Cancer Causes Control DOI 10.1007/s10552-014-0401-7

ORIGINAL PAPER

Validity of a self-administered food frequency questionnaire in the estimation of heterocyclic aromatic amines Motoki Iwasaki • Tomomi Mukai • Ribeka Takachi • Junko Ishihara • Yukari Totsuka • Shoichiro Tsugane

Received: 28 November 2013 / Accepted: 13 May 2014 Ó Springer International Publishing Switzerland 2014

Abstract Background Clarification of the putative etiologic role of heterocyclic aromatic amines (HAAs) in the development of cancer requires a validated assessment tool for dietary HAAs. This study primarily aimed to evaluate the validity of a food frequency questionnaire (FFQ) in estimating HAA intake, using 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) level in human hair as the reference method. Methods We first updated analytical methods of PhIP using liquid chromatography-electrospray ionization/tandem mass spectrometry (LC-ESI/MS/MS) and measured

Electronic supplementary material The online version of this article (doi:10.1007/s10552-014-0401-7) contains supplementary material, which is available to authorized users. M. Iwasaki (&)  T. Mukai  R. Takachi  J. Ishihara Division of Epidemiology, Research Center for Cancer Prevention and Screening, National Cancer Center, 5-1-1 Tsukiji, Chuo-ku, Tokyo 104-0045, Japan e-mail: [email protected] R. Takachi Division of Social and Environmental Medicine, Department of Community Preventive Medicine, Niigata University Graduate School of Medical and Dental Sciences, Niigata, Japan J. Ishihara Department of Nutrition Management, Sagami Women’s University, Sagamihara, Kanagawa, Japan Y. Totsuka Division of Cancer Development System, National Cancer Center Research Institute, Tokyo, Japan S. Tsugane Research Center for Cancer Prevention and Screening, National Cancer Center, Tokyo, Japan

44 fur samples from nine rats from a feeding study as partverification of the quantitative performance of LC-ESI/MS/ MS. We next measured PhIP level in human hair samples from a validation study of the FFQ (n = 65). HAA intake from the FFQ was estimated using information on intake from six fish items and seven meat items and data on HAA content in each food item. Correlation coefficients between PhIP level in human hair and HAA intake from the FFQ were calculated. Results The animal feeding study of PhIP found a significant dose–response relationship between dosage and PhIP in rat fur. Mean level was 53.8 pg/g hair among subjects with values over the limit of detection (LOD) (n = 57). We found significant positive correlation coefficients between PhIP in human hair and HAA intake from the FFQ, with Spearman rank correlation coefficients of 0.35 for all subjects, 0.21 for subjects with over LOD values, and 0.34 for subjects with over limit of quantification. Conclusion Findings from the validation study suggest that the FFQ is reasonably valid for the assessment of HAA intake. Keywords Food frequency questionnaire  Heterocyclic aromatic amines  Validity Abbreviations CV CI 7,8-DiMeIQx 4,8-DiMeIQx dG-C8-PhIP

Coefficients of variation Confidence interval 2-Amino-3,7,8-trimethylimidazo [4,5-f]quinoxaline 2-Amino-3,4,8-trimethylimidazo [4,5-f]quinoxaline N2-(deoxyguanosine-8-yl)-2-amino1-methyl-6-phenylimidazo[4,5-b] pyridine

123

Cancer Causes Control

DR ESI FFQ HAAs IQ LC-ESI/MS LC-ESI/MS/MS LC/MS/MS LOD LOQ MeIQ MeIQx MRM PhIP psi SPE Trp-P-1

Dietary record Electrospray ionization Food frequency questionnaire Heterocyclic aromatic amines 2-Amino-3-methylimidazo [4,5-f]quinoline Liquid chromatography-electrospray ionization/mass spectrometry Liquid chromatography-electrospray ionization/tandem mass spectrometry Liquid chromatography tandem mass spectrometry Limit of detection Limit of quantification 2-Amino-3,4-dimethylimidazo [4,5-f]quinoline 2-Amino-3,8-dimethylimidazo [4,5-f]quinoxaline Multiple reaction monitoring 2-Amino-1-methyl-6-phenylimidazo [4,5-b]pyridine Pound per square inch Solid phase extraction 3-Amino-1,4-dimethyl-5H-pyrido [4,3-b]indole

Introduction Heterocyclic aromatic amines (HAAs), which are formed from the reaction of creatine or creatinine, amino acids, and sugars in meat and fish cooked at high temperatures, are mutagenic and carcinogenic in non-human primates [1– 3]. However, the findings of epidemiological studies that have specifically examined the association between HAA intake and cancer risk have been inconsistent [4–8]. This might partly reflect the difficulty in assessing dietary HAA intake, given that the HAA composition of cooked meat and fish varies according to cooking technique, temperature, cooking time, and meat type [9, 10]. In contrast, biomarkers reflect exposure to the human body without the measurement errors introduced by self-reported dietary intake. One possible specimen source for biological monitoring of individual exposure in humans is human hair [11, 12]. We previously established a method to measure 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) in human hair by liquid chromatography-electrospray ionization/mass spectrometry (LC-ESI/MS) [13] and used this measurement as reference level to validate a self-administered food frequency questionnaire (FFQ) in the estimation of dietary HAA intake [14]. One limitation of our

123

previous work is the poor selectivity of PhIP, which results from methodological issues related to LC-ESI/MS. In the present study, we applied this previous method to the quantification of HAAs by liquid chromatography-electrospray ionization/tandem mass spectrometry (LC-ESI/ MS/MS), which improves the selectivity of analytes and provides confirmation of the target analyte via the product ion spectrum. A second limitation is that the validation study included only 20 healthy volunteers. Middle-aged or older Japanese generally consume more fish than animal meat, and meat is more often prepared by chopping and stir-frying than grilling [15, 16]. In consequence, the foods contributing to HAA intake in Japan differ from those in western countries [17], and dietary assessment tools for HAA intake should be population specific. These findings also indicated the desirability of replicating our previous work in a larger number of subjects from the general population. The present study primarily aimed to evaluate the validity of an FFQ in estimating dietary HAA intake, using PhIP level in human hair as the reference method based on data from a validation study among examinees of the cancer screening program at the Research Center for Cancer Prevention and Screening, National Cancer Center, Japan. We first updated the analytical methods for PhIP in human hair by LC-ESI/MS/MS and verified quantitative performance of PhIP by investigating the dose–response relationship of PhIP level in rat fur samples from a feeding study.

Materials and methods Study population Study subjects were participants in a validation study of an FFQ used for a case–control study of colorectal adenoma in Tokyo [18, 19]. Details of this study have been described previously [20]. Briefly, subjects were selected from examinees of the cancer screening program at the Research Center for Cancer Prevention and Screening, National Cancer Center, Japan, from January 2004 through July 2006 who met the following criteria: (1) age between 40 and 69 years; (2) residence in Tokyo and suburban prefectures; and (3) no previous or present diagnosis of cancer, cardiovascular disease, or diabetes mellitus. Subjects were stratified by sex and age (40–49, 50–59, and 60–69 years) and randomly numbered for recruiting priority. Among the 896 candidates invited, 187 agreed to participate in the study (response rate: 20.9 %). After excluding those who could not attend the study orientation, 144 men and women provided weighed dietary records (DR) over four consecutive days; a self-administered semi-quantitative FFQ;

Cancer Causes Control

serum and EDTA-2Na plasma samples; 24-h urine samples; and hair samples between May 2007 and April 2008. The hair samples were collected in plastic ziplock bags and stored at 4 °C until processed. The study was approved by the Institutional Review Board of the National Cancer Center, Tokyo, Japan. All participants provided written informed consent for participation at the study orientation.

Dietary assessment for HAAs intake Details of the estimation of HAA intake from the FFQ have been reported elsewhere [17]. The FFQ consists of 138 food and beverage items with 9 frequency categories and standard portions/units, and were asked about the usual consumption of listed foods during the previous year. Frequency response choices for food items were less than once per month, 1–3 times per month, 1–2 times per week, 3–4 times per week, 5–6 times per week, once per day, 2–3 times per day, 4–6 times per day, and 7 or more times per day. Standard portion sizes were specified for each food item in three ‘‘amount’’ choices of small (50 % smaller), medium (standard), and large (50 % larger). Daily food intake was calculated by multiplying frequency by standard portion and relative size for each food item. For fish, the FFQ included 19 kinds of fresh or processed fish items or fish groups. Of these, HAA intake was estimated using salted fish, semi-dried split fish, salmon, horse mackerel or sardines, Pacific saury or mackerel, and eel of six groups, all of which are usually grilled. In addition, participants were asked about grilled skin consumption using a five-quantity category of almost none, one-third, half, two-thirds, to almost all. HAA intake from fish consumption was then estimated based on the proportion of grilled to total fish consumption, the rate of grilled skin consumption, the ratio of skin to flesh, and data on HAA content in the skin and flesh. For meat, HAA intake was estimated using 7 of 18 meat items in the FFQ: pan-fired and grilled beef, stir-fried pork and pork liver, grilled chicken and chicken liver, and bacon. Participants also reported preferred doneness levels (very well-done, well-done, medium, medium rare, and rare) for pan-fried and grilled beef. HAA intake from meat consumption was estimated based on this information and data on HAA content. Dietary intake of the following seven HAAs was calculated: 3-amino-1,4-dimethyl-5H-pyrido[4,3-b]indole (Trp-P-1), 2-amino-3-methylimidazo[4,5f]quinoline (IQ), 2-amino-3,4-dimethylimidazo[4,5-f]quinoline (MeIQ), 2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline (MeIQx), 2-amino-3,7,8-trimethylimidazo[4,5-f]quinoxaline (7,8-DiMeIQx), 2-amino-3,4,8-trimethylimidazo[4,5-f]quinoxaline (4,8-DiMeIQx), and PhIP. Total HAA intake was defined as the sum of the seven HAAs above.

Animal studies Male Brown Norway rats (6 weeks old) were purchased from Japan SLC (Shizuoka, Japan) and provided with food (CE-2 pellet diet, CLEA Japan, Inc., Tokyo, Japan) and tap water ad libitum and quarantined for one week. Rats were maintained under controlled conditions: 12-h light/dark cycle, 22 ± 2 °C room temperature, and 55 ± 10 % relative humidity. The experiments were conducted according to the ‘‘Guidelines for Animal Experiments in the National Cancer Center’’ of the Committee for Ethics of Animal Experimentation of the National Cancer Center. PhIP was added at a concentration of 20 or 200 ppm to CE-2 basal powder diet. Rats were housed 2–3 per plastic cage and quarantined for 1 week. Before PhIP administration, the back fur of rats was shaved with an animal electric shaver. Starting at 7 weeks of age, the rats were fed either a basal diet (n = 5) or experimental diets containing 20 or 200 ppm PhIP (n = 5 for each group) for 8 weeks and then changed to a basal diet for a further 8 weeks. After 4, 8, 12, and 16 weeks from the first treatment with PhIP, the back fur of all rats was shaved, and fur samples were pretreated in the same way as for human hair samples described below. For DNA adduct analysis, five rats of each group were killed every 4 weeks, and colons were excised and stored at -80 °C until DNA extraction. Body weights and diet consumption were recorded every 4 weeks. Determination of PhIP in human hair and rat fur samples Here, we briefly describe the analytical method for PhIP in human hair and rat fur. Details of laboratory analysis are included in the supplementary material. The washed and dried sample was weighed (1 g for human hair and 2 mg for rat fur), and 1 mol/L NaOH (100 mL) and internal standard ([2H3C]PhIP) (2 ng for human hair and 30 ng for rat fur) were added. This solution was incubated at 80 °C for 1 h. The sample was neutralized to pH 7–9 with 6 mol/L HCl, extracted using InertSepÒ SlimJ Aroma-Blue, and the eluate was concentrated under vacuum. After liquid–liquid extraction cleanup, the sample residue was dissolved in mobile phase (40 lM ammonium acetate (pH4.0): MeOH, 1:1, v/v) (500 lL for human hair and 750 lL for rat fur) and then filtrated. For human hair samples, an aliquot of 300 lL (1.2 ng of internal standard per injection) was injected into LC-ESI/MS/MS. For rat fur samples, the sample was diluted 10-fold with solvent and injected at 200/500 lL (0.8 ng of internal standard per injection) into the LC-ESI/MS/MS.

123

Cancer Causes Control

Chromatography was performed with a Shimadzu series VP model equipped with an Inertsil ODS-4 column (2.1 mm 9 15 cm, 5 lm particle size) at a temperature of 40 °C. Chromatographic separation was performed based on the previously described method [13]. In brief, the analytes were separated on a gradient with a columnswitching technique. The mass spectral data were obtained by ESI–MS using an API2000 (AB Sciex, Foster City, CA, USA) equipped with a modified TurboIonSprayÒ source. Analysis of processed samples was performed in positive ion mode, using multiple reaction monitoring (MRM) with the following transitions: PhIP and [2H3C]PhIP (precursor ion/product ion), 225.0/210.0, and 228.0/210.0. Quantification was conducted using internal standard calibration with [2H3C]PhIP. The calibration curve was established as the area ratio between analyte and internal standard versus added amount of analyte per sample. The standard solution was prepared by adding the standards, diluted in mobile phase (40 lM ammonium acetate (pH4.0): MeOH, 1:1, v/v), as follows: 0.027, 0.054, 0.1, 0.2, 0.4, 0.8 pg/lL for human hair sample and 0.2, 0.4, 0.8, 1.67, 3.33, 16.7, 33.3, 40.0 pg/lL for rat fur sample. The amount of internal standard was 4 pg/lL. The coefficient of determination values of the slopes of the calibration curves exceeded 0.99. The limit of detection (LOD) and the limit of quantification (LOQ) for the present analytical method were 13 and 45 pg/g hair, respectively.

PhIP level in human hair was natural log-transformed to produce approximately normal distributions. All measurement values of PhIP in human hair were used for the primary analyses, and values over the LOD or LOQ were used for additional analyses. HAA intake by the FFQ was adjusted for total energy intake by the residual method. First, we calculated mean PhIP level in human hair and mean intakes of HAAs according to the FFQ. Spearman rank correlation coefficients and 95 % confidence intervals were calculated between PhIP level in human hair and HAA intake according to the FFQ. Next, crude and energyadjusted values of HAA intake by the FFQ were divided into tertile categories. Adjusted geometric mean levels of PhIP in human hair according to tertile categories of dietary HAA intake were calculated using a multivariable linear regression model with adjustment for sex, age (continuous), and body mass index (continuous). A linear trend for mean levels was calculated in the multivariable linear regression model using categories of dietary intake levels as ordinal variables. In addition, to validate the categorization of subjects, we computed the number of subjects classified into the same, adjacent, and extreme categories by joint classification by quartile. All statistical analyses were performed with SAS software version 9.3 (SAS Institute, Inc., Cary, NC).

Results DNA adduct analysis We measured PhIP-DNA adduct levels in rat colon, which is a target organ for putative HAA-associated cancer. The method used to analyze DNA adducts is detailed in the supplementary material. Statistical analysis Mean PhIP level in rat fur was calculated according to dosage and time of fur collection. A linear trend for mean levels was calculated in the linear regression model using dosages as ordinal variables. We excluded subjects for whom hair samples were not available or who reported extremely low or high total energy intake by the FFQ (\800 or C4,000 kcal per day), leaving 139 men and women for analysis. To avoid interference related to unknown compounds, we excluded subjects whose hair was waved or who used hair dyes. In addition, we also excluded outliers of PhIP levels in human hair, defined as below or above a value equal to three times the interquartile range, leaving 50 men and 15 women for inclusion in the present analyses.

123

LC-ESI/MS/MS analysis of HAAs in human hair and rat fur To improve the quantitative performance of PhIP in human hair, we updated a previously established method of measuring PhIP in human hair by LC-ESI/MS/MS. Typical chromatograms of LC-ESI/MS/MS analyses of authentic PhIP are presented in Fig. 1a. A daughter fragment ion for PhIP was observed at m/z 210, and molecular weight transition (i.e., parent to primary daughter ion) was m/z 225 ? 210, as previously reported [21]. The product ion spectrum of the analyte obtained from the rat fur collected at 8 weeks at the dose of 200 ppm was in excellent agreement with the spectrum of the authentic PhIP and confirmed the identity of the analyte as PhIP (Fig. 1a). A peak corresponding to PhIP (retention time = 11.6 min for rat fur and 11.9 min for human hair) was readily found in the chromatogram of rat fur collected at 8 weeks at the dose of 200 ppm (Fig. 1b) and human hair (Fig. 1c). Figure 1d shows a representative example of chromatograms from a sample exposed to hair dye. The peak was low in intensity, and isobaric interference precluded reliable measurement.

Cancer Causes Control

Fig. 1 a Product ion spectra of 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) and the analyte isolated from rat fur; b liquid chromatography-electrospray ionization/tandem mass spectrometry (LC-ESI/MS/MS) chromatogram multiple reaction monitoring (MRM) traces of PhIP and [2H3C]PhIP isolated from rat fur

collected at 8 weeks; c LC-ESI/MS/MS chromatograms MRM traces of PhIP and [2H3C]PhIP isolated from human hair; d LC-ESI/MS/MS chromatograms MRM traces of PhIP and [2H3C]PhIP isolated from human hair which had been exposed to hair dye

Reproducibility of the analytical method

days (Supplementary Table 1). Intra- and inter-day coefficients of variation (CV) for rat fur samples were 6.1 and 10.3 %, respectively, and those for human hair samples were 9.8 and 10.0 %, respectively.

Reproducibility of the analytical method was assessed by measuring PhIP level of duplicate samples on four different

123

A

160 140

Control 20 ppm

120

200 ppm

100 80 60 40 20 0

B Adduct level (Adduct/106 nucleotides)

Fig. 2 Accumulation and disappearance of 2-amino-1methyl-6-phenylimidazo[4,5b]pyridine (PhIP) in rat fur (a) and N2-(deoxyguanosine-8yl)-2-amino-1-methyl-6phenylimidazo[4,5-b]pyridine (dG-C8-PhIP) levels in the colon (b) of rats fed a PhIPcontaining diet. Rats were treated with 20 or 200 ppm of PhIP for 8 weeks and then changed to a basal diet for the subsequent 8 weeks

PhIP amount (µg/g fur)

Cancer Causes Control

0 week 4 week 8 week 12 week 16 week

To verify the quantitative performance of the updated analytical method using rat fur samples with a controlled level of exposure to PhIP, we investigated the dose– response relationship between three exposure levels (control, 20 ppm, and 200 ppm) in fur collected at four different times (Supplementary Table 2 and Fig. 2a). A positive dose–response relationship was observed between dosage and PhIP accumulation in rat fur regardless of the time of fur collection. Mean PhIP levels were highest in fur collected at 8 weeks regardless of dosage, which corresponds to the end of the PhIP feeding period, although notably levels in fur collected at 16 weeks was somewhat higher than that in fur collected at 12 weeks at the dose of 200 ppm. Similar results were found when we used PhIP levels per melanin content (data not shown).

DNA adduct analysis To clarify whether the PhIP levels in rat fur samples represented the exposure level of PhIP in target tissue, we also examined PhIP-DNA adduct levels in the rat colon, a target organ for putative HAA-associated cancer. A typical chromatogram of LC-ESI/MS/MS analyses of dG-C8-PhIP detected in the colon obtained from a rat receiving 200 ppm of PhIP for 8 weeks is shown in Supplementary Fig. 1. Levels of dG-C8-PhIP observed in rats fed the diet containing 20 or 200 ppm of PhIP for 8 weeks were 1.15 ± 0.65 and 9.11 ± 3.65 per 106 nucleotides, respectively. Figure 2b shows the time course of PhIP-DNA adduct levels. When PhIP was given to rats for 8 weeks, dG-C8-PhIP levels gradually increased with an increase in feeding period, but when PhIP feeding was stopped, PhIPDNA adduct level rapidly decreased to 1–2 % over 4 weeks, consistent with the decrease in PhIP in rat fur samples (Fig. 2a).

123

12

20 ppm

10

200 ppm

8 6 4 2 0

4 week

Treatment with PhIP

Dose–response relationship between dosage and PhIP accumulated in rat fur

14

8 week 12 week 16 week

Treatment with PhIP

Table 1 Characteristics of study subjects Men (n = 50)

Women (n = 15)

Mean

Mean

95 % confidence interval

95 % confidence interval

Age (year)

58.9

(56.8, 61.0)

57.2

(53.4, 61.0)

Body mass index (kg/m2)

23.1

(22.4, 23.8)

21.2

(19.9, 22.5)

Energy (kcal)

2,108

(1,905, 2,311)

1,965

(1,685, 2,246)

Protein (g)

75.2

(66.2, 84.1)

77.9

(66.6, 89.2) (51.4, 75.5)

Fat (g)

65.2

(56.1, 74.3)

63.5

Carbohydrate (g)

265.9

(238.0, 293.9)

256.1

(219.2, 293.0)

Meats (g)

65

(49, 81)

55

(38, 73)

Processed meat (g)

7

(5, 9)

8

(5, 12)

Red meat (g)

41

(29, 53)

29

(18, 40)

Fish (g) HAA (ng)

75 44.1

(57, 93) (34.4, 53.8)

77 35.4

(56, 97) (25.6, 45.3)

PhIP (ng)

27.5

(21.3, 33.6)

20.8

(14.7, 26.8)

MeIQ (ng)

4.9

(3.9, 5.9)

4.3

(3.2, 5.5)

MeIQx (ng)

6.2

(4.8, 7.6)

4.8

(3.4, 6.1)

Trp-P-1 (ng)

3.1

(2.2, 4.1)

3.4

(1.7, 5.1)

7,8-DiMeIQx (ng)

1.8

(1.0, 2.6)

1.9

(1.0, 2.7)

4,8-DiMeIQx (ng)

0.6

(0.3, 0.9)

0.3

(0.1, 0.6)

IQ (ng)

0.04

(0.03, 0.1)

0.03

(0.02, 0.1)

HAA heterocyclic aromatic amines; PhIP 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine; MeIQ 2-amino-3,4-dimethylimidazo[4,5-f] quinoline; MeIQx 2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline; Trp-P-1 3-amino-1,4-dimethyl-5H-pyrido[4,3-b]indole; 7,8-DiMeIQx 2-amino-3,7,8-trimethylimidazo[4,5-f]quinoxaline; 4,8-DiMeIQx 2-amino-3,4,8-trimethylimidazo[4,5-f]quinoxaline; IQ 2-amino-3methylimidazo[4,5-f]quinoline

Comparison of PhIP level in human hair with dietary HAA intake estimated from the FFQ Characteristics of study subjects are presented in Table 1. Data from 50 men and 15 women were analyzed, with a mean age of 58.9 and 57.2 years, respectively. Mean meat and HAA intake tended to be higher for men (65 g and

Cancer Causes Control Table 2 Distribution of PhIP level in human hair samples Number of subjects

Mean

SD

Median

Interquartile range

Crude level (pg/g hair) All subjects Men and women

65

48.0

42.6

37.5

(22.0, 57.7)

Men

50

49.4

44.3

39.1

(22.0, 67.4)

Women

15

43.5

37.4

37.0

(18.4, 57.7)

Subjects with over LOD values Men and women

57

53.8

42.4

39.7

(27.0, 67.4)

Men

43

56.4

43.9

44.3

(27.0, 72.7)

Women

14

46.0

37.6

37.2

(25.7, 57.7)

Subjects with over LOQ values Men and 24 87.1 women

47.4

74.7

(56.9, 98.3)

Men Women

20

86.9

48.3

74.7

(55.1, 98.3)

4

88.4

49.3

67.4

(57.7, 119.0)

Level per melanin content (pg/mg melanin) All subjects Men and women

65

0.57

0.46

0.43

(0.27, 0.70)

Men

50

0.58

0.48

0.43

(0.27, 0.71)

Women

15

0.53

0.39

0.44

(0.27, 0.66)

Subjects with over LOD values Men and women

57

0.63

0.46

0.48

(0.32, 0.81)

Men

43

0.65

0.49

0.53

(0.32, 0.83)

Women

14

0.56

0.38

0.44

(0.31, 0.66)

0.49

0.86

(0.66, 1.16)

Subjects with over LOQ values Men and women

24

Men

20

0.96

0.52

0.76

(0.64, 1.16)

4

1.06

0.35

1.04

(0.82, 1.30)

Women

0.98

SD standard deviation, LOD limit of detection, LOQ limit of quantification

44.1 ng, respectively) than for women (55 g and 35.4 ng, respectively). Mean PhIP intake was highest among the 7 HAAs for both men and women. Since mean intake of 7,8DiMelQx, 4,8-DiMeIQx, and IQ was low, we did not use these for further analyses. PhIP level in human hair was over LOD in 57 and over LOQ in 24 of 65 subjects. Mean PhIP levels were 53.8 pg/g hair among subjects with over LOD values and 87.1 pg/g hair among subjects with over LOQ values (Table 2). Spearman rank correlation coefficients of HAA intake from the FFQ and PhIP level in human hair are shown in

Table 3. We found significant positive correlation coefficients between HAA intake (both crude and energyadjusted intake) and PhIP level (both crude level and level per melanin content) among all subjects (n = 65) regardless of type of HAA, except Trp-P-1. Correlation coefficients ranged from 0.25 to 0.38. Similarly, significant positive correlation coefficients were observed between energy-adjusted HAA intake and PhIP level (both crude level and level per melanin content) among men (n = 50) regardless of type of HAA, except Trp-P-1. We observed somewhat lower correlation coefficients among subjects with over LOD or LOQ values, and these were not statistically significant. For example, correlation coefficients (95 % confidence interval) between energy-adjusted PhIP intake and PhIP level per melanin content were 0.21 (0.05, 0.45) among subjects with over LOD values and 0.34 (-0.07, 0.66) among subjects with over LOQ values. Geometric mean levels of PhIP in human hair by tertile category of dietary HAAs intake are presented in Table 4. We found a significant positive association between HAA intake (both crude and energy-adjusted intake) and PhIP level (both crude level and level per melanin content) among all subjects (n = 65) regardless of type of HAA, except Trp-P-1. Among subjects with over LOD values, significant positive associations were observed between energy-adjusted PhIP, MeIQ, and MeIQx intake and PhIP level per melanin content only. Similarly, significant positive associations were seen between energy-adjusted PhIP intake and PhIP level per melanin content only among subjects with over LOQ. Table 5 shows a comparison of HAA intake with PhIP level in human hair based on joint classification by quartile. We calculated percentages of classification into the same, adjacent, and extreme categories by quartile among all subjects (n = 65) and those with over LOD values (n = 57) only. Percentages of classification into the same, adjacent, and extreme categories between energy-adjusted HAA intake and PhIP level per melanin content were 33.8, 75.4, and 4.6 %, respectively, among all subjects. Corresponding percentages among subjects with over LOD values were 31.6, 70.2, and 7.0 %, respectively.

Discussion In this study, we first confirmed that our updated analytical method quantified PhIP in human hair. Further, our animal feeding study of PhIP found a significant positive dose– response relationship between dosage and PhIP accumulation in rat fur, which suggests verification of the quantitative performance of the updated analytical method. In addition, the levels of dG-C8-PhIP in the colon, a target organ for putative HAA-associated cancer, revealed a similar pattern of PhIP amounts detected in rat fur. This

123

Cancer Causes Control Table 3 Spearman rank correlation coefficients (CCs) and 95 % confidence interval (CI) of HAA intake and PhIP level in human hair Number of subjects

Crude PhIP level in hair

PhIP level per melanin content in hair

Crude intake

Energy-adjusted intake

Crude intake

Energy-adjusted intake

CC

CC

CC

CC

95 % CI

95 % CI

95 % CI

95 % CI

HAA intake All subjects Men and women Men

65 50

0.29 0.23

(0.05, 0.50) (-0.05, 0.48)

0.36 0.33

(0.12, 0.55) (0.06, 0.56)

0.26 0.21

(0.02, 0.48) (-0.07, 0.46)

0.34 0.32

(0.11, 0.54) (0.04, 0.55)

Women

15

0.48

(-0.04, 0.80)

0.40

(-0.14, 0.76)

0.49

(-0.03, 0.80)

0.42

(-0.12, 0.77)

Subjects with over LOD values Men and women

57

0.14

(-0.13, 0.38)

0.20

(-0.07, 0.44)

0.15

(-0.11, 0.40)

0.22

(-0.05, 0.45)

Men

43

0.07

(-0.23, 0.36)

0.15

(-0.16, 0.43)

0.12

(-0.19, 0.40)

0.17

(-0.13, 0.45)

Women

14

0.36

(-0.21, 0.75)

0.29

(-0.29, 0.71)

0.38

(-0.19, 0.76)

0.31

(-0.26, 0.72)

Subjects with over LOQ values Men and women

24

0.16

(-0.26, 0.53)

0.15

(-0.27, 0.52)

0.25

(-0.17, 0.59)

0.23

(-0.19, 0.58)

Men

20

0.13

(-0.33, 0.54)

0.18

(-0.28, 0.58)

0.28

(-0.18, 0.64)

0.23

(-0.24, 0.61)

4

0.34

(-0.92, 0.98)

-0.53

(-0.99, 0.88)

0.75

(-0.76, 0.99)

0.00

(-0.96, 0.96)

Men and women

65

0.29

(0.05, 0.50)

0.35

(0.11, 0.54)

0.27

(0.03, 0.48)

0.35

(0.11, 0.54)

Men

50

0.24

(-0.04, 0.48)

0.34

(0.07, 0.57)

0.22

(-0.06, 0.47)

0.33

(0.06, 0.56)

0.45

(-0.08, 0.78)

0.33

(-0.22, 0.72)

0.45

(-0.08, 0.78)

0.37

(-0.17, 0.74)

Women PhIP intake All subjects

Women 15 Subjects with over LOD values Men and women

57

0.14

(-0.13, 0.38)

0.18

(-0.09, 0.42)

0.16

(-0.11, 0.40)

0.21

(-0.05, 0.45)

Men

43

0.09

(-0.22, 0.38)

0.15

(-0.16, 0.43)

0.13

(-0.18, 0.41)

0.18

(-0.12, 0.46)

Women

14

0.32

(-0.25, 0.73)

0.20

(-0.37, 0.66)

0.33

(-0.24, 0.73)

0.26

(-0.31, 0.70)

Subjects with over LOQ values Men and women

24

0.18

(-0.24, 0.54)

0.27

(-0.15, 0.61)

0.28

(-0.14, 0.61)

0.34

(-0.07, 0.66)

Men

20

0.17

(-0.30, 0.57)

0.27

(-0.20, 0.64)

0.31

(-0.16, 0.66)

0.30

(-0.16, 0.66)

4

0.34

(-0.92, 0.98)

0.34

(-0.92, 0.98)

0.75

(-0.76, 0.99)

0.75

(-0.76, 0.99)

Men and women

65

0.31

(0.07, 0.52)

0.38

(0.15, 0.57)

0.27

(0.03, 0.48)

0.36

(0.13, 0.56)

Men

50

0.28

(0.00, 0.52)

0.36

(0.09, 0.58)

0.24

(-0.04, 0.49)

0.33

(0.06, 0.56)

Women

15

0.40

(-0.14, 0.76)

0.47

(-0.06, 0.79)

0.46

(-0.07, 0.79)

0.55

(0.05, 0.83)

Women MeIQ intake All subjects

Subjects with over LOD values Men and women

57

0.14

(-0.12, 0.39)

0.22

(-0.04, 0.46)

0.14

(-0.12, 0.39)

0.24

(-0.02, 0.47)

Men Women

43 14

0.10 0.29

(-0.21, 0.39) (-0.28, 0.71)

0.19 0.37

(-0.12, 0.46) (-0.20, 0.75)

0.12 0.37

(-0.19, 0.41) (-0.20, 0.75)

0.20 0.47

(-0.10, 0.48) (-0.08, 0.80)

Subjects with over LOQ values Men and women

24

0.15

(-0.27, 0.52)

0.15

(-0.27, 0.52)

0.21

(-0.21, 0.57)

0.20

(-0.22, 0.56)

Men

20

0.19

(-0.27, 0.59)

0.24

(-0.23, 0.61)

0.30

(-0.16, 0.66)

0.26

(-0.21, 0.63)

4

-0.75

(-0.99, 0.76)

-0.75

(-0.99, 0.76)

-0.34

(-0.98, 0.92)

-0.34

(-0.98, 0.92)

Men and women

65

0.25

(0.01, 0.47)

0.30

(0.06, 0.51)

0.25

(0.01, 0.47)

0.32

(0.08, 0.52)

Men

50

0.19

(-0.09, 0.45)

0.28

(0.01, 0.52)

0.21

(-0.07, 0.46)

0.31

(0.03, 0.54)

Women

15

0.35

(-0.20, 0.73)

0.36

(-0.19, 0.73)

0.36

(-0.19, 0.74)

0.41

(-0.13, 0.76)

0.15

(-0.12, 0.39)

0.23

(-0.03, 0.46)

0.17

(-0.10, 0.41)

0.24

(-0.02, 0.47)

Women MeIQx intake All subjects

Subjects with over LOD values Men and women

123

57

Cancer Causes Control Table 3 continued Number of subjects

Crude PhIP level in hair

PhIP level per melanin content in hair

Crude intake

Energy-adjusted intake

Crude intake

Energy-adjusted intake

CC

CC

CC

CC

95 % CI

95 % CI

95 % CI

95 % CI

Men

43

0.10

(-0.20, 0.39)

0.23

(-0.07, 0.50)

0.15

(-0.16, 0.43)

0.24

(-0.06, 0.50)

Women

14

0.21

(-0.37, 0.66)

0.24

(-0.34, 0.68)

0.22

(-0.35, 0.67)

0.30

(-0.28, 0.72)

Subjects with over LOQ values Men and women

24

0.09

(-0.33, 0.48)

0.16

(-0.26, 0.53)

0.18

(-0.24, 0.54)

0.17

(-0.25, 0.53)

Men

20

0.08

(-0.38, 0.50)

0.11

(-0.35, 0.53)

0.19

(-0.27, 0.59)

0.10

(-0.36, 0.52)

4

0.34

(-0.92, 0.98)

0.34

(-0.92, 0.98)

0.75

(-0.76, 0.99)

0.75

(-0.76, 0.99)

Men and women

65

0.19

(-0.05, 0.42)

0.20

(-0.05, 0.42)

0.09

(-0.16, 0.33)

0.09

(-0.15, 0.33)

Men Women

50 15

0.20 0.13

(-0.08, 0.46) (-0.41, 0.61)

0.22 0.12

(-0.06, 0.47) (-0.42, 0.60)

0.11 0.04

(-0.17, 0.38) (-0.48, 0.54)

0.13 -0.01

(-0.15, 0.39) (-0.52, 0.50)

Women Trp-P-1 intake All subjects

Subjects with over LOD values Men and women

57

0.00

(-0.26, 0.26)

0.02

(-0.24, 0.28)

-0.05

(-0.30, 0.22)

-0.04

(-0.29, 0.23)

Men

43

0.01

(-0.29, 0.31)

0.04

(-0.26, 0.34)

0.01

(-0.29, 0.31)

0.03

(-0.28, 0.32)

Women

14

-0.04

(-0.56, 0.50)

-0.03

(-0.55, 0.51)

-0.16

(-0.64, 0.40)

-0.19

(-0.65, 0.38)

-0.08

(-0.47, 0.33)

-0.07

(-0.46, 0.34)

-0.12

(-0.50, 0.30)

Subjects with over LOQ values Men and women

24

-0.12

(-0.50, 0.30)

Men

20

-0.03

(-0.46, 0.42)

0.01

(-0.43, 0.45)

0.09

(-0.37, 0.51)

0.03

(-0.42, 0.47)

4

-0.34

(-0.98, 0.92)

-0.75

(-0.99, 0.76)

-0.75

(-0.99, 0.76)

-

-

Women

HAA heterocyclic aromatic amines; PhIP 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine; MeIQ 2-amino-3,4-dimethylimidazo[4,5-f]quinoline; MeIQx 2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline; Trp-P-1 3-amino-1,4-dimethyl-5H-pyrido[4,3-b]indole; LOD limit of detection; LOQ limit of quantification

finding suggests that hair samples can be used as a surrogate tissue for evaluation of exposure levels, as previously reported [12, 22]. Finally, as the primary aim of this study, a validation study observed significant positive correlation coefficients between PhIP in human hair and HAA intake from an FFQ. This suggests that our FFQ for middle-aged or older Japanese is reasonably valid for the assessment of HAA intake. We quantitated PhIP in human hair of 65 examinees of a cancer screening program using LC-ESI/MS/MS and showed a mean level of 53.8 pg/g hair among examinees with over LOD values. This mean level was lower than that measured by LC-ESI/MS in our previous work (1,376 pg/g hair) [13, 14]. This difference in mean level might be explained by the difference in analytical methods, LC-ESI/ MS/MS and LC-ESI/MS. Given that the product ion spectrum by LC-ESI/MS/MS verified the identity of the analyte as PhIP, the mean level by LC-ESI/MS might be overestimated, probably due to interference related to unknown compounds. In contrast, Bessette et al. [12] established a rapid and simple method for measuring PhIP in human hair which uses a tandem solvent/solid-phase extraction cleanup method to isolate PhIP. In a pilot study

of 12 volunteers, PhIP was detected among six meat-eaters at levels ranging from 290 to 890 pg/g hair [12], which is higher than the PhIP level in the present study. This difference is likely to reflect differences in HAA intake, resulting from simple differences in meat intake between Americans and Japanese; as an example, red meat intake (excluding processed meat) by the control group in a case– control study of colorectal adenoma in Hawaii was 54.9 g per day among men, versus 41 g per day among men in our present study [23]. The impact of different analytical methods on the difference in analytical values remains unknown, and an effect cannot be ruled out. We found a significant positive dose–response relationship between dosage and PhIP accumulated in rat fur regardless of time of fur collection. Given that the animal feeding study controlled the level of exposure to PhIP, the observed dose–response relationship indicates that our analytical method is valid. Further, levels decreased over time after feeding PhIP at the dose of 20 ppm was stopped, albeit that levels were somewhat higher in fur collected at 16 than at 12 weeks at 200 ppm. The reason for this unexpected pattern is unclear but might be due to chance. On the other hand, a relatively rapid increase in PhIP level

123

Cancer Causes Control Table 4 Adjusted geometric mean (GM)a and 95 % confidence interval (CI) of PhIP level in human hair according to HAA intake Crude intake

Energy-adjusted intake

Crude PhIP level (pg/g hair) GM

PhIP level per melanin content (pg/mg melanin)

Crude PhIP level (pg/g hair)

PhIP level per melanin content (pg/mg melanin)

95 % CI

p for trend

GM

95 % CI

p for trend

GM

95 % CI

p for trend

\0.01

\0.01

\0.01

GM

95 % CI

p for trend

\0.01

All subjects (n = 65) HAA Lowest

22.5

(14.5, 34.7)

0.28

(0.18, 0.43)

19.5

(13.1, 28.9)

0.25

(0.17, 0.37)

Middle

33.3

(22.5, 49.3)

0.41

(0.27, 0.61)

37.3

(24.7, 56.5)

0.40

(0.26, 0.61)

Highest

50.2

(33.1, 76.2)

0.59

(0.38, 0.90)

54.0

(36.8, 79.2)

0.68

(0.46, 1.01)

Lowest

21.9

(14.2, 33.7)

0.28

(0.18, 0.44)

22.0

(14.7, 33.1)

0.27

(0.18, 0.41)

Middle

33.0

(22.7, 47.9)

0.38

(0.26, 0.56)

34.5

(23.1, 51.6)

0.40

(0.27, 0.60)

Highest

56.0

(36.3, 86.2)

0.68

(0.44, 1.06)

52.9

(35.2, 79.5)

0.65

(0.43, 0.99)

PhIP \0.01

\0.01

\0.01

MeIQ Lowest

24.5

(15.9, 37.7)

0.32

(0.20, 0.49)

19.2

(12.9, 28.5)

0.24

(0.16, 0.37)

Middle

30.3

(20.3, 45.2)

0.36

(0.24, 0.55)

38.8

(25.2, 59.7)

0.46

(0.29, 0.72)

Highest

52.3

(34.6, 79.0)

0.61

(0.40, 0.93)

53.3

(36.6, 77.7)

0.62

(0.42, 0.92)

\0.01

0.02

\0.01

\0.01

\0.01

MeIQx Lowest

27.3

(17.6, 42.3)

0.32

(0.21, 0.50)

24.7

(16.1, 38.0)

0.29

(0.19, 0.44)

Middle

29.0

(19.6, 43.0)

0.02

0.36

(0.24, 0.53)

0.02

32.6

(21.4, 49.5)

0.02

0.38

(0.25, 0.57)

Highest

53.3

(34.4, 82.4)

0.64

(0.41, 1.00)

48.1

(31.7, 73.0)

0.63

(0.42, 0.96)

Lowest

26.8

(17.3, 41.4)

0.35

(0.22, 0.54)

24.6

(15.8, 38.2)

0.32

(0.20, 0.50)

Middle

37.3

(24.1, 57.8)

0.47

(0.30, 0.74)

38.8

(25.3, 59.5)

0.48

(0.31, 0.74)

Highest

40.3

(26.0, 62.3)

0.44

(0.28, 0.68)

41.4

(27.0, 63.3)

0.46

(0.29, 0.70)

0.44

(0.30, 0.63)

36.2

(25.8, 50.7)

0.47 0.58

(0.34, 0.65) (0.40, 0.83)

39.3 48.7

(27.8, 55.6) (35.3, 67.0)

\0.01

Trp-P-1 0.17

0.44

0.09

0.26

All subjects with over LOD values (n = 57) HAA Lowest

39.3

(27.9, 55.4)

Middle Highest

37.6 49.5

(27.8, 50.8) (35.2, 69.8)

0.29

0.24

0.19

0.43

(0.30, 0.62)

0.43 0.61

(0.30, 0.62) (0.43, 0.86)

Lowest

43.8

(31.1, 61.6)

0.49

(0.34, 0.71)

34.5

(25.0, 47.4)

0.41

(0.29, 0.57)

Middle

34.6

(25.7, 46.7)

0.44

(0.32, 0.60)

39.1

(28.5, 53.7)

0.44

(0.32, 0.62)

Highest

50.4

(35.9, 70.7)

0.59

(0.41, 0.85)

51.9

(37.8, 71.3)

0.65

(0.47, 0.91)

Lowest

38.8

(27.8, 54.2)

0.44

(0.31, 0.63)

35.2

(25.5, 48.4)

0.39

(0.28, 0.54)

Middle

39.7

(29.0, 54.5)

0.50

(0.36, 0.70)

38.3

(27.5, 53.4)

0.48

(0.34, 0.68)

Highest

45.3

(32.6, 62.8)

0.52

(0.37, 0.74)

51.8

(37.6, 71.2)

0.64

(0.45, 0.89)

Lowest

36.8

(26.0, 51.9)

0.41

(0.28, 0.59)

34.3

(24.8, 47.5)

0.40

(0.28, 0.56)

Middle

42.0

(31.1, 56.8)

0.51

(0.37, 0.70)

38.6

(28.3, 52.6)

0.43

(0.31, 0.60)

Highest

44.6

(31.5, 63.1)

0.55

(0.38, 0.80)

52.4

(38.2, 71.9)

0.68

(0.49, 0.94)

Lowest

39.2

(28.0, 54.9)

0.47

(0.33, 0.67)

41.0

(29.4, 57.2)

Middle Highest

46.4 38.1

(33.7, 63.8) (27.4, 52.9)

0.56 0.44

(0.40, 0.79) (0.31, 0.63)

40.9 41.6

(29.1, 57.5) (30.1, 57.4)

0.14

PhIP 0.51

0.43

0.06

0.04

MeIQ 0.48

0.48

0.08

0.03

MeIQx 0.38

0.21

0.05

0.02

Trp-P-1

123

0.86

0.78

0.95

0.49

(0.35, 0.70)

0.51 0.47

(0.36, 0.74) (0.33, 0.66)

0.79

Cancer Causes Control Table 4 continued Crude intake

Energy-adjusted intake

Crude PhIP level (pg/g hair) GM

95 % CI

p for trend

PhIP level per melanin content (pg/mg melanin)

Crude PhIP level (pg/g hair)

GM

GM

95 % CI

p for trend

0.36

PhIP level per melanin content (pg/mg melanin)

95 % CI

p for trend

0.56

GM

95 % CI

p for trend

0.13

All subjects with over LOQ values (n = 24) HAA Lowest

75.2

(49.1, 115.1)

0.75

(0.50, 1.11)

80.4

(53.0, 121.9)

0.84

(0.56, 1.26)

Middle

95.1

(69.3, 130.3)

0.49

1.20

(0.90, 1.61)

92.2

(64.9, 130.9)

1.03

(0.73, 1.45)

Highest

88.7

(60.7, 129.6)

0.95

(0.67, 1.35)

92.6

(64.5, 133.0)

1.19

(0.84, 1.69)

Lowest

75.2

(49.1, 115.1)

0.75

(0.50, 1.11)

78.0

(52.7, 115.4)

0.90

(0.63, 1.29)

Middle

95.1

(69.3, 130.3)

1.20

(0.90, 1.61)

83.9

(62.5, 112.5)

0.92

(0.70, 1.20)

Highest

88.7

(60.7, 129.6)

0.95

(0.67, 1.35)

113.7

(79.5, 162.6)

1.47

(1.06, 2.04)

Lowest

75.8

(50.2, 114.5)

0.79

(0.53, 1.18)

77.5

(50.3, 119.5)

0.81

(0.53, 1.24)

Middle

102.4

(75.3, 139.4)

1.22

(0.90, 1.64)

91.8

(64.9, 129.8)

1.07

(0.76, 1.50)

Highest

77.5

(53.8, 111.7)

0.91

(0.63, 1.29)

91.8

(63.1, 133.4)

1.10

(0.76, 1.59)

Lowest

84.8

(56.6, 127.0)

0.86

(0.58, 1.27)

73.9

(48.9, 111.8)

0.83

(0.54, 1.25)

Middle

90.7

(64.8, 126.8)

1.17

(0.84, 1.62)

91.9

(64.7, 130.5)

1.03

(0.73, 1.47)

Highest Trp-P-1

91.5

(62.3, 134.4)

1.00

(0.69, 1.46)

95.6

(68.2, 134.0)

1.15

(0.82, 1.62)

Lowest

99.5

(68.3, 144.9)

1.07

(0.72, 1.60)

103.3

(71.8, 148.7)

1.13

(0.77, 1.66)

Middle

94.2

(66.8, 132.9)

1.06

(0.73, 1.53)

86.5

(58.1, 128.8)

0.99

(0.65, 1.51)

Highest

77.2

(54.1, 110.1)

0.98

(0.67, 1.42)

78.0

(54.5, 111.5)

0.97

(0.66, 1.41)

PhIP 0.49

0.36

0.06

0.01

MeIQ 0.95

0.59

0.45

0.18

MeIQx 0.73

0.23

0.53

0.66

0.26

0.22

0.15

0.52

HAA heterocyclic aromatic amines; PhIP 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine; MeIQ 2-amino-3,4-dimethylimidazo[4,5-f]quinoline; MeIQx 2-amino-3,8-dimethylimidazo[4,5-f]quinoxaline; Trp-P-1 3-amino-1,4-dimethyl-5H-pyrido[4,3-b]indole; LOD limit of detection; LOQ limit of quantification a

Adjusted for sex, age, and body mass index

between 0 and 4 weeks and rapid decline between 8 and 12 weeks were observed. In the present animal study, the back fur of rats was shaved completely every 4 weeks from the first treatment with PhIP, meaning that PhIP in fur reflected exposure to PhIP in the past 4 weeks. This rapid increase and decline indicates that PhIP is rapidly absorbed and eliminated after exposure. This does not necessarily indicate that PhIP level in hair is a short-term biomarker since the setting of the animal study differs from that of human hair collected from non-vegetarians. Namely, relatively long hair is assumed to reflect repeated exposure to PhIP over an extended period of time, as reflective of the daily life and dietary habits of meat-eaters, and PhIP level in hair is therefore considered to represent cumulative exposure to PhIP over that period. We found significant positive correlation coefficients between PhIP in human hair and HAA intake from the FFQ, which suggests that the FFQ is reasonably valid for the assessment of HAA intake. Nevertheless, the

correlation coefficients were lower than those in our previous validation study [14]. For example, Spearman rank correlation coefficients in the previous study were 0.47 between PhIP level per melanin content in human hair and PhIP intake and 0.51 between PhIP level per melanin content in human hair and HAA intake, versus respective values of 0.35 and 0.34 in the present study, albeit that these were still statistically significant. In addition, the correlation coefficients were somewhat lower among subjects with over LOD or LOQ values and were not statistically significant. Although the reason for these findings is unclear, they might be a chance finding associated with the relatively small number of subjects. In addition to the difference in analytical methods, assessment of HAA intake might also have affected the difference in findings between the two studies. The FFQ used in our present study, which was initially developed for the Japan Public Health Center (JPHC)-based prospective study, was modified for a middle-aged urban population. In particular, 10

123

123

36.9

33.8 35.4

32.3

PhIP

MeIQ MeIQx

Trp-P-1

66.2

73.8 66.2

72.3

73.8

Adjacent categoryb (%)

6.2

6.2 6.2

4.6

4.6

Extreme categoryc (%)

36.8

29.8

40.4

19.3

PhIP

MeIQ

MeIQx

Trp-P-1

59.6

71.9

70.2

66.7

68.4

12.3

10.5

12.3

8.8

10.5

29.8

35.1

21.1

28.1

22.8

33.8

26.2 35.4

35.4

35.4

Same categorya (%)

57.9

71.9

70.2

70.2

70.2

64.6

73.8 64.6

72.3

72.3

Adjacent categoryb (%)

14.0

8.8

10.5

7.0

8.8

9.2

7.7 7.7

6.2

6.2

Extreme categoryc (%)

19.3

31.6

28.1

28.1

29.8

24.6

36.9 29.2

35.4

35.4

Same categorya (%)

64.9

78.9

71.9

68.4

71.9

70.8

75.4 75.4

72.3

75.4

Adjacent categoryb (%)

14.0

12.3

8.8

12.3

10.5

9.2

4.6 6.2

3.1

3.1

Extreme categoryc (%)

17.5

36.8

26.3

33.3

31.6

27.7

38.5 36.9

30.8

33.8

Same categorya (%)

63.2

73.7

70.2

68.4

70.2

66.2

75.4 75.4

75.4

75.4

Adjacent categoryb (%)

14.0

8.8

5.3

7.0

7.0

10.8

6.2 7.7

4.6

4.6

Extreme categoryc (%)

PhIP level per melanin content (pg/mg melanin)

c

b

a

Subjects were classified into the opposite extreme categories between intake from food frequency questionnaire and PhIP level in human hair

Subjects were classified into the same categories or the adjacent categories between intake from food frequency questionnaire and PhIP level in human hair

Subjects were classified into the same categories between intake from food frequency questionnaire and PhIP level in human hair

HAA Heterocyclic aromatic amines; PhIP 2-Amino-1-methyl-6-phenylimidazo[4,5-b]pyridine; MeIQ 2-Amino-3,4-dimethylimidazo[4,5-f]quinoline; MeIQx 2-Amino-3,8-dimethylimidazo[4,5-f]quinoxaline; Trp-P-1 3-Amino-1,4-dimethyl-5H-pyrido[4,3-b]indole; LOD limit of detection

33.3

HAA

All subjects with over LOD values (n=57)

35.4

HAA

All subjects (n = 65)

Same categorya (%)

Crude PhIP level (pg/g hair)

Crude PhIP level (pg/g hair)

PhIP level per melanin content (pg/mg melanin)

Energy-adjusted intake

Crude intake

Table 5 Comparison of HAA intake with PhIP level in human hair based on joint classification by quartile (%)

Cancer Causes Control

Cancer Causes Control

foods mainly consumed in specific areas or at specific times were excluded (luncheon meats, vivipara, qing-gengcai [bok choy], leaf mustard, bitter gourd, chard, loofah, mugwort, yushi-tofu [soft, boiled tofu], and Okinawa soba) and 11 foods consumed throughout the year in urban areas were added (beef, stir-fried; chicken, stir-fried; chicken, stew; low-fat milk; Japanese amberjack; Welsh onion; eggplant; edible burdock; konnyaku foods [devil’s tongue]; nama-age [fried slices of drained tofu]; and jam, strawberry, or marmalade) [20]. Given that HAAs were detected in stir-fried pork [24] and that more than 50 % of total HAA intake was derived from pork in Okinawa [17], stirfried beef and chicken might contribute to HAA intake. In this study population, the contribution of stir-fried beef and chicken intake to meat intake from nine items used for HAA estimation is \20 %, and the contribution to HAA intake might be accordingly relatively small if HAA content in stir-fried beef and chicken are comparable to other meats. Nevertheless, measurement errors due to lack of information on HAA intake from stir-fried beef and chicken cannot be ruled out. Since the HAA content of stirfried beef and chicken is to our knowledge unavailable, we assigned HAA values in stir-fried pork to stir-fried beef and chicken and compared PhIP level in human hair with dietary HAA intake estimated from the FFQ as a sensitive analysis. In general, Spearman rank correlation coefficients between HAA intake and PhIP levels were somewhat lower in the sensitivity analyses than in the main analyses, although significant positive associations were observed, in particular for energy-adjusted HAA, PhIP, and MeIQ intake and PhIP level (both crude level and level per melanin content) among all subjects (data not shown). This suggests that improving the validity of the FFQ in estimating HAA intake requires food-specific HAA contents. Another potential explanation for the difference in findings is differences in study populations: The previous study was conducted in volunteers aged 25–57 years versus examinees of a cancer screening program aged 40–69 years in the present study. Although the present study population was well-characterized, participants were randomly selected and recruited from examinees of a cancer screening program, and the proportion of response was not necessarily high. This might have resulted in a higher proportion of health-conscious subjects than in the actual population, and health-conscious subjects might somewhat differ regarding the type and amount of meat and fish, and cooking methods. Another factor related to study population was the difference in hair color among subjects. Although the present studies excluded subjects who used hair dyes, the shading of black hair varied, and some samples included a degree of gray hair. Considering that melanin level might be related to hair color and that PhIP has strong affinity for melanin [25], the influence of

misclassification due to hair color might be partially reduced by adjustment for melanin. However, we cannot exclude the possibility of misclassification due to hair color, and this remains a challenge to the assessment of PhIP exposure using hair samples. In conclusion, our updated analytical method quantitates PhIP in human hair and thereby allows the use of hair in the assessment of individual exposure level as a long-term biomarker of HAAs. Our findings from the validation study suggest that the FFQ is reasonably valid for the assessment of HAA intake, which allows its use in future epidemiological studies of the association between HAA intake and cancer risk by ranking individuals by dietary intake of HAAs. Acknowledgments We thank Dr. Loic Le Marchand (Cancer Research Center of Hawaii, University of Hawaii), Dr. Robert J. Turesky (New York State Department of Health), Dr. Hiroyuki Kataoka (School of Pharmacy, Shujitsu University), and Dr. Hiroaki Itoh (Juntendo University Faculty of Medicine) for their helpful advice on the assay of HAA in hair. This study was supported by National Cancer Centre Research and Development Fund, a Grant-in-Aid for Scientific Research (C) (24501366) from the Japan Society for the Promotion of Science, and a Grant-in-Aid for the Third-Term Comprehensive Ten-Year Strategy for Cancer Control from the Ministry of Health, Labour and Welfare of Japan. Conflict of interest

The authors have no conflict of interest.

References 1. Sugimura T, Wakabayashi K, Nakagama H, Nagao M (2004) Heterocyclic amines: mutagens/carcinogens produced during cooking of meat and fish. Cancer Sci 95:290–299 2. Hasegawa R, Sano M, Tamano S et al (1993) Dose-dependence of 2-amino-1-methyl-6-phenylimidazo[4,5-b]-pyridine (PhIP) carcinogenicity in rats. Carcinogenesis 14:2553–2557 3. Shirai T, Sano M, Tamano S et al (1997) The prostate: a target for carcinogenicity of 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) derived from cooked foods. Cancer Res 57:195–198 4. Sinha R, Peters U, Cross AJ et al (2005) Meat, meat cooking methods and preservation, and risk for colorectal adenoma. Cancer Res 65:8034–8041 5. Wu K, Giovannucci E, Byrne C et al (2006) Meat mutagens and risk of distal colon adenoma in a cohort of U.S. men. Cancer Epidemiol Biomarkers Prev 15:1120–1125 6. Ollberding NJ, Wilkens LR, Henderson BE, Kolonel LN, Le Marchand L (2012) Meat consumption, heterocyclic amines and colorectal cancer risk: the Multiethnic Cohort Study. Int J Cancer 131:E1125–E1133 7. Cross AJ, Peters U, Kirsh VA et al (2005) A prospective study of meat and meat mutagens and prostate cancer risk. Cancer Res 65:11779–11784 8. Daniel CR, Cross AJ, Graubard BI et al (2012) Large prospective investigation of meat intake, related mutagens, and risk of renal cell carcinoma. Am J Clin Nutr 95:155–162 9. Skog KI, Johansson MA, Jagerstad MI (1998) Carcinogenic heterocyclic amines in model systems and cooked foods: a review on formation, occurrence and intake. Food Chem Toxicol 36:879–896

123

Cancer Causes Control 10. Iwasaki M, Kataoka H, Ishihara J et al (2010) Heterocyclic amines content of meat and fish cooked by Brazilian methods. J Food Compost Anal 23:61–69 11. Kobayashi M, Hanaoka T, Hashimoto H, Tsugane S (2005) 2-Amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) level in human hair as biomarkers for dietary grilled/stir-fried meat and fish intake. Mutat Res 588:136–142 12. Bessette EE, Yasa I, Dunbar D, Wilkens LR, Le Marchand L, Turesky RJ (2009) Biomonitoring of carcinogenic heterocyclic aromatic amines in hair: a validation study. Chem Res Toxicol 22:1454–1463 13. Hashimoto H, Hanaoka T, Kobayashi M, Tsugane S (2004) Analytical method of 2-amino-1-methyl-6-phenylimidazo[4,5b]pyridine in human hair by column-switching liquid chromatography-mass spectrometry. J Chromatogr B Analyt Technol Biomed Life Sci 803:209–213 14. Kobayashi M, Hanaoka T, Tsugane S (2007) Validity of a selfadministered food frequency questionnaire in the assessment of heterocyclic amine intake using 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) levels in hair. Mutat Res 630:14–19 15. Tsugane S, Sasaki S, Kobayashi M, Tsubono Y, Sobue T (2001) Dietary habits among the JPHC study participants at baseline survey. Japan Public Health Center-based Prospective Study on Cancer and Cardiovascular Diseases. J Epidemiol 11:S30–S43 16. Sasaki S, Kobayashi M, Tsugane S (2003) Validity of a selfadministered food frequency questionnaire used in the 5-year follow-up survey of the JPHC Study Cohort I: comparison with dietary records for food groups. J Epidemiol 13:S57–S63 17. Kobayashi M, Hanaoka T, Nishioka S, Kataoka H, Tsugane S (2002) Estimation of dietary HCA intakes in a large-scale population-based prospective study in Japan. Mutat Res 506–507: 233–241 18. Otani T, Iwasaki M, Ikeda S et al (2006) Serum triglycerides and colorectal adenoma in a case–control study among cancer

123

19.

20.

21.

22.

23.

24.

25.

screening examinees (Japan). Cancer Causes Control 17: 1245–1252 Yamaji T, Iwasaki M, Sasazuki S, Sakamoto H, Yoshida T, Tsugane S (2012) Association between plasma 25-hydroxyvitamin D and colorectal adenoma according to dietary calcium intake and vitamin D receptor polymorphism. Am J Epidemiol 175:236–244 Takachi R, Ishihara J, Iwasaki M et al (2011) Validity of a selfadministered food frequency questionnaire for middle-aged urban cancer screenees: comparison with 4-day weighed dietary records. J Epidemiol 21:447–458 Goodenough AK, Schut HA, Turesky RJ (2007) Novel LC-ESI/ MS/MS(n) method for the characterization and quantification of 20 -deoxyguanosine adducts of the dietary carcinogen 2-amino-1methyl-6-phenylimidazo[4,5-b]pyridine by 2-D linear quadrupole ion trap mass spectrometry. Chem Res Toxicol 20:263–276 Steffensen IL, Janak K, Hegstad S et al (2003) Incorporation of the food mutagen 2-amino-1-methyl-6-phenylimidazo[4,5-b]pyridine (PhIP) into fur and correlation with intestinal tumourigenesis in Min/?mice. Pharmacol Toxicol 92:131–136 Voutsinas J, Wilkens LR, Franke A et al (2013) Heterocyclic amine intake, smoking, cytochrome P450 1A2 and N-acetylation phenotypes, and risk of colorectal adenoma in a multiethnic population. Gut 62:416–422 Kataoka H, Nishioka S, Kobayashi M, Hanaoka T, Tsugane S (2002) Analysis of mutagenic heterocyclic amines in cooked food samples by gas chromatography with nitrogen-phosphorus detector. Bull Environ Contam Toxicol 69:682–689 Hegstad S, Reistad R, Haug LS, Alexander J (2002) Eumelanin is a major determinant for 2-amino-1-methyl-6-phenylimidazo[4,5b]pyridine (PhIP) incorporation into hair of mice. Pharmacol Toxicol 90:333–337