Acrylamide and Health Outcomes

Background

Acrylamide forms in foods rich in starch when cooked/processed at temperature >120°C (e.g., roasting, frying, grilling) in absence of moisture. Acrylamide may also form in foods from acrolein during the degradation of amino acids, carbohydrates, lipids, and organic acids. Of note, the content of acrylamide in diet may vary strongly depending on the food matrix and processing methods (home and restaurants cooking and/or industrial processing). Acrylamide is prevalent in a wide-range of commonly consumed foods (e.g. coffee, biscuits, breakfast cereals, bread especially crispy bread, potato products and vegetables crisps) - some of these are also part of a healthy diet. Acrylamide is also produced in heated tobacco.

Acrylamide-induced neurotoxic adverse effects have been well-documented in animal and occupational studies. From its inception, acrylamide is a chemical compound used in the industry to produce plastics (polyacrylamides). However, its occupational exposure is considered of low concern due to successful intervention and preventive measures. Cumulative exposure to acrylamide in workers via inhalation and/or dermal absorption have been observed to give rise central and peripheral neurotoxicity, supporting the concern related to cumulative dietary acrylamide exposure in the development of neurodegenerative disorders in the general population. Yet, to date, no epidemiological studies have investigated whether dietary acrylamide exposure associates to neurodegenerative diseases including dementia, Alzheimer´s, and Parkinson´s diseases.

Growing body of evidence also indicate that acrylamide´s mode of action might lead to cardiometabolic alteration which in turn, play a critical role in atherosclerosis and CVD development. Still, epidemiological evidence investigating dietary acrylamide and incidence of CVD including myocardial infarction are lacking.

In theory, acrylamide can exert toxicity in all the human organs. In vivo and vitro studies show that acrylamide-induced toxicity may be evident at low doses. However, it is proposed that the detrimental effects will exacerbate at high doses or most likely at prolonged/cumulative exposure such the dietary exposure resulting in the clinical manifestation of the diseases here investigated. There is substantial evidence showing that toxicity of acrylamide is mediated by glycidamide, a highly reactive compound and the main metabolite of acrylamide, via the formation of DNA adducts. Also, production of oxidative stress and formation of adducts with protein (especially alkylation) are considered plausible molecular initiating events to trigger cumulative neurotoxicity. For the acrylamide-induced neurotoxicity, acrylamide is suspected to promote neuroinflammation, induce neurite degeneration, inhibit axonal transport, alter the level and turnover of neurotransmitters (e.g., enhanced activity of acetylcholinesterase and dopamine depletion) and directly inhibit neurotransmission which ultimately may result in learning and memory impairment, anxiogenic responses, and disturbed motor coordination. Consequently, this can give rise to neurodegenerative disease.

The importance of using objective measures of dietary intake in epidemiological studies to improve accuracy and limit biases and acknowledged difficulties related to the estimation of dietary exposures from self-reported questionnaire and food databases has been stressed. Biomarkers of acrylamide exposure have been identified and validated in blood and in urines. The blood biomarkers i.e., blood hemoglobin adducts are considered more suitable biomarkers of exposure, as they may reflect long term of exposure (previous four months) and are not influenced by daily fluctuations.

Finally, screening for potential acrylamide-related metabolomics and proteomics may contribute, in epidemiological studies, to strengthen the possible exposure-health outcome association and understand its mode of action. To our knowledge, little effort has been made in this direction in this field. In unpublished data based on the EUROMIX (n=120 healthy subject), an association was found between hemoglobin adducts and 4 proteins (JAM-B, CNTN5, CLEC10A and EPHB6) related to neurological alteration.

Specific aims:

To assess the relationship between dietary acrylamide exposure, measured through validated biomarkers (i.e., acrylamide and glycidamide hemoglobin adducts) and the risk of dementia including Alzheimer´s disease (n=700 cases) - Aim 1; Parkinson´s disease (n =190 cases) - Aim 2 and myocardial infarction (n= 390 cases) - Aim 3 in two population-based Swedish cohorts, The Cohort of 60-year-olds (60YO) and The Swedish Mammography Cohort (SMC) employing a case-cohort design.

In addition, exposure-affected OMICS' signatures (mainly proteins and metabolomics) will be used to explore molecular pathways potentially underlying diseases development and to support biological plausibility. Acrylamide-related metabolic and proteins biomarkers will be assessed in relation to risk of the diseases under investigation (only SMC) - Aim 4

Study design: The project will be based on two relatively large population-based cohorts, The Cohort of 60-year-olds (60YO) and the Swedish Mammography Cohort (SMC). Both cohorts are well-established in Sweden. Due to the high cost of acrylamide biomarkers measurements, a case-cohort design will be employed. The total estimated case-cohort sample will consist of approximately 1,740 participants. Two sub-cohorts will be randomly selected from the full cohorts (n = 442 around 5 % of the baseline population in 1997-1999, for 60YO, and 2003-2009, for SMC). All the cases of each of the diseases under investigation (i.e., dementia including Alzheimer´s, Parkinson´s disease and myocardial infarction) occurring outside the sub-cohorts will be included. The cases will include participants free of any diagnosis of each of the diseases investigated at the time of sampling. Incident cases of the diseases under investigation will be identified through linkage to the National Patient Register and Cause of Death Register till the end of follow-up (2022 for both cohorts). The incident cases of dementia and Alzheimer´s disease will also be retrieved through the linkage to the Swedish registry for cognitive/dementia disorders (SveDem). The integration of this quality register will improve the sensitivity of dementia diagnosis which has been shown to be low when detected only from the National Patient Register and the Cause of Death Register. The statistical power calculation (%) based on the expected incident cases (n) for each of the diseases under investigation and assumed Hazard Ratios (HR) as follows:

cases expected: Myocardial infarction: n=390; Dementia (including Alzheimer´s disease) n=710 and Parkinson´s disease n= 192

Power calculation:

HR 1.25: MI: 50%,Dementia: 60% and Parkinson: 34% HR 1.50: MI 90%, Dementia:95%, Parkinson: 70% HR: 1.75: MI 99%, Dementia: 100% and Parkinson 91%

Detailed information on the 60YO and SMC are provided here:

https://ki-se.proxy.kib.ki.se/en/imm/the-cohort-of-60-year-olds; https://www.simpler4health.se/.

Exposures: Acrylamide biomarkers measured in blood, i.e., acrylamide and glycidamide hemoglobin adducts. Whole blood samples (approximately n = 1,740) will be collected (0.5 ml) from samples stored in freezer at -80°C at Uppsala University biobank, for the SMC, and at Karolinska Institutet (KI) biobank, for the 60YO, and sent to a certified laboratory in Stockholm. The assessment will be carried out with a validated method in Törnqvist's laboratory. Briefly, the acrylamide and glycidamide hemoglobin adducts to N-terminal valine (AA-val and Gly-val, respectively) will be measured in blood by LC-MS based methods, representing the exposure during the lifespan of the erythrocyte. The case/control status of the samples will be unknown during the analysis. Laboratory batch will be adjusted for in all analyses. Each of two acrylamide biomarkers will be modeled as continuous (per 10 pmol/g increment) and in categories (tertiles or quartiles) to assess potential nonlinear dose-response.

Outcome definition: Incident cases of dementia (including Alzheimer´s disease), Parkinson´s disease and myocardial infarction. Based on the International Classification of disease version 10 (ICD10), the diagnoses included are the following:

Neurodegenerative disorders:

• Dementia (in Alzheimer´s disease, vascular, related to other disease, unspecified) including Alzheimer´s disease, other specified (including Lewis bodies) and unspecified neurodegenerative diseases: F00-F03, G30, G31.8, G31.9
• Parkinson´s disease:G20

Cardiovascular disease:

• Myocardial infarction:I21

Metabolomic and Proteomic data: Metabolites and proteins markers are already available in the SMC sample. Large scale untargeted liquid chromatography-mass spectrometry and OLINK panels (Proseek Multiplex CVD II, CVD III and Metabolism) were employed to obtain these data from blood sample collected at baseline. For detailed information on the methodology and analytical process, please refer to previously published research26,27. A total of approximately 10,000 metabolites and 250 proteins will be available for all the individuals selected for the case-cohort sample (approximately n=900).

Time plan and implementation:

The project will last four years (Jan 2024-Dec 2026).

Data analysis and statistics:

Project I: acrylamide and glycidamide hemoglobin adducts in relation to risk of neurodegenerative and cardiovascular diseases.

Cox proportional hazards weighted regression models will be employed to estimate hazard ratios and 95% confidence intervals (CI) for each of the health outcomes under investigation in relation to acrylamide and glycidamide biomarkers, respectively, with follow-up from the date of acrylamide biomarker measurement till December, 31rst, 2022. Weighted likelihood approach using Borgan II weights will be applied28. The weighted method will allow to make inference on the full cohort28. Models will be adjusted considering several covariates such as sex (only for 60YO), age (only for SMC), education, physical activity, smoking, alcohol consumption, body mass index, dietary factors and other important confounders retrieved a priori from the literature. Analysis will be performed separately in the two cohorts. Then, results will be combined considering the possible heterogeneity between cohorts using random effect models. The main analysis will be repeated separately in smokers and non-smokers. Smokers have three to four times higher levels of acrylamide compared to the non-smokers. Conducting this analysis separately is important to better distinguish the risks associated with different sources of acrylamide exposure, specifically from the diet versus smoking. Since acrylamide may disrupt hormonal levels, menopausal status will be also considered in the analysis. To investigate the shape of the possible relationship between acrylamide biomarkers and the diseases considered, each of the associations with multivariate restricted cubic splines will be also modelled. Assuming that the true hazard ratio (HR) is 1.5 and with a significance level α =0.05, by using 5% of the total cohort, ≥90% of statistical power for analysis involving dementia and myocardial infarction and 70% for Parkinson disease are expected. If the true HR will be ≤1.25, the statistical power will decrease and analysis related to Parkinson´s disease may be underpowered if not combined with dementia including Alzheimer's disease. Multivariable adjustments as well as the employment of exposures in categories may somewhat reduce the power.

Additionally, the association between dietary factors and acrylamide levels (Sub-project), taking advantage of the longitudinal dietary information present in SMC, will be investigated. Multi-adjusted quantile regression models will be employed to model changes in the diet during 1989-1990, 1997, 2009 in relation to blood acrylamide levels. Foods that may have high content of acrylamide have been investigated, however it will be important to clarify which dietary determinants/dietary pattern may be relevant in the association with acrylamide levels.

Project II: associations between acrylamide hemoglobin adducts and OMICS data (Phase I) and related metabolites/proteins and incidence of neurodegenerative and myocardial infarction (Phase II).

The proposed analysis will be performed in SMC only. The expected sample size for this analysis will be around n=900 of which n=236 for the subsample and n=420 cases for neurodegenerative disease and n=160 cases for myocardial infarction.

Phase I: Partial Least Squares analysis adapted for large-scale OMICS data to minimize false positive associations, followed by partial Spearman rank correlation analysis adjusted for covariates will be used to investigate the association between OMICS' data and acrylamide and glycidamide biomarkers29. Phase II: Each of those proteins and metabolites identified in Phase I will be then modeled as independent variables in relation to the risk of each of the diseases under investigation, using multi-adjusted Cox proportional hazards weighted regression models accounting for multiple testing corrections. To facilitate interpretation, significant proteins and metabolites will then be plotted with their correlation with each of the blood biomarkers of acrylamide and the diseases under investigation.