effects on human cytochrome P-450 2A6 activity. Drug Metab Dispos. 2000;28:28–33. [PubMed]
28. John BA, Chasseaud LF, Wood SG, Forlot P. Metabolism of the anti-psoriatic agent 5-methoxypsoralen in humans: comparison with rat and dog. Xenobiotica. 1992;22:1339–51. [PubMed]
29. Navarro SL, Peterson S, Chen C, et al. Cruciferous vegetable feeding alters UGT1A1 activity: diet- and genotype-dependent changes in serum bilirubin in a controlled feeding trial. Cancer Prev Res. 2009;2:345–52. [PMC free article] [PubMed]
30. Zeman FJ. Clinical Nutrition and Dietetics. New York, NY: Macmillan; 1983.
31. Frankenfeld D, Roth-Yousey L, Compher C. Comparison of predictive equations for resting metabolic rate in healthy nonobese and obese adults: a systematic review. J Am Diet Assoc. 2005;105:775–89. [PubMed]
32. Ainsworth BE, Haskell WL, Whitt MC, et al. Compendium of physical activities: an update of activity codes and MET intensities. Med Sci Sports Exerc. 2000;32:S498–S504. [PubMed]
33. Nyéki Á, Buclin T, Biollaz J, Decosterd LA. NAT2 and CYP1A2 phenotyping with caffeine: head-to-head comparison of AFMU vs. AAMU in the urine metabolite ratios. Br J Clin Pharmacol. 2003;55:62–7. [PMC free article] [PubMed]
34. Butler MA, Lang NP, Young JF, et al. Determination of CYP1A2 and NAT2 phenotypes in human populations by analysis of caffeine urinary metabolites. Pharmacogenetics. 1992;2:116–27. [PubMed]
35. Rostami-Hodjegan A, Nurminen S, Jackson PR, Tucker GT. Caffeine urinary metabolite ratios as markers of enzyme activity: a theoretical assessment. Pharmacogenetics. 1996;6:121–49. [PubMed]
36. Tucker GT, Rostami-Hodjegan A, Jackson PR. Determination of drug-metabolizing enzyme activity in vivo: pharmacokinetic and statistical issues. Xenobiotica. 1998;28:1255–73. [PubMed]
37. Sinués B. Five caffeine metabolite ratios to measure tobacco-induced CYP1A2 activity and their relationships with urinary mutagenicity and urine flow. Cancer Epidemiol Biomarkers Prev. 1999;8:159–66. [PubMed]
38. Chung FL, Jiao D, Getahun SE, Yu MC. A urinary biomarker for uptake of dietary isothiocyanates in humans. Cancer Epidemiol Biomarkers Prev. 1998;7:103–8. [PubMed]
39. Laird NM, Skinner J, Kenward M. An analysis of two-period crossover designs with carry-over effects. Stat Med. 1992;11:1967–79. [PubMed]
40. Little RJ. Modeling the drop-out mechanism in repeated-measures studies. J Am Stat Assoc. 1995:90.
41. Shin A, Shrubsole MJ, Rice JM, et al. Meat intake, heterocyclic amine exposure, and metabolizing enzyme polymorphisms in relation to colorectal polyp risk. Cancer Epidemiol Biomarkers Prev. 2008;17:320–9. [PMC free article] [PubMed]
42. Crump C, Chen C, Appelbaum FR, et al. Glutathione S-transferase theta 1 gene deletion and risk of acute myeloid leukemia. Cancer Epidemiol Biomarkers Prev. 2000;9:457–60. [PubMed]
43. Kadlubar FF, Young JF, Lang NP, Caporaso NE, Sinha R, Landi MT. Correspondence re: letter to the editor by B. K. Tang and W. Kalow on CYP1A2 phenotyping using caffeine. Cancer Epidemiol. Biomarkers & Prev. 5:231, 1996. Cancer Epidemiol Biomarkers Prev. 1996;5:757–8. [PubMed]
44. Lampe JW, Chen C, Li S, et al. Modulation of human glutathione S-transferases by botanically defined vegetable diets. Cancer Epidemiol Biomarkers Prev. 2000;9:787–93. [PubMed]
45. Hudson CE, Schulte BA, Sutter TR, Norris JS. Steroid hormones modulate expression of cytochrome P450 enzymes in male hamster reproductive tract and leiomyosarcomas. Carcinogenesis. 2001;22:763–70. [PubMed]
46. Scandlyn MJ, Stuart EC, Rosengren RJ. Sex-specific differences in CYP450 isoforms in humans. Expert Opin Drug Metab Toxicol. 2008;4:413–24. [PubMed]
47. Aklillu E, Carrillo JA, Makkonnen E, et al. Genetic polymorphism of CYP1A2 in Ethiopians affecting induction and expression: characterization of novel haplotypes with single-nucleotide polymorphisms in intron 1. Mol Pharmacol. 2003;64:659–69. [PubMed]
48. Larsen JT, Brosen K. Consumption of charcoal-broiled meat as an experimental tool for discerning CYP1A2-mediated drug metabolism in vivo. Basic Clin Pharmacol Toxicol. 2005;97:141–8. [PubMed]
49. Ghotbi R, Christensen M, Roh HK, Ingelman-Sundberg M, Aklillu E, Bertilsson L. Comparisons of CYP1A2 genetic polymorphisms, enzyme activity and the genotype-phenotype relationship in Swedes and Koreans. Eur J Clin Pharmacol. 2007;63:537–46. [PubMed]
50. Sachse C, Bhambra U, Smith G, et al. Polymorphisms in the cytochrome P450 CYP1A2 gene (CYP1A2) in colorectal cancer patients and controls: allele frequencies, linkage disequilibrium and influence on caffeine metabolism. Br J Clin Pharmacol. 2003;55:68–76. [PMC free article] [PubMed]
51. Jiang Z, Dragin N, Jorge-Nebert LF, et al. Search for an association between the human CYP1A2 genotype and CYP1A2 metabolic phenotype. Pharmacogenet Genomics. 2006;16:359–67. [PubMed]
52. Zhang Y, Kolm RH, Mannervik B, Talalay P. Reversible conjugation of isothiocyanates with glutathione catalyzed by human glutathione transferases. Biochem Biophys Res Commun. 1995;206:748–55. [PubMed]
53. Moyer AM, Salavaggione OE, Hebbring SJ, et al. Glutathione S-transferase T1 and M1: gene sequence variation and functional genomics. Clin Cancer Res. 2007;13:7207–16. [PubMed]
54. Koenigs LL, Trager WF. Mechanism-based inactivation of P450 2A6 by furanocoumarins. Biochemistry. 1998;37:10047–61. [PubMed]
55. Koenigs LL, Trager WF. Mechanism-based inactivation of cytochrome P450 2B1 by 8-methoxypsoralen and several other furanocoumarins. Biochemistry. 1998;37:13184–93. [PubMed]
56. von Weymarn LB, Zhang QY, Ding X, Hollenberg PF. Effects of 8-methoxypsoralen on cytochrome P450 2A13. Carcinogenesis. 2005;26:621–9. [PubMed]
57. Shapiro TA, Fahey JW, Wade KL, Stephenson KK, Talalay P. Human metabolism and excretion of cancer chemoprotective glucosinolates and isothiocyanates of cruciferous vegetables. Cancer Epidemiol Biomarkers Prev. 1998;7:1091–100. [PubMed]