SEARCH

SEARCH BY CITATION

Keywords:

  • acrylamide;
  • diet;
  • prostate cancer

Abstract

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgements
  8. References

Acrylamide, a probable human carcinogen, is formed during the cooking of many commonly consumed foods. Data are scant on whether dietary acrylamide represents an important cancer risk in humans. We studied the association between acrylamide and prostate cancer risk using 2 measures of acrylamide exposure: intake from a food frequency questionnaire (FFQ) and acrylamide adducts to hemoglobin. We also studied the correlation between these 2 exposure measures. We used data from the population-based case-control study Cancer of the Prostate in Sweden (CAPS). Dietary data was available for 1,499 cases and 1,118 controls. Hemoglobin adducts of acrylamide were measured in blood samples from a subset of 170 cases and 161 controls. We calculated odds ratios (ORs) for the risk of prostate cancer in high versus low quantiles of acrylamide exposure using logistic regression. The correlation between FFQ acrylamide intake and acrylamide adducts in non-smokers was 0.25 (95% confidence interval: 0.14–0.35), adjusted for age, region, energy intake, and laboratory batch. Among controls the correlation was 0.35 (95% CI: 0.21–0.48); among cases it was 0.15 (95% CI: 0.00–0.30). The OR of prostate cancer for the highest versus lowest quartile of acrylamide adducts was 0.93 (95% CI: 0.47–1.85, p-value for trend = 0.98). For FFQ acrylamide, the OR of prostate cancer for the highest versus lowest quintile was 0.97 (95% CI: 0.75–1.27, p trend = 0.67). No significant associations were found between acrylamide exposure and risk of prostate cancer by stage, grade, or PSA level. Acrylamide adducts to hemoglobin and FFQ-measured acrylamide intake were moderately correlated. Neither measure of acrylamide exposure—hemoglobin adducts or FFQ—was associated with risk of prostate cancer. © 2008 Wiley-Liss, Inc.

In 2002, researchers at Stockholm University reported that high levels of acrylamide are formed during the cooking of many commonly consumed foods that are prepared at high temperatures.1 This finding was confirmed by others, and acrylamide has now been detected in a diverse set of foods including breads, cereals, fried potato products, and coffee.2–4 In fact, acrylamide is ubiquitous in the human diet, with more than thirty percent of calories consumed coming from foods with detectable levels of acrylamide.5 Because acrylamide is classified as a probable human carcinogen,6 the discovery caused alarm that exposure to acrylamide in diet could be an important cancer risk factor.

The data establishing acrylamide as a probable carcinogen are based on animal and in vitro studies. Rats given acrylamide in water7, 8 show increased tumor rates at daily doses of about 1 mg/kg body weight. Furthermore, glycidamide, the primary metabolite of acrylamide, is a mutagenic compound that is reactive with DNA.9, 10 A recent prospective study among Dutch women found an increased risk of ovarian cancer overall and of endometrial cancer among never-smoking women with increased acrylamide intake,11 which suggests acrylamide may also act through a hormonal pathway.

Studies of human exposure to acrylamide have generally found no association between exposure and cancer risk. Nine studies have examined the association between dietary acrylamide intake and risk of cancers at various sites: colorectal, kidney, bladder, breast, oral, esophageal, larynx, ovarian, endometrial, and prostate.11–19 Of these, the only report of a significant association between higher acrylamide intake and cancer risk was for ovarian and endometrial cancer in a cohort of Dutch women.11

These studies all estimated acrylamide intake using food-frequency questionnaires (FFQs) to assess diet. It is not well-known, however, whether FFQs can accurately measure acrylamide intake in the diet. Acrylamide content in a particular food varies with specific cooking and processing methods for both homemade and commercially prepared foods, which makes measuring individual intake difficult. Several reports20–22 have attempted to assess the validity of FFQs by comparing them to acrylamide adducts to hemoglobin (Hb), a biomarker of acrylamide exposure that represents the internal dose of acrylamide over approximately 4 months.23 Results of these studies have varied.

In this study, we build on the epidemiological evidence using data from a population-based case-control study of prostate cancer in Sweden. First we assessed FFQ-assessed acrylamide intake by comparing intakes with acrylamide adducts to hemoglobin in a subset of the study population. Second, we studied the association between acrylamide exposure and prostate cancer risk using 2 measures of acrylamide exposure: FFQ intake and hemoglobin adducts. In addition to total prostate cancer, we also examine the risk of prostate cancer subtypes. In addition, this is the first report to use Hb adducts of acrylamide to study the association between acrylamide exposure and prostate cancer risk in humans.

Material and methods

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgements
  8. References

Participants

The Cancer of the Prostate in Sweden (CAPS) study is a population-based case-control study of prostate cancer previously described in detail.24 Cases were drawn from 4 of the 6 regional cancer registries in Sweden and diagnosed between 2001 and 2002. Participants from the northern and central regions were between 35 and 79 years old, and those from the southern regions were between 35 and 65 years. Cases were incident cases of pathologically or cytologically verified prostate cancer. Cases were informed about the study and asked to participate through their treatment physicians. The average time period between date of diagnosis and date when the questionnaire was sent out to the cases was 5 months. Clinical data on TNM (tumor, nodes, and metastasis) stage, Gleason score, and serum prostate-specific antigen (PSA) level at diagnosis was obtained from linkage to the National Prostate Cancer Registry. Clinical data was available for 95% of cases in the study.

Control subjects were randomly selected from the Swedish Population Registry, which maintains complete coverage of the population, and were frequency matched to cases by five-year age groups and region of residence. Controls were contacted by mail and received the same information about the study as cases.

Overall, 1895 prostate cancer cases were invited to participate. Of those, 1,499 (79%) completed a baseline lifestyle and health questionnaire and 1,400 (74%) donated a blood sample. Of 1,684 invited controls, 1,130 (67%) completed a questionnaire and 879 (52%) donated blood.

All study participants gave informed consent at the time of enrollment in the study. The study was approved by the ethics committees at Karolinska Institutet and Umeå University.

Dietary assessment

As part of the baseline questionnaire, all participants completed a self-administered 261-item FFQ that assessed usual intake of foods over the previous 12 months. Data from the Swedish National Food Administration was used to calculate total energy intake and intake of nutrients based on questionnaire responses.

Data on the median acrylamide content of foods from the Swedish National Food Administration25 was used to calculate usual intake of acrylamide through diet. Acrylamide intake was calculated based on the consumption of 18 foods and the assigned content of acrylamide per kilo food were as follows: fried potatoes (292 μg/kg), chips and popcorn (744 μg/kg), ground meat dishes (64 μg/kg), cereal (184 μg/kg), crispbread (rye) (527 μg/kg), crispbread (wheat) (93 μg/kg), black pudding (40 μg/kg), buns and cookies (115 μg/kg), pancakes (21 μg/kg), sausage (40 μg/kg), fish sticks (30 μg/kg), coffee 25 μg/kg), bread made of rye (122 μg/kg), crackers (205 μg/kg) and muesli (31 μg/kg). Some values are the average value based on several analyses, for example, the content for crisp bread (rye) is the average value from 3 different types of crisp breads. The acrylamide content in cakes and bread made from wheat was less then 30 μg/kg, and the content was therefore set to zero in our calculations. Acrylamide intake was calculated by multiplying the acrylamide content of a serving of the food by the frequency of consumption of that food and summing across all acrylamide-containing items on the FFQ. The resulting acrylamide intake is in micrograms per day. Acrylamide and other intakes were calorie-adjusted using the residual method.26

We restricted our study to the 2,617 men who completed the FFQ component of the questionnaire. Sixteen were excluded because of unreasonably high or low energy intakes, and 1 was excluded because of missing acrylamide data. After these exclusions, we had 1,499 cases and 1,118 controls with dietary acrylamide information.

Measurement of acrylamide adducts in blood

As a biomarker of acrylamide exposure, acrylamide adducts to Hb were measured in blood samples from a random sample of 377 men in the CAPS study. Participants received a blood sampling kit along with the questionnaire and informed consent form. They were given 4 tubes (heparin, plasma and EDTA-treated) and were instructed to donate blood at the nearest clinic. Unprocessed blood samples were sent by overnight mail to the Umeå Biobank, where samples were divided into plasma, serum, white and red blood cell components and stored in a freezer at −80°C until the time of analysis.

Red blood cell samples were analysed for acrylamide adducts to hemoglobin. Briefly, the globin was precipitated from the red blood cell samples, and the globin samples were then prepared for analysis of Hb adducts to N-terminal valine according to the N-alkyl Edman method.27, 28 Calibration curves were established from calibration samples prepared according to the procedure, using reference globin with known acrylamide-adduct level.28

The samples were processed in 4 batches of approximately 100 samples each. Throughout the analysis control samples of 2 different concentrations were run about 15 times in each batch (in addition to the calibration curve). The mean coefficient of variation (CV) was 3.9% within batches. The case/control status of the samples was unknown during the analysis. By chance, more case samples were included in batches 1 and 2, and more controls were in batches 3 and 4. Mean acrylamide adduct levels decreased over the 4 batches. Within each batch, mean adduct levels were similar for cases and controls, suggesting that the difference between batches was due to laboratory drift. Laboratory batch was adjusted for in all analyses.

Eleven samples were not processed because the cells were clotted. Men who reported using tobacco products (cigarettes, pipes, or snuff) at the time of the questionnaire (N = 40) were excluded from the statistical analysis, as tobacco users are exposed to much higher levels of acrylamide through tobacco than through the diet. (Mean levels of acrylamide adducts in these men was 152 pmol/g globin compared with 52 pmol/g in non-smoking men.) After exclusions, 170 cases and 161 controls were used in the analysis of Hb adducts of acrylamide.

Statistical analysis

For the comparison of FFQ acrylamide intake and acrylamide adduct levels, partial Pearson correlation coefficients between FFQ acrylamide intake and acrylamide adducts to Hb were calculated in the subset of men with blood measurements. We evaluated correlations separately among cases and controls. Correlations were adjusted for age, region, and laboratory batch. Acrylamide intake and Hb adducts were log-transformed to improve normality.

To measure the association between acrylamide adduct levels and risk of prostate cancer we used unconditional logistic regression models with indicator variables for quartile of acrylamide adducts. Quartiles were created based on the distribution among the controls. Age group and region, which were matching factors in this study, were included in all models, as was laboratory batch. Fully adjusted models also included variables for BMI (continuous) and former smoking (yes/no). To test for a dose-response trend across quartiles of acrylamide adducts, we also modelled adducts as a continuous variable to see the effect of a 10 pmol/g increase on cancer risk.

To measure the association between dietary acrylamide intake and risk of prostate cancer, we used unconditional logistic regression models with indicator variables for quintile of calorie-adjusted acrylamide intake. Quintiles were created based on the distribution of intake among the controls. Age group and region were included in all models. Fully adjusted models also included variables for BMI (continuous), former and current smoking (yes/no indicators), education (4 categories), zinc intake (ordinal quartiles), and total energy intake (continuous). Employment status and civil (marital) status were also considered as potential confounders. Several other nutrients and foods were considered as potential confounders including: alcohol, alpha-linolenic acid, calcium, vitamin D, folate, phytoestrogens, red meat, fish, and tomato intake. None of these was included in the final models, as they had little effect on the acrylamide-prostate cancer effect estimates or precision. Data on all confounders was collected in the self-administered mailed questionnaire. To test for a dose-response trend across quintiles of acrylamide intake, we modelled acrylamide as a continuous variable using the median intake in each quintile. Using the quintile medians reduces the impact of outliers and measurement error in the FFQ. We used this continuous variable to estimate the effect of a 10 μg/day increase in acrylamide intake.

All statistical analysis was done using SAS 9.1.

Results

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgements
  8. References

Characteristics of the study population

Cases and controls were similar in acrylamide intake estimated from the FFQ and in acrylamide adducts to hemoglobin (Table I). Mean intake was 44.5 μg/day among controls and 43.8 μg/day among cases. In the subset of men with acrylamide adduct measurements, the mean adduct level adjusted for laboratory batch was 53.7 pmol/g globin among controls and 54.7 pmol/g globin among cases. Cases were more likely to come from the Northern regions of Sweden. Cases and controls were similar in age, education, BMI, height, smoking status, and diet, including daily intakes of dairy, red meat, fish, fruits, vegetables, and total energy intake.

Table I. Characteristics of the Study Population by Disease Status
 Full study populationBlood cohort
CasesControlsCasesControls
  • 1

    Adduct levels are adjusted for laboratory batch.

  • 2

    Advanced case: Death from CaP, or N1, or M1, or T4 or T3. Localized cancer: T1–T2, N0/M0.

N1,4891,111170161
Age (mean ± SD)66.8 ± 7.367.7 ± 7.567.5 ± 7.467.3 ± 7.3
Region (%)
 North31182816
 South69827284
Education level (%)
 9 years of less46464248
 10–12 years40424642
 13+ years14111210
BMI (mean ± SD)26.2 ± 3.426.2 ± 3.426.0 ± 3.126.5 ± 3.3
Height (cm) (mean ± SD)177 ± 8176 ± 8176 ± 14176 ± 7
Smoking
 Never smokers (%)39384348
 Former smokers (%)49485652
 Current smokers (%)1112
Dietary characteristics (mean ± SD)
 Acrylamide intake (μg/d)43.8 ± 13.744.5 ± 14.545.1 ± 12.843.9 ± 13.4
 Acrylamide in μg/kg/day0.54 ± 0.180.56 ± 0.200.56 ± 0.170.54 ± 0.18
 Acryl. adduct (pmol/g globin)1nana54.7 ± 16.553.7 ± 14.9
 Total energy intake (kcal/d)2283 ± 6462219 ± 6562287 ± 6432293 ± 600
 Zinc (mg/d)11.6 ± 1.911.7 ± 1.811.3 ± 1.711.8 ± 1.8
Food intakes (serv/d) (mean ± SD)
 Dairy5.7 ± 3.15.6 ± 3.15.7 ± 2.95.6 ± 2.8
 Red meat1.4 ± 0.71.3 ± 0.71.3 ± 0.71.4 ± 0.7
 Fish0.5 ± 0.40.5 ± 0.40.5 ± 0.30.5 ± 0.3
 Crispbread2.4 ± 2.02.6 ± 2.12.6 ± 1.82.5 ± 1.7
 Other bread2.9 ± 1.82.8 ± 1.83.0 ± 1.82.8 ± 1.7
 Fruit1.9 ± 1.41.7 ± 1.22.0 ± 1.32.0 ± 1.4
 Vegetables2.7 ± 1.82.5 ± 1.72.5 ± 1.52.8 ± 1.6
 Baked goods1.0 ± 0.90.9 ± 0.81.0 ± 0.90.9 ± 0.8
 Coffee3.1 ± 1.93.1 ± 2.03.1 ± 1.93.0 ± 1.8
Disease characteristics among cases [N (% of cases)]
 CaP mortality205 (14)23 (14)
 Advanced cases2535 (36)62 (36)
 Localized cancer21020 (69)111 (65)
Gleason grade sum (mean ± SD)6.5 ± 1.26.4 ± 1.1
PSA Level (mean ± SD)88.5 ± 359.764.6 ± 168.5

Acrylamide intake ranged from 8 to 125 μg/day, or 0.08 to 1.59 μg/kilogram body weight per day. The top food contributors to acrylamide intake were crispbread, coffee, other bread, fried potatoes, and buns and cakes (Fig. 1). Crispbread, coffee, other bread, and fried potatoes were also the major contributors to variation in acrylamide intake between quintiles. There was a significant (p < 0.0001) correlation between acrylamide intake and intake of carbohydrates, fiber, and zinc (all positive) and alcohol (negative). Acrylamide intake was not correlated with age or height and was weakly correlated with BMI (r = 0.05, p = 0.01).

thumbnail image

Figure 1. Contribution of foods to acrylamide intake. [Color figure can be viewed in the online issue, which is available at www.interscience.wiley.com.]

Download figure to PowerPoint

Correlation between Hb Adducts and FFQ acrylamide

The partial correlation between FFQ acrylamide intake and acrylamide adducts to Hb was 0.25 (95% CI: 0.14–0.35), adjusted for age, region, energy intake, and laboratory batch. Among controls only, the correlation was 0.35 (95% CI: 0.21–0.48). Among cases, it was 0.15 (95% CI: 0.00–0.30). Correlations between acrylamide adducts and acrylamide intake measured in μg/kg body weight per day were almost identical to the correlations with acrylamide intake in μg/day. Adjustment for energy intake improved the correlations by reducing within-person measurement error. Without adjustment for calories, the correlation between FFQ acrylamide and adduct levels was 0.19 (95% CI: 0.08–0.29), adjusted for age, region, and batch.

Association between acrylamide adducts to hemoglobin and prostate cancer risk

As shown in Table II, no significant association was seen between quartile of acrylamide Hb adducts and prostate cancer risk. Adjusting for age, region, BMI, former smoking, and laboratory batch, the relative risk for the highest versus lowest quartile of acrylamide adducts was 0.93 (95% CI: 0.47–1.85). For a 10 pmol/g globin increase in acrylamide adducts, the relative risk of prostate cancer was 1.00 (CI: 0.86-1.16, p = 0.98). Among never smokers only, the association was similar; for a 10 pmol/g globin increase in acrylamide adducts, the relative risk of prostate cancer was 0.91 (CI: 0.75–1.11, p = 0.36).

Table II. Relative Risk of Prostate Cancer by Quartile of Acrylamide Hemoglobin Adducts
 Quartile of acrylamide hemoglobin adductsPer 10 unit increase in level1
Q1 (Lowest)Q2Q3Q4 (Highest)
  • Basic model: adjusted for age in 5-year intervals, region (north/south), and laboratory batch. Multivariable model: adjusted for age in 5-year intervals, region (north/south), laboratory batch, BMI (continuous) and former smoking (yes/no).

  • 1

    Acrylamide adducts modeled as continuous variable.

Median Hb adducts (pmol/g)32395056 
Number of case/contol40/4031/4147/4152/39 
Basic model
 OR1.000.781.120.981.02
 95% CI0.39–1.560.57–2.170.50–1.920.87–1.18
 p0.480.750.960.85
Multivariable model
 OR1.000.740.980.931.00
 95% CI0.37–1.490.50–1.930.47–1.850.86–1.16
 p0.400.950.840.98

No association was found between acrylamide adducts and specific prostate cancer subtypes including advanced disease, localised disease, high- or low-grade disease, or high- or low-PSA disease (Table III).

Table III. Relative Risk of Prostate Cancer Subtypes for given Increase in Hb Adducts or FFQ Acrylamide Intake
 Prostate cancer subtypesPSA level
AdvancedLocalizedHigh-GradeLow-GradeHighLow
  • Advanced disease: death, N1, M1 or T4 or T3. Localized disease: T1–T2 with N0 and M0. High-grade disease: Gleason sum 7–10. Low-grade disease: Gleason sum 2–6. High PSA disease: >10. Low PSA disease: ≤10.

  • 1

    Hb adducts modeled as continuous variable. Results adjusted for age in 5-year intervals, region (north/south), laboratory batch, BMI (continuous) and former smoking.

  • 2

    Acrylamide intake modeled as continuous variable using median intake in each quintile. Results adjusted for age in 5-year intervals, region (north/south), education (4 categories), former and current smoking, BMI (continuous), zinc intake (ordinal quartiles) and energy intake (continuous).

Per 10 pmol/g globin increase in Hb adducts1
 Number of case/contol62/162111/16168/16184/161100/16163/161
 OR1.090.941.190.861.040.95
 95% CI0.89–1.330.78–1.130.98–1.450.70–1.060.87–1.250.77–1.18
 p0.400.500.080.150.650.64
Per 10 μg/d increase in FFQ acrylamide intake2
 Number of case/contol516/1066988/1066612/1066675/1066223/1066589/1066
 OR0.980.980.990.990.961.05
 95% CI0.90–1.070.91–1.060.91–1.080.91–1.070.88–1.030.96–1.14
 p0.640.660.840.730.250.33

Association between dietary acrylamide from FFQ and prostate cancer risk

For dietary acrylamide calculated from the FFQ, there was no association between higher intakes and prostate cancer risk (Table IV). The relative risk for the highest versus lowest quintile of acrylamide intakes was 0.97 (CI: 0.75–1.27), adjusting for age, region, BMI, education, smoking, and zinc and energy intake. For a 10 μg/day increase in acrylamide intake, the RR of prostate cancer was 0.99 (CI: 0.92–1.06, p = 0.67). Similar results were seen when the analysis was limited to non-smokers only and never-smokers only. Among non-smokers, a 10 μg/day increase in acrylamide intake was associated with a relative risk of 1.02 (CI: 0.94–1.09, p = 0.68). Among never-smokers, a 10 μg/day increase in acrylamide intake was associated with a relative risk of 0.97 (CI: 0.86–1.08, p = 0.94).

Table IV. Relative Risk of Prostate Cancer by Quintile of Dietary Acrylamide Intake
 Quintile of dietary acrylamide intakePer 10 μg/day increase1
Q1 (Low)Q2Q3Q4Q5 (High)
  • Basic model: adjusted for age in 5-year intervals and region (north/south). Multivariable model: adjusted for age in 5-year intervals, region (north/south), education (4 categories), former and current smoking, BMI (continuous), zinc intake (ordinal quartiles) and energy intake (continuous).

  • 1

    Acrylamide intake modeled as continuous variable using median intake in each quintile.

Intake range (μg/d)8–3333–4040–4747–5656–125 
Number of case/contol294/210326/216269/214294/216255/210 
Basic model
 OR1.001.170.991.030.940.97
 95% CI0.91–1.490.77–1.270.81–1.320.73–1.210.91–1.04
 p0.210.930.810.650.39
Multivariable model
 OR1.001.140.991.060.970.99
 95% CI0.89–1.470.76–1.280.82–1.370.75–1.270.92–1.06
 p0.310.920.650.840.67

No association was found between acrylamide intake and specific prostate cancer subtypes including advanced disease, localised disease, high- or low-grade disease, or high- or low-PSA disease (Table III). Again, these results were similar when restricted to non-smokers only.

Intake of the 5 foods that contribute most to acrylamide intake in this study was examined (Table V). We found no association between crispbread, other bread, or coffee intake and prostate cancer risk, but a suggestion of increased risk for men in the highest tertile of intake of fried potatoes and buns/cakes. However, no p-value for trend was significant. Furthermore, this increased risk was unchanged when dietary acrylamide was also included in the models, suggesting that any association between these foods and prostate cancer risk is due to chance or to components other than acrylamide.

Table V. Relative Risk of Prostate Cancer by Category of Intake of High-Acrylamide Foods
 Category of food intakep-trend1
1 (Low)23 (High) 
  • Models adjusted for age in 5-year intervals, region (north/south), education (4 categories), former and current smoking, BMI (continuous), zinc intake (ordinal quartiles) and energy intake (continuous).

  • 1

    p-value for food intake as a continuous variable using median intake for each category.

Crispbread
 Number of case/control511/349554/421373/296 
 OR1.000.980.94 
 95% CI0.81–1.190.75–1.180.58
 Also adjusted for dietary acrylamide
  OR1.000.980.96 
  95% CI0.80–1.210.74–1.250.76
Other bread
 Number of case/control589/486353/248496/332 
 OR1.001.111.01 
 95% CI0.90–1.360.83–1.240.97
 Also adjusted for dietary acrylamide
  OR1.001.111.01 
  95% CI0.90–1.36083–1.240.96
Coffee
 Number of case/control579/428320/250539/388 
 OR1.000.941.01 
 95% CI0.76–1.160.83–1.220.86
 Also adjusted for dietary acrylamide
  OR1.000.941.04 
  95% CI0.75–1.170.83–1.300.66
Fried potatoes
 Number of case/control240/246749/511449/309 
 OR1.001.361.31 
 95% CI1.10–1.691.03–1.660.16
 Also adjusted for dietary acrylamide
  OR1.001.361.32 
  95% CI1.09–1.691.03–1.690.14
Buns and cakes
 Number of case/control356/318626/434456/314 
 OR1.001.251.26 
 95% CI1.02–1.530.99–1.590.26
 Also adjusted for dietary acrylamide
  OR1.001.241.25 
  95% CI1.01–1.520.98–1.580.28

Discussion

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgements
  8. References

We found a moderate correlation between acrylamide adducts to Hb and FFQ acrylamide intake, but no association between either Hb adducts or FFQ acrylamide intake and prostate cancer risk. In addition, there was no association between either exposure measure and risk of specific subtypes of prostate cancer, either by grade or stage. This is the first study to examine the association between Hb adducts of acrylamide and prostate cancer risk in humans, and it is the first study to examine acrylamide exposure and risk of various subtypes of prostate cancer.

Correlation between Hb Adducts and FFQ acrylamide

We found a moderate correlation between acrylamide adducts to Hb and FFQ acrylamide intake. This correlation was higher among controls than among cases. This difference may be due to recall or selection bias or to recent changes in the diets of cases after diagnosis.

Three other articles have examined the correlation between acrylamide adducts and FFQ-calculated acrylamide intake with varying results. In the Malmö Diet and Cancer Cohort, Wirfalt, et al.21 found a correlation of 0.60 in non-smoking men. This correlation is probably overstated as some participants were selected based on high or low consumption of acrylamide-rich foods, introducing greater between-person variation in intake. Kütting et al.29 found low (0.17–0.18) but statistically significant correlations between adduct levels and FFQ acrylamide in a study of 828 non-smokers. These results were not adjusted for age or energy intake, and in fact, these correlations are quite similar to our unadjusted results. This underscores the importance of adjustment for energy intake in studies using FFQs, as energy adjustment has been shown to reduce within-person measurement error and improve FFQ performance.30 Bjellaas et al.22 found no correlation between adduct levels and FFQ acrylamide among 50 people. The length of the FFQ and the number of items used to calculate acrylamide intake were not stated. In addition, age, sex, and energy intake were not adjusted for in the correlation analysis.

It is important to note that FFQ acrylamide and acrylamide adducts to hemoglobin do not measure the same thing. FFQ measurements estimate dietary intake of acrylamide, whereas hemoglobin adducts also take into account the absorption and metabolism of dietary acrylamide. In addition, the FFQ assesses the past year of intake, whereas Hb adducts reflect the past several months of exposure. Therefore, the 2 measures would not likely be perfectly correlated even in the absence of measurement error. In addition, it is possible that measuring adducts of both acrylamide and glycidamide might result in higher correlations between adducts and the FFQ, as this would help account for some differences in metabolism between individuals. Our observed correlations can be seen as an estimate of the lower bound of FFQ validity.

Acrylamide adduct levels and prostate cancer risk

This is the first study to examine the association between acrylamide adduct levels and prostate cancer risk in humans. We found no association between adduct levels of acrylamide and risk of overall prostate cancer or prostate cancer subtypes. Our ability to study specific subtypes such as advanced or localised disease was limited by low sample sizes in the subgroup with blood data.

The major limitation of our biomarker analysis is related to the biologically relevant time period of acrylamide exposure. Because of the case-control design, blood samples for cases were collected after cancer diagnosis and do not necessarily reflect acrylamide exposure during development of the cancer. In addition, hemoglobin adducts of acrylamide reflect only the previous several months of exposure rather than the long-term dietary exposure that is likely most relevant to the development of prostate cancer.

Dietary acrylamide and prostate cancer risk

We found no association between FFQ-measured acrylamide intake and risk of overall prostate cancer. We also found no association between intake of the major acrylamide-contributing foods and prostate cancer risk. Our findings are similar to those of Pelucchi et al.17 in a network of hospital-based case-control studies in Italy and Switzerland and to those of Hogervorst et al.19 in a prospective cohort study in the Netherlands, who found no association between acrylamide intake and prostate cancer risk overall or in non-smoking men only.

Prostate cancer is a clinically heterogeneous disease, and risk factors may play different roles in disease incidence and progression.31 Therefore it is important to examine prostate cancer risk according to stage and grade of prostate cancer when possible. Ours is the first study to examine acrylamide intake and risk of prostate cancer subtypes. The risk estimates remained null across different stages, grades, and PSA-levels of cases.

Random measurement error in acrylamide intake is likely. Because the FFQ was not originally designed to assess preparation of foods, food items with different cooking methods and substantially different acrylamide contents are sometimes grouped into a single FFQ question. In addition, the acrylamide content of foods varies widely depending on specific processing and cooking parameters, resulting in large differences in acrylamide content between different brands of the same foods and even between different production batches within the same brand. However, our analysis suggests that FFQ acrylamide intake does provide useful information on acrylamide exposure. Even with error in our measurement of absolute acrylamide intakes, we are likely to correctly rank individuals in the highest and lowest quintiles of intake.

Differential measurement error is also likely. The difference between cases and controls in the correlation between FFQ acrylamide and adduct levels suggest that FFQ accuracy varies by disease status. It is not clear how such recall bias might affect our results. Further study of acrylamide intake in prospective cohorts will be necessary to eliminate the possibility of recall bias.

Strengths of our study include its large size, an FFQ with information on most commonly consumed high-acrylamide food items, and information on subtypes of prostate cancer. In addition, we used both FFQ and biomarker measurements of acrylamide exposure, which strengthens our findings of no significant association between dietary acrylamide and prostate cancer risk.

Conclusions

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgements
  8. References

It is important to note that these results are not at odds with risk estimates based on toxicological studies. Such studies suggest very small increases in cancer risks for the levels of acrylamide taken in the diet. Using animal study data, Hagmar and Törnqvist estimate a relative risk of 1.015–1.05 for humans consuming acrylamide at levels greater than 1 μg/kg/day, which represents less than 2% of the population.32 Even very large case-control or cohort studies do not have the power to rule out such small relative risks.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgements
  8. References

K.M.W. is partially supported by NCI/NIH Training Grant T32 CA09001.

References

  1. Top of page
  2. Abstract
  3. Material and methods
  4. Results
  5. Discussion
  6. Conclusions
  7. Acknowledgements
  8. References