The Quantitation of Metabolites of Quercetin Flavonols in Human Urine"2

Vol. 5, 711-720, September 1996 Cancer Epidemiology, Biomarkers & Prevention 711

Myron Gross, 3 Michelle Pfeiffer, Margaret Martini, analytical procedure is an essential first step for Deborah Campbell, Joanne Slavin, and John Potter validation of the metabolites as biomarkers of quercetin Division of Epidemiology, School of Public Health, University of Minnesota, intake. Minneapolis, Minnesota 55454-1015 [M. G., M. P., J. P.], and Department of Food Science and Nutrition, University of Minnesota, St. Paul, Minnesota 55108-6099 [M. M., D. C., J. S.] Introduction Human diets contain a wide range of compounds with possible "semi-essential" functions. One group of such compounds is Abstract the flavonoids. Flavonoids have multiple chemical and biolog- The flavonoid quercetin, or its metabolites, inhibit ical actions, including antioxidant (1-4), chelation (5, 6), an- chemical carcinogenesis in rodents and may have a role ticarcinogenic (7, 8), bacteriostatic (9), and secretory activities in the prevention of human cancers. Quercetin exposure (10, 11). Several enzymes, including trypsin, leucine amino- in human populations results from the dietary intake of transferase, phenosulfotransferase, DNA topoisomerase, phos- various plant foods; high concentrations of quercetin are phoinositol kinase, NADPH diaphorase, H+,K~+)-ATPase, and found in apples, onions, tea, and red wine. Determination xanthine oxidase, are modulated by flavonoids (12-18). Fur- of the relationship between dietary intake and cancer risk thermore, some flavonoids may act as antiviral, antitumor, depends on the characterization of quercetin intake. The anti-inflammatory and antiallergenic agents (19-21). development and use of biomarkers for quercetin intake Flavonoids have a widespread distribution among food may provide a basis for the objective classification of this plants. In general, fruits and vegetables are good dietary sources exposure. One possible biomarker is metabolic products of flavonoids (22-28). The flavonoids are found primarily in of quercetin. We report the development of a high- the outer layers of fruits and vegetables; the peel and outer performance liquid chromatography (HPLC)-based assay leaves have the highest concentrations (22, 23, 29-32). Human for quantitation of quercetin metabolites in human urine. consumption of flavonoids is estimated at 1 g or more per day The metabolites include 3,4-dihydroxyphenylacetic acid for the average American diet (33). (homoprotocatechuic acid), metahydroxyphenylacetic One flavonoid, quercetin, is of particular interest because acid, and 4-hydroxy-3-methoxyphenylacetic acid of its anticarcinogenic activities (7) and a significant quantita- (homovanillic acid). The assay has only two major steps, tive presence in human foods. Quercetin is found in numerous ether extraction and HPLC analysis, and is suitable for fruits and vegetables (22-24), the highest concentrations oc- analysis of large sample numbers. Analytical curring in apples, onions, and tea (34-36). Biological activities characteristics of the assay include a sensitivity of less of quercetin include the inhibition of cell proliferation in ani- than 1 ?tg, precision with coefficients of variation < 10%, mals and cell cultures (37) and prevention of chemically in- and metabolite recoveries >90%. The mean duced tumors. Early reports of carcinogenic activity by quer- concentrations of 3,4-dihydroxyphenylacetic acid, cetin are based on bacterial mutagenicity tests (38). metahydroxyphenylacetic acid, and homovanillic acid in Extrapolation of the results to humans, in this case, led to an two human urine samples are approximately 0.7, 4.8, and erroneous conclusion. Tests of the carcinogenicity of quercetin 2.8/xg/ml, respectively. The identification of each in mammalian species, with few exceptions (39, 40), have metabolite is confirmed by HPLC, UV absorbance scans, proven to be negative (41, 42), and epidemiological data do not and gas chromatography-mass spectrometry analysis. implicate carcinogenic activity for quercetin in humans (43). These results verify the occurrence of quercetin Interpretations of more recent data from in vitro experiments, metabolites in human urine and the feasibility of suggesting carcinogenic activity for quercetin, do not consider quercetin metabolite quantitation, by the assay described the absence of quercetin absorption (44) and almost complete herein, for epidemiological studies. Development of the metabolism of quercetin in the human gastrointestinal tract (45). Thus, the evidence to date does not support carcinogenic activities for quercetin in humans (46, 47) but rather a chemopreventive effect (48). The chemopreventive effect may be Received 5/5/96; revised 6/11/96; accepted 6/13/96. mediated by metabolites of quercetin. Also, quercetin or its The costs of publication of this article were defrayed in part by the payment of metabolites may have cytotoxic activities (49). page charges. This article must therefore be hereby marked advertisement in Because diets associated with a reduction in cancer risk accordance with 18 U.S.C. Section 1734 solely to indicate this fact. generally contain numerous potential anticarcinogens (43, 50, 1 Supported by Grant P01 CA50305 (a Colon Cancer Prevention Research Unit) from the National Cancer Institute. 51), biomarkers of dietary quercetin would be useful in studies 2 A portion of the results reported herein were presented at the annual meeting of of diet-cancer relationships. Studies on the metabolism of quer- the American Association for Cancer Research in 1994 in San Francisco. cetin suggest degradation by intestinal microbes and relatively 3 To whom requests for reprints should be addressed, at Division of Epidemiol- low absorption of quercetin (52, 53), limiting the use of quer- ogy, School of Public Health, University of Minnesota, 1300 South Second Street, Suite 300, Minneapolis, MN 55454-1015. cetin as a biomarker. However, metabolites of quercetin have

-i

o / /

CID t'~ w i O.1 /

it_ / O / =~ / o / / ...... ""'"

l ....""'""

i i i I i i | I i i 1

0.00 20.00 60.00 75,02 Time (minutes) B

i /

e- ..." en :. N

.: e- ¢g / i , !I I!/•

ll [ P~ t! , i ; ~ i~ i * I l" • i' r 'f

I 1 0.00 20.00 40,00 60.00 75.0~ Time (minutes)

Fig. 1. Chromatograms of standards and urine sample extracts. A, chromatogram from the analysis of a mixture containing 5/zg of each standard, prepared from crystalline compounds. B and C, chromatograms from the analysis of urine sample extracts• The extracts yielding the chromatograms shown in B and C came from separate urine sample collections by two individuals.

i E

I i go I'd Q) ! i=i .: m i L ! 0 .0 i IZ i i " I // i' ~'t ] .,..- "

I I! n! , ...... -I li ,= ~ rl ~i ...... !...... !i i ii ~ i\ ...... H!,"fl ..... !, ! 11 :~ ~ ..'" 'I I r ,

i i i --i ] "r ] ~ ' ' ' 'I i !

t-~ ~- Time (minutes)

Fig. 1. Continued.

potential as biomarkers of quercetin and quercetin-containing ford, IL). Acetonitrile and methanol were purchased from EM food intake. Metabolism of quercetin yields DHPAA, 4, mH- Science (Gibbstown, NJ). PAA, and HVA (54). These three metabolites are excreted in Preparation of Standard Solutions and Determination of the urine of rats, rabbits, and humans (54-56). Calibration Curves. Extinction coefficients for DHPAA, mH- A prerequisite for the evaluation of metabolic products PAA, HVA, and p-coumaric acid were determined as follows: from quercetin as biomarkers is the development of suitable (a) each crystalline compound was dried to a constant weight at analytic methods. Presently available methods are inadequate 90°C and solubilized with methanol; (b) the A maximum was for the identification and quantitation of quercetin metabolites determined for each compound; and (c) absorbance measure- (54, 57-60). We report the development of a quantitative ments at the A maxima were determined for DHPAA, mHPAA, method for the detection of quercetin metabolites in human urine samples. The method is an HPLC-based assay, using an HVA, and p-coumaric acid solutions, at known concentrations on-line photodiode detector. by weight. The A maxima and extinction coefficient (1%, 1 cm), respectively, for the compounds were as follows: 272 nm and 172 (DHPAA), 272 nm and 127 (mHPAA), 272 nm and Materials and Methods 174 (HVA), and 303 nm and 1176 (p-coumaric acid). Response Materials. DHPAA, mHPAA, HVA, and p-coumaric acid factors were determined for each analyte, using coumaric acid were purchased from Sigma Chemical Co. (St. Louis, MO). as an internal standard. Ethyl ether anhydrous was obtained from Mallinckrodt Spe- The routine preparation of DHPAA, HVA, mHPAA, and cialty (Paris, KY). HPLC-grade TFA and BSTFA with 1% p-coumaric acid standard solutions was done by solubilization trimethylchlorosilane were from Pierce Chemical Co. (Rock- of crystalline compounds in methanol, filtration, measurement of absorbance at the respective A maxima and calculation of concentrations, using the respective compound's extinction co- 4 The abbreviations used are: DHPAA, 3,4-dihydroxyphenylacetic acid (homo- efficient. Each standard solution was evaluated for purity by protocatechuic acid); HPLC, high-performance liquid chromatography; mHPAA, HPLC analysis, as described in "HPLC Analysis" below, and metahydroxyphenylacetic acid; HVA, 4-hydroxy-3-methoxyphenylacetic acid all standards were greater than 97% pure. All standard solutions (homovanillic acid); TFA, trifluoroacetic acid; BSTFA, N,O,-bis(trimethylsilyl)- triflouroacetamide; GC, gas chromatography; CV, coefficient of variation. were stable for at least 6 months with storage at -70°C.

Collection of Urine Samples. Two male volunteers collected A 24-h urine samples for 3 consecutive days. The subjects, ages 24 and 40, consumed their habitual diets and maintained their usual light exercise schedule during the collection period. Urine was collected in 1.0-liter polypropylene bottles containing 10.0 ml of 5% acetic acid. Acetic acid was added to retard bacteria growth and stabilize the metabolites of quercetin. Samples were stored in a refrigerator at approximately 4°C until delivered to the laboratory, generally 15 h after the completion of each 24-h collection. Three-day urine pools were prepared for each indi-

,/% vidual from frozen 24-h collections and stored at -70°C. Collections and pools were prepared twice from each individ- o i k t ual. These investigations were done according to the principles of the Declaration of Helsinki (61), which describes ethical

,- : principles and recommendations regarding the orientation of \..,/ ~,~. research with human beings. . ,,, '~.._~_.~ Extraction of DHPAA, mHPAA, and HVA. Twenty-ml ali- quots of urine were adjusted with 5.0 normal hydrochloric acid to pH 1.5 and saturated with NaC1. Fifty/xg ofp-coumaric acid in methanol (30 /xl) were added as an internal standard, and each sample was extracted with 20 ml of anhydrous ethyl ether. After initial mixing and venting, the sample was mixed con- tinuously for 10 min on a reciprocal shaker at medium speed il (IKA-Works, Cincinnati, OH). The aqueous layer was col- !i lected, after low-speed centrifugation (1650 rpm), for phase I <, separation, and re-extracted with 20 ml of anhydrous ethyl I ether; the resulting ether extract was combined with the initial ether extract. The combined extracts were evaporated under N nitrogen and reconstituted with methanol/0.1% TFA (50 /xl), !i followed by 950/xl of H20-0.1% TFA. The reconstituted sam- Z - ples were centrifuged at 12,000 rpms in a Beckman microfuge (Fullerton, CA) for 5 min and passed through a 0.45-/xm, organic filter from Millipore (Milford, MA). HPLC analysis was performed using 100 /xl of the resulting solution. All procedures were done at room temperature. Notably, the combined ether extracts could be stored overnight at 4°C, without ~--.-S-..d '" ; ,.;.~ .~...... ~ ...... any effect on analyte concentration. HPLC Analysis. The HPLC system consisted of a Vydak C18 column (Hesperia, CA) with a Chrom Tech C 2 guard column ,,,0-... (Apple Valley, MN), Beckman 507 autosampler (Fullerton, CA), Beckman 168 diode array detector module, Beckman 116 programmable solvent module, and a Waters 470 scanning fluorescence detector (Millipore, Milford, MA). Both the diode array and fluorescence detectors had scanning capabilities• A !.. chromatogram was developed at a 1.0 ml/min flow rate with a gradient consisting of 0.1% TFA from 0 to 30 min, 0.1% TFA and --*10% methanol-0.1% TFA from 30-35 min in a linear gradient, and 0.1% TFA- 10% methanol and --4). 1% TFA-20% i *** methanol from 35-65 min in a linear gradient. Analyte detec- i . tion was by absorbance at 280 nm and fluorescence at 310 nm with an excitation wavelength of 280 nm. In addition, absorbance and fluorescence scans were done for analyte identifica- tit\d2 tion and verification of standard compound spectrometric prop- ~'... erties. The column was washed between each analysis with ;LL..LZLZ. :..;_; :;:L-.= ...... 40% methanol/20%-0.1% TFA/40% acetonitrile for approximately 10 min. i 220.00 300.00 Mass Spectrometric Analysis. Mass spectra were acquired 400.00 with a Finnigan MAT (San Jose, CA); the MAT 95 mass Wiivelenoth (nm) spectrometer was equipped with a Hewlett Packard HP 5890 Series II gas chromatograph (Avondale, PA). The ion source Fig. 2. UV/visible scans of standards and comigrating compounds from urine temperature was 200°C. An electron energy of 70 eV was used sample extracts. Scans were taken of standard and comigrating compounds for acquiring electron ionization GC-mass spectrometry data. analyzed in chromatograms in Fig. 1, A and B. A, overlays of DHPAA (Fig. IA) and the corresponding comigrating compound (Fig. 1B). B and C, similar overlays Using a splitless injector maintained at 250°C, samples were for mHPAA and HVA, respectively. introduced into the ion source via a DB-5 fused silica capillary

i00- 384 .I

80"

60"

267.2 40" 73:0 179.1

20- 36%1 386.2

237.1 45.0 077.0 104.0 147.0 .I ] 281.0 308.9 ...... 1.. ll.r-,,r--.,I t,. ~Lo .... k,., I,.~., F , ...... L: 100 200 300 400 M/Z

B 73.1 I00" 291;.i

281.1 80-

164.1 60"

40- Ls.o 147.o 252.1 20- 133,. 0 L! 191.i 237.1 1 45.0 89.0 105.1 I 207 .i I 308.9 ..1t, I,,...... L ,IIiF~-~'2',.. , I , t, J_ ,[ , -, , , , ' ' ' ' ' 50 ' I "' ' I00 150 200 250 300 350 M/Z

326.1 I00-

80- 209.1 73.0 60-

311.i 40" 26~.2 179.1 327 .i 20" __75-0 1 1 149.0 206.0 296.0 1 28 0 ~5.0 L,8910 104.0 I 163.1 1193.1--~ 23,7.1 28 i .I, ,[-~,, ,,, !,fl;_ ,Jlt, ,,I!, ,,L, I, ,., ,L., .I,.. k, ,. k.,., ' ; ' ~150 ' ' i00 150 200 250 300 35O M/Z

Fig. 3. (Continues on next page.) Mass-spectrometric analysis of standards and comigrating compounds from urine sample extracts. Spectra are for DHPAA (A) and the corresponding comigrating compound (D), mHPAA (B) and the corresponding comigrating compound (E) and 3) HVA (C) and the corresponding comigrating compound (/9.

[D 384.1 i00 -

80-

6o ~

-a 267 .I .4 40- 73. 179.0 I

369.1 20- I] 386.1 237.0 1041.0 147j .0 I l'C93.1

, , , , , , , I .,, ,.. If,, u !,., !. L ...... 100 200 300 400 M/Z

73. i00-

80"

-M

60"

-a

4O" 164.1 296.1 281.1 I

20"

133.0 [--165.0 191.0 252.1 45.0 7_7.0 105 0 I ti , 207.0 237.1 1253.1 299.1 327.0 , ,,J, ..... L ,, ,,.i ,~..L I ~ L !.. ,[- Lr- r- .... , , . , ,, . . ,' - , , , . 50 100 150 200 250 300 350 M/Z

F 73.0 i00-

326.1 80- 209.1

60-

311.i 40- 179.1

267.2 2..._7.2 20-

44.0 296.1 217.0 j 59i0 77-0 I04.0 148.0 206.0 237.1 [{6___8.1 ...... JaL, t ,? I J ~ 50 i00 150 200 250 300 350 M/Z

Fig. 3. Continued.

.t2 greater than 0.1 units for 5-/xg samples (Fig. 1A). The retention times of DHPAA, mHPAA, HVA, and p-coumaric acid were approximately 26.0, 43.0, 48.0, and 58.0 rnin, respectively. Fig. 1, B and C, shows chromatograms from the analysis

E of urine samples from two individuals. Each urine sample extract contained compounds with retention times, relative to the internal standard, identical to those of standards for DH- Od 0.06 PAA, mHPAA, and HVA. A comparison of chromatograms in ¢.-o= Fig. 1, B and C, with each other and the chromatogram in Fig. GI .0 1A illustrate important considerations for appropriate applica- o tion of this assay. (a) The analysis for chromatograms in Fig. 1, ..CI < A and C, were done on different days; analysis for chromatograms on Fig. 1, A and B, were done on the same day. Slight shifts in absolute retention times of the analytes were seen between analysis days; compare Fig. 1, A and B, with Fig. 1C. 0.00 | | ! ! ! Thus, automated data analysis procedures required the use of

0.2 relative retention times. (b) Significant differences were found B in the complexity of chromatograms for samples from the different individuals; compare Fig. 1, B and C. The chromatogram in Fig. 1B was more complex than that in Fig. 1C. Based on the analysis of numerous samples (> 100; data not shownS), chromatogram C represents a typical chromatogram; the more complex chromatograms were relatively rare (<5%). However,

0.1- verification of analyte identity by UV/visible absorbance scans was essential for the analysis of the more complex chromatograms. UV/visible scans of absorbance peaks comigrating with standard compounds had spectra identical to the standards (see below). In contrast, surrounding absorbance peaks had spectra different from those of the standard compounds. (c) The retention time of the internal standard, p-coumaric acid, was not

0.0 , i affected by the sample matrix (Fig. 1B); it comigrated with the standard (Fig. 1A). In addition, p-coumaric acid migrated in an 0.3 area of the chromatogram without other peaks as determined by the analysis of a urine sample lacking added p-coumaric acid (data not shown). Significant amounts of each analyte were present in each urine sample; estimated amounts of DHPAA, mHPAA, and E 0.2 ¢- HVA were 0.87, 4.46 and 3.36 /xg/ml, respectively (Fig. 1B) and 0.44, 5.09, and 2.30/xg/ml, respectively (Fig. 1C). Verification of DHPAA, mHPAA, and HVA identification in urine samples was done by several methods. UV/visible r- scans of DHPAA, mHPAA, and HVA from the chromato- e'~ 0.1 graphic analysis shown in Fig. 1A and comigrating analytes

< from the chromatogram shown on Fig. 1B are shown in Fig. 2A (DHPAA), B (mHPAA), and C (HVA). Overlays of the scans indicated identical spectral properties for the standards and

0,0 ! i comigrating compounds. Slight differences at the ends of ab- 0 sorbance spectra resulted from the normalization of scans for concentration differences between standards and urine extract concentration (ug/lOOul) analytes. The h maxima for the standard and urinary compound was identical for each analyte. It was 272 nm for DHPAA, Fig. 4. Dose-response curves for DHPAA (A), mHPAA (B), and HVA (C). mHPAA, and HVA. Additional evaluation of DHPAA, mHPAA, and HVA identification in urine samples was done by the analysis of urine samples after the addition of DHPAA, mHPAA, and HVA from column (25 M × 0.25 mm, 0.25 /xm film thickness; J & W standard solutions of these compounds. A comparison of chro- Scientific, Folsum, CA). The column was heated from 80 to matograms from analysis of the original urine sample (no added 300°C at 5°C/min, with a 5-min hold at 300°C. The transfer line compounds) and the samples with added DHPAA, mHPAA, or between the gas chromatograph and the mass spectrometer was HVA indicated an increase in the respective peaks identified as maintained at 300°C.

Results A mixture of DHPAA, mHPAA, HVA, and p-coumaric acid 5 M. D. Gross, M. D. Martini, M. D. Campbell, G. Granditis, J. Potter, and J. (internal standard) was analyzed by HPLC analysis. Each ana- Slavin. Urinary metabolites of quercetin flavonols as biomarkers of fruit and lyte produced sharp, symmetrical peaks with absorbances vegetable intake, manuscript in preparation.

Table I Assay precision CVs (%)

DHPAA mHPAA HVA Pool Within day Between day Overall Within day Between day Overall Within day Between day Overall

1 3.18 6.70 7.40 4.65 3.10 6.60 2 3.40 7.90 8.70 2.08 3.80 4.40 2.60 4.50 5.30

DHPAA, mHPAA, and HVA in the original urine sample (data Table 2 Extraction efficiency not shown). Number of extractions Percentage recovery of coumaric acid DHPAA, mHPAA, and HVA were derivatized with BSTFA containing 1% trimethylchlorosilane, and each com- 2 90.5 pound was analyzed by GC-mass spectrometry. Full-scan mass 3 94.5 spectra from the analysis are shown in Fig. 3 A-C. An identical 4 77.3 analysis was done on compounds isolated from urinary extracts that comigrated with DHPAA, mHPAA, and HVA in HPLC analysis. The mass spectra from these analyses are shown in Extraction efficiency was determined for the use of 2-4 Fig. 3 D-F. DHPAA and the HPLC-comigrating compound extracts. Recovery ofp-coumaric acid was utilized as a measure from urinary extracts had retention times of 22.49 and 22.44 of extraction efficiency. The results are shown in Table 2. More min, respectively, in GC analysis. The mass spectra for these than 90% of p-coumaric acid was recovered after two or three analytes is shown on Fig. 3, A and D, respectively. The mo- extractions; a slightly larger recovery was found with three lecular ion found for DHPAA (m/z = 384.1) was in good extractions. Recovery of p-coumaric acid was lower with the agreement with the theoretical molecular ion (m/z = 384.47) use of four extractions. for the compound. The molecular ion (m/z = 384.1) and all An evaluation was done of extract stability. Extractions major fragmentation ions from analysis of the HPLC-comigrat- were done on day 0 and stored at 4°C for 0-2 days. HPLC ing compound were in good agreement with those found for analysis of the extracts was done on days 0, 1, and 2. All DHPAA. Differences in ion intensities were a result of sample analyses were done in duplicate. Results of these analyses are amounts used in the analysis. shown on Table 3. HVA and mHPAA concentrations were MHPAA and the HPLC-comigrating compound from uri- essentially unchanged after the 2-day storage period. DHPAA nary extracts had retention times of 17.46 and 17.43 min, concentrations were lower after storage for 1 (81%) and 2 respectively, in GC analysis. The molecular ion of mHPAA (87.3%) days. The difference between DHPAA concentrations (m/z = 296.1) was in good agreement with the predicted moon days 1 and 2 was not significant. lecular ion for the compound (m./z = 296.32). The molecular DHPAA, mHPAA, and HVA were detected by fluores- ion (m/z = 296.1) and all major fragmentation ions from cence emission. This method of detection was more sensitive analysis of the HPLC-comigrating compound were in good than detection by UV absorbance. However, fluorescence had a agreement with those found for mHPAA (Fig. 3, B and E). lower signal-to-noise ratio than UV absorbance. Thus, fluores- Good agreement was found between the retention times in GC cence emission was not a better method of detection than UV analysis for HVA and the HPLC-comigrating compound (21.29 absorption for this assay. versus 21.27 min). The molecular ions for the HPLC-comigrating compound and HVA and fragmentation patterns were in good agreement. The predicted, HVA, and HPLC-comigrating Discussion compound molecular ions were 326.39, 326.1 and 326.1 m/z, This article reports the development of an HPLC-based method respectively. In summary, the results from HPLC, UV/visible for the detection and quantitation of DHPAA, mHPAA, and scans and mass spectral analysis indicated that human urine HVA in human urine samples. The method consists of two samples contained DHPAA, HPAA, and HVA. major steps: ether extraction at acid pH and analysis of the Analytical characteristics of the HPLC-based assay were extract by HPLC. These two steps are sufficient and necessary evaluated for the quantitation of DHPAA, mHPAA, and HVA. for the detection and quantitation of DHPAA, mHPAA, and Dose-response curves for DHPAA, mHPAA, and HVA are HVA in human urine samples. The assay procedure allows for shown on Fig. 4A, B, and C, respectively. Less than 0.25/zg of the quantitation of each and all analytes in a single HPLC each analyte/100-p,1 injection was detectable, and dose-reanalysis. Dose-response curves for standards and samples are sponse curves were linear. The lowest r a value from linear steep and linear. Assay sensitivity is high and sufficient for the regression analysis of the dose-response curves was 0.996. analysis of human urine samples. Approximately 0.25 ~g of Precision of the assay was determined by replicate analysis each analyte can be detected by the assay. High recoveries of a second set of urine pools from the two volunteers. Multiple (>90%) are found in routine analysis. Precision of the assay is analyses of each urine pool were done on each of 4 days. One high, with an overall coefficient of variation of less than 10% urine pool contained mHPAA and HVA; it did not contain for each analyte. These analytic characteristics are appropriate DHPAA. The other urine pool contained each of the three for the routine detection of DHPAA, mHPAA, and HVA in analytes. CVs for the analysis are shown in Table 1. The overall human urine. CVs were less than 10% for all analytes in either pool. Be- The presence and identity of each analyte in human urine tween-day CVs were slightly higher than within-day CVs for all samples are verified by several criteria, which are: (a) urinary analytes. Similar results were obtained from the analysis of analytes comigrate with respective crystalline standards in either pool. Also, replicate analysis indicated that peak height HPLC analysis; (b) identical UV/visible scans are found for was more precise than peak area for the quantitation of metab- standards and urinary analytes; (c) spiking urine samples with olites (data not shown). DHPAA, mHPAA, and HVA increases the expected peak

Table 3 Extraction stability

Analysis DHPAA mHPAA HVA time Concentration % of day 0 Concentration % of day 0 Concentration % of day 0

Day 0 1.42 100.0 8.53 100.0 5.95 100.0 Day 1 1.15 81.0 8.85 103.8 5.75 96.5 Day 2 1.24 87.3 8.65 101.4 5.80 97.3

heights in HPLC analysis; and (d) the mass spectra of urinary of the quercetin in foods exists in the form of glycosides; the analytes match the mass spectra of respective standards. The most commonly occurring glycoside of quercetin is rutin. Rutin detection of flavonoid metabolites in human urine samples is and other quercetin flavonols have similar metabolic profiles consistent with previous reports. Other investigators have found (63). Thus, multiple flavonols with an identical parent com- DHPAA and HVA in human urine samples (59, 60). Further- pound, quercetin, may yield several identical phenolic metab- more, similar concentrations of DHPAA and HVA are found in olites, such as HVA, DHPAA, and mHPAA. this and previous studies. Overall, the noted technical precautions for the assay are The HPLC-based method has several advantages over relatively minor, and the assay has numerous desirable charac- other analytical approaches. These advantages include that: (a) teristics. It has high sensitivity, precision, and analyte recover- the method does not require derivatization of analytes; (b) the ies. The assay is quantitative and relatively simple. Automation method is quantitative and sensitivity of the assay is high; (c) of several steps in the assay allows for the simultaneous anal- simultaneous extraction of multiple samples and automated ysis of multiple samples. These characteristics make the routine HPLC analysis minimize technician time requirements for the analysis of DHPAA, mHPAA, and HVA in human studies assay; (d) only small amounts of urine (20 ml) are needed for feasible. This assay is the first analytical method that provides the assay; (e) the assay is simple (it requires only two steps); (f) for the evaluation of DHPAA, mHPAA, and HVA as biomar- samples and standards are stable for at least 6 months with kers in human studies. It is a practical method for screening storage at -70°C; (g) unlike other assays, this assay measures human populations and the measurement of quercetin metabo- mHPAA; and (h) the assay simultaneously measures DHPAA, lites in human studies. mHPAA, and HVA. Several aspects of the assay require attention for the main- Acknowledgments tenance of high-precision analysis. (a) Determination of inter- We thank Dr. Edward Larka for the mass-spectral analysis of urinary DHPAA, nal standard recovery is essential. Low recoveries (<70%) mHPAA, and HVA. We are indebted to Michael Chaussee for performing several generally indicate inadequate performance of the extraction initial experiments for the development of this assay. procedure. Spurious data can result from analysis with low recoveries; i.e., p-coumaric acid may not be a perfect internal References standard. Thus, we recommend reanalysis of samples with recoveries below 70%; this should be a low percentage of the 1. Heimann, W., and Reiff, F. Beziehung zwischen chemischer Konstitution und antioxygener Wirkung bei Flavonolen. Fette Seifen, 55: 451-458, 1953. samples. (b) If chromatograms are complex, the identity of each 2. Pratt, D., and Watts, B. Antioxidant activity of vegetable extracts. I. Flavone analyte should be verified by UV/visible scans. The scans can aglycons. J. Food Sci., 29: 27-33, 1964. be done automatically with many scanning and photodiode 3. Clegg, K., and Morton, A. The phenolic compounds of blackcurrant juice and detectors. (c) Peak broadening will occur without the use of a their protective effect on ascorbic acid II. The stability of ascorbic acid in model column-cleaning procedure. That is, adequate column cleaning systems containing some of the phenolic compounds associated with blackcurrant is essential for optimal assay performance. juice. J. Food Technol., 3: 277-284, 1968. Application of the assay in studies of human populations 4. Lea, C., and Swoboda, P. On the antioxidant activity of the flavonols gos- sypetin and quercetagetin. Chem. Ind., 47: 1426-1428, 1956. requires consideration of concentration units. Significant day- 5. Masquelier, J. Action comparee de divers facteurs vitaminiques P sur to-day variations in urinary excretion volumes occur within and l'oxydation de l'acide ascorbique par les ions cuivriques. Bull. Soc. Chim. Biol., between subjects. Correction for urinary output is essential. 33: 302-303, 1951. Thus, urinary DHPAA, mHPAA, and HVA concentrations 6. Masquelier, J. Effets physiologiques des constituants non alcoliques du vin. should be adjusted for total urinary output or creatinine con- Cah. Nutr. Diet., 5: 57-64, 1970. centrations. Another concern for application of the assay is the 7. Wattenberg, L., and Leong, J. Inhibition of the carcinogenic action of 7,12- formation of DHPAA, HVA, and perhaps MHPAA from the dimethylbenz(a)anthracene by/3-naphthoflavone. Proc. Soc. Exp. Biol. Med., 28: metabolism of catecholamines and certain amino acids. For any 940-943, 1968. 8. Salup, R., Sicker, D., Wolmark, N., Herberman, R., and Hakala, T. Chemo- individual, these metabolic pathways produce a relatively con- immunotherapy of metastatic murine renal cell carcinoma using flavone acetic stant amount of DHPAA and HVA. Although dietary intake of acid and interleukin-2. J. Urol., 147: 1120-1123, 1992. quercetin can substantially increase DHPAA, mHPAA, and 9. Bohm, K. Die Flavonoide. Editio Cantor, Aulendorf, 1967. HVA over the amounts produced from catecholamines and 10. Nguyen, T. and Canada, A. Citrus flavonoids stimulate secretion by human amino acid metabolism, the best estimates of dietary determi- colonic Ts4 cells. J. Nutr., 123: 259-268, 1993. nants may involve experimental designs that incorporate meas- 11. Nguyen, T., Canada, A., Heintz, G., Gettys, T., and Cohn, J. Stimulation of urements at two or more time points and evaluate changes in secretion by the T84 colonic epithelial cell line with dietary flavonols. Biochem. quercetin metabolite excretion. Between individuals, age, gen- Pharmacol., 41: 1879-1886, 1991. der, drug use, body weight, and disease conditions may produce 12. Parellada, J., and Guinea, M. Flavonoid inhibitors of trypsin and leucine aminopeptidase: a proposed mathematical model for IC5o estimation. J. Nat. Prod. slight changes in HVA (62), and perhaps mHPAA and DHPAA (Lloydia), 58: 823-829, 1995. excretion, and should be adjusted for in population studies. 13. Walle, T., Eaton, E., and Walle, U. Quercetin, a potent and specific inhibitor Interpretation of results from human dietary studies should of the human P-form phenosulfotransferase. Biochem. Pharmacol., 50:731-734, allow for the presence of multiple glycosides of quercetin. Most 1995.

14. Constantinou, A., Mehta, R., Runyan, C., and Moon, R. Flavonoids as DNA 40. Zhu, B. T., and Liehr, J. G. Quercetin increases the severity of estradiol- topoisomerase antagonists and poisons: structure-activity relationships. J. Nat. induced tumorigenesis in hamster kidney. Toxicol. Appl. Pharmacol., 125: 149- Prod. (Lloydia), 58: 217-225, 1995. 158, 1994. 15. Singhal, R. L., Yeh, Y. A., Praja, N., Olah, E., Sledge, G. W., Jr., and Weber 41. Ambrose, A. M., Robbins, D. J.., and DeEds, F. Comparative toxicities of G. Quercetin down-regulates signal transduction in human breast carcinoma cells. quercetin and quercitrin. J. Am. Pharm. Assoc. Sci Ed., 41:119-122, 1952. Biochem. Biophys. Res. Commun., 208: 425-431, 1995. 42. Dunnick, J. K., and Hailey, J. R. Toxicity and carcinogenicity studies of 16. Tamura, M., Kagawa, S., and Tsuruo, Y. Effects of flavonoid compounds on quercetin, a natural component of foods. Fundam. Applied Toxicol., 19: 423- the activity of NADPH diaphorase prepared from the mouse brain. Jpn. J. Phar- 431, 1992. macol., 65: 371-373, 1994. 43. Steinmetz, K., and Potter, J. Vegetables, fruit, and cancer. I. Epidemiology. 17. Murakami S, Murametsu M, and Otomo S. Inhibition of gastric H+,K(+)- Cancer Causes & Control, 2: 325-357, 1991. ATPase by quercetin. J. Enzym. Inhib., 5: 293-298, 1992. 44. Gugler, R., Leschik, M., and Dengler, H. J. Disposition of quercetin in man 18. Chang, W. S., Lee, Y. J., Lu, F. J., and Chiang, H. C. Inhibitory effects of after single oral and intravenous doses. Eur. J. Clin. Pharmacol., 9: 229-234, flavonoids on xanthine oxidase. Anticancer Res., 13: 2165-2170, 1993. 1975. 19. Yoshida, M., Sakai, T., Hosokawa, N., Marui, N., Matsumoto, K., Fujioka, 45. Krishnamurty, H. G., Cheng, K. J., Jones, G. A., Simpson, F. J., and Watkin, A., Nishino, H, and Aoike, A. The effect of quercetin on cell cycle progression J. E. Identification of products produced by the anaerobic degradation of rutin and and growth of human gastric cancer cells. FEBS Lett., 260: 10-13, 1990. related flavonoids by Butyrivibrio sp. C3. Can. J. Microbiol., 16: 759-767, 1970. 20. Middleton, E., Jr., and Kandaswami, C. Effects of flavonoids on immune and 46. Ito, N. Is quercetin carcinogenic? (Letter). Jpn. J. Cancer Res., 83: 312-313, inflammatory cell functions. Biochem. Pharmacol., 43:1167-1179, 1992. 1992. 21. Bors, W., Heller, W., Michel, C., and Saran, M. Flavonoids as antioxidants: 47. Hirono, I. Is quercetin carcinogenic? (Letter). Jpn. J. Cancer Res., 83: determination of radical-scavenging efficiencies. Methods Enzymol., 186: 343- 313-314, 1992. 355, 1990. 48. Akagi, K., Hirose, M., Hoshiya, T., Mizoguchi, Y., Ito, N., and Shirai, T. 22. Walker, J. Flavonoid pigments in the skins of New Zealand apples. N. Z. J. Modulating effects of ellagic acid, vanillin and quercetin in a rat medium term Sci., 7: 585-588, 1964. multi-organ carcinogenesis model. Cancer Lett., 94:113-121, 1995. 23. Dick, D., Redden, P., DeMarco, A., Lidster, P., and Grindley, T. Flavonoid 49. Geahlen, R., Koonchanok, N., Mclaughlin, J., and Pratt, D. Inhibition of glycosides of spartan apple peel. J. Agric. Food Chem., 35: 529-531, 1987. protein-tyrosine kinase activity by flavonoids and related compounds. J. Nat. Prod. (Lloydia), 52: 982-986, 1989. 24. Ryan, J., and Coffin, D. Flavonol glucuronides from red raspberry, Rubus idaeus (rosaceae). Phytochemistry (Oxf.), 10: 1675-1677, 1971. 50. Steinmetz, K., and Potter, J. Vegetables, fruit, and cancer. II. Mechanisms. Cancer Causes & Control, 2: 427-442, 1991. 25. Shrikhande, A., and Francis, F. Flavonol glycosides of sour cherries. J. Food Sci., 38: 1035-1037, 1973. 51. Wattenberg, L., Lipkin, M., Boone, C., and Kelloff, G. Chemoprevention of cancer by naturally occurring and synthetic compounds. In: Cancer Prevention, 26. Ryan, J. Flavonol glycosides of the cultivated strawberry. J. Food Sci., 36: Chap. 2., pp. 19-34. Boca Raton, FL: CRC Press, 1992. 867-870, 1971. 52. Hollman, P. C. H., de Vries, J. H. M., van Leeuwen, S. D., Mengelers, M. 27. Herrmann, K. Occurrence and contents of flavonoids in fruit. Erwerbsobst- J. B., and Katan, M. B. Absorption of dietary quercetin glycosides and quercetin bau, 32: 32-37, 1990. in healthy ileostomy volunteers. Am. J. Clin. Nutr., 62: 1276-1282, 1995. 28. Duggan, M. Identity and occurrence of certain flavonol glycosides in four 53. Kuhnau, J. The flavonoids. A class of semi-essential food components: their varieties of pears. J. Agric. Food Chem. 17: 1098-1101, 1969. role in human nutrition. World Rev. Nutr. Diet., 24: 117-191, 1976. 29. Goupy, P., Varoquaux, P., Nicolas, J., and Macheix, J. Identification and 54. Booth, A., Murray C., Jones, F., and DEEDS, F. The metabolic fate of rutin localization of hydroxycinnamoyl and fiavonol derivatives from endive (Cicho- and quercetin in the animal body. J. Biol. Chem., 223: 251-257, 1956. rium endiva L. cv. geante maraichere) leaves. J. Agric. Food Chem., 38:2116- 2121, 1990. 55. DEEDS, F. Flavonoid metabolism. In: M. Florkin and E. H. Stotz (eds.), Comprehensive Biochemistry, Vol. 20, Chap. 2. Amsterdam: Elsevier, 1968. 30. Herrmann, K. Ueber den gehalt und die lokalisation der phenolischen in- haltsstoffe im gemuese. Plant Foods Hum. Nutr. 25: 231-246, 1976. 56. Murray, C., Booth, A., DEEDS, F., and Jones F. Absorption and metabolism of rutin and quercetin in the rabbit. J. Am. Pharm. Assoc., 43: 361-364, 1954. 31. Henning, W., and Herrmann, K. Flavonolglykoside der stisskirschens (Prunus avium L.) und der sauerkirschen (Prunus cerasus L. ) Z. Lebensm. Unters. Forsch., 57. Karoum, F., and Sandier, M. High-resolution gas chromatographic analysis 170: 433-444, 1980. of phenolic acids and alcohols using capillary columns. Clin. Chim. Aeta, 32: 391-397, 1971. 32. Williams, B., and Wender, S. The isolation and identification of kaempferol and quercetin from strawberries (fragaria chiloensis). Am. Chem. Soc. J., 74: 58. Karoum, F., Anah, C., Ruthven, C., and Sandier, M. Further observations on 5919-5920, 1952. the gas-chromatographic measurement of urinary phenolic and indolic metabolites. Clin. Chim. Acta, 24: 341-348, 1969. 33. Brown, J. A review of the genetic effects of naturally occurring flavonoids, anthraquinones and related compounds. Mutat. Res., 75: 243-277, 1980. 59. Takeda, N. Simultaneous analyses of monoamines and their metabolites in urine specimens of patients with neuroblastoma. Comp. Pharmacol. Toxicol., 3: 34. Hertog, M., Hollman, P., and Katan, M. Content of potentially anticarcino- 373-377, 1993. genic flavonoids of 28 vegetables and 9 fruits commonly consumed in the Netherlands. J. Agric. Food Chem., 40: 2379-2383, 1992. 60. Mashige, F., Ohkubo, A., Matsushima, Y., Takano, M., Tsuchiya, E., Kanazawa, H., Nagata, Y., Takai, N., Shinozuka, N., and Sakuma, I. High 35. Hertog, M., Hollman, P., and van de Putte, B. Content of potentially anti- performance liquid chromatographic determination of catecholamine metabolites carcinogenic flavonoids of tea infusions, wines, and fruit juices. J. Agric. Food and 5-hydroxyindoleacetic acid in human urine using a mixed-mode column and Chem., 41: 1242-1246, 1993. an eight-channel electrode electrochemical detector. J. Chromatogr. Biomed. 36. Hertog, M. Flavonols and flavones in foods and their relation with cancer and Appl., 658 (1): 63-68, 1994. coronary heart disease (Thesis). Bilthoven, the Netherlands: Agricultural Univer- 61. World Medical Assembly. The declaration of Helsinki: recommendations sity Wageningen, 1994. guiding medical doctors in biomedical research involving human subjects. 37. Suolinna, E. M., Buchsbaum, R. N., and Racker, E. The effect of flavonoids Adopted by the 18th World Medical Assembly, Helsinki, Finland, 1964; revised on aerobic glycolysis and growth of tumor cells. Cancer Res., 35: 1865-1872, by the 29th World Medical Assembly, Tokyo, Japan, 1975. 1975. 62. Cole, M., Craft, A., Parker, L., Bell, S., Sevious, J., McGill, A., and Dale, G. 38. Bjeldanes, L. F., and Chang, G. W. Mutagenic activity of quercetin and Urinary creatinine adjusted reference ranges for homovanillic and vanillylman- related compounds. Science (Washington DC), 197: 577-578, 1977. delic acid in children and adults. Clin. Chim. Acta, 1: 19-32, 1995. 39. Pamukcu, A. M., Yalciner, S., Hatcher, J. F., and Bryan, G. T. Quercetin, 63. Manach, C., Morand, C., Texier, O., Favier, M., Agulla, G., Demigne, C., a rat intestinal and bladder carcinogen present in bracken fern (Pteridium Regerat, F., and Remesy, C. Quercetin metabolites in plasma of rats fed diets aquilinum). Cancer Res., 40: 3468-3472, 1980. containing rutin or quercetin. J. Nutr., 125: 1911-1922, 1995.

M Gross, M Pfeiffer, M Martini, et al.

Cancer Epidemiol Biomarkers Prev 1996;5:711-720.

Updated version Access the most recent version of this article at: http://cebp.aacrjournals.org/content/5/9/711

E-mail alerts Sign up to receive free email-alerts related to this article or journal.

Reprints and To order reprints of this article or to subscribe to the journal, contact the AACR Publications Subscriptions Department at [email protected].

Permissions To request permission to re-use all or part of this article, use this link http://cebp.aacrjournals.org/content/5/9/711. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC) Rightslink site.