Toxicity testing of poorly soluble particles, lung overload and lung cancer.

In 2013, an ECETOC Task Force evaluated scientific understanding of the 'lung overload' hypothesis. As there is no evidence that humans develop lung tumours following exposure to poorly soluble particles (PSPs), emphasis was given to the observed higher sensitivity and specificity of rat lung responses and potential impacts of this on human risk assessment. Key arguments and outcomes are summarised here, together with discussion of additional findings published since 2013. Inhalation exposure to PSPs in all species is associated with localised pulmonary toxicity initiated by a persistent pro-inflammatory response to particle deposition. Events in the rat indicate a plausible adverse outcome pathway for lung tumour development following exposure to PSPs under overload conditions. A different particle lung translocation pattern compared to rats make humans less sensitive to developing comparable lung overload conditions and appears to also preclude tumour formation, even under severe and prolonged exposure conditions. Evidence continues to suggest that the rat lung model is unreliable as a predictor for human lung cancer risk. However, it is a sensitive model for detecting various thresholded inflammatory markers, with utility for non-neoplastic risk assessment purposes. It is noteworthy that preventing inflammatory rat lung responses will also inhibit development of neoplastic outcomes.

Threshold and non-threshold chemical carcinogens: A survey of the present regulatory landscape.

For the proper regulation of a carcinogenic material it is necessary to fully understand its mode of action, and in particular whether it demonstrates a threshold of effect. This paper explores our present understanding of carcinogenicity and the mechanisms underlying the carcinogenic response. The concepts of genotoxic and non-genotoxic and threshold and non-threshold carcinogens are fully described. We provide summary tables of the types of cancer considered to be associated with exposure to a number of carcinogens and the available evidence relating to whether carcinogenicity occurs through a threshold or non-threshold mechanism. In light of these observations we consider how different regulatory bodies approach the question of chemical carcinogenesis, looking in particular at the definitions and methodologies used to derive Occupational Exposure Levels (OELs) for carcinogens. We conclude that unless proper differentiation is made between threshold and non-threshold carcinogens, inappropriate risk management measures may be put in place - and lead also to difficulties in translating carcinogenicity research findings into appropriate health policies. We recommend that clear differentiation between threshold and non-threshold carcinogens should be made by all expert groups and regulatory bodies dealing with carcinogen classification and risk assessment.

A comparison of the results from intra-pleural and intra-peritoneal studies with those from inhalation and intratracheal tests for the assessment of pulmonary responses to inhalable dusts and fibres.

The aim of this paper is to compare results from inhalation studies with those from intraperitoneal and intrapleural tests, where available, for a number of fibrous and particulate test materials. The objective is to determine how well intraperitoneal/intrapleural studies predict the pathological responses observed in more standard in vivo studies of pulmonary toxicity, with a particular focus on carcinogenicity. Published toxicity data was obtained for a number of materials including asbestos, wollastonite, MMVFs (including glass fibres, stone wools and RCF), silicon carbide whiskers, potassium octatitanate, quartz, kevlar, polypropylene and titanium dioxide. For some of the fibrous material reviewed, there is conformity between the results of intraperitoneal and inhalation tests such that they are either consistently positive or consistently negative. For the remaining fibrous materials reviewed, intraperitoneal and inhalation tests give different results, with positive results in the intraperitoneal test not being reflected by positive inhalation results. It is suggested that the intraperitoneal test can be used to exonerate a dust or fibre (because if negative in the intraperitoneal test it is extremely unlikely to be positive in either inhalation or intratracheal tests) but should not be used to positively determine that a dust or fibre is carcinogenic by inhalation. We would argue against the use of intraperitoneal tests for human health risk assessment except perhaps for the purpose of exoneration of a material from classification as a carcinogen.

Regulatory risk assessment approaches for synthetic mineral fibres.

Exposure to synthetic mineral fibres (SMF) may occur in a number of workplace scenarios. To protect worker health, a number of different organisations worldwide have assessed the health risk of these materials and established workplace exposure limits. This paper outlines the basic principles of risk assessment and the scientific methods used to derive valid (justifiable) occupational exposure limits (OELs) and goes on to show how, for SMF, and particularly for refractory ceramic fibre (otherwise known as aluminosilicate wool, RCF/ASW), the methods used and the associated outcomes differ widely. It is argued that the resulting differences in established OELs prevent consistent and appropriate risk management of SMF worldwide, and that development of a transparent and harmonised approach to fibre risk assessment and limit-setting is required.

Sources of variability in biomarker concentrations.

Human biomonitoring has become a primary tool for chemical exposure characterization in a wide variety of contexts: population monitoring and characterization at a national level, assessment and description of cohort exposures, and individual exposure assessments in the context of epidemiological research into potential adverse health effects of chemical exposures. The accurate use of biomonitoring as an exposure characterization tool requires understanding of factors, apart from external exposure level, that influence variation in biomarker concentrations. This review provides an overview of factors that might influence inter- and intraindividual variation in biomarker concentrations apart from external exposure magnitude. These factors include characteristics of the specific chemical of interest, characteristics of the likely route(s) and frequency of exposure, and physiological characteristics of the biomonitoring matrix (typically, blood or urine). Intraindividual variation in biomarker concentrations may be markedly affected by the relationship between the elimination half-life and the intervals between exposure events, as well as by variation in characteristics of the biomonitored media such as blood lipid content or urinary flow rate. Variation across individuals may occur due to differences in time of sampling relative to exposure events, physiological differences influencing urinary flow or creatinine excretion rates or blood characteristics, and interindividual differences in metabolic rate or other factors influencing the absorption or excretion rate of a compound. Awareness of these factors can assist researchers in improving the design and interpretation of biomonitoring studies.

Biomonitoring for workplace exposure to copper and its compounds is currently not interpretable.

This paper sets out to explore the requirements needed to recommend a useable and reliable biomonitoring system for occupational exposure to copper and its inorganic compounds. Whilst workplace environmental monitoring of copper is used to measure ambient air concentrations for comparison against occupational exposure limits, biological monitoring could provide complementary information about the internal dose of workers, taking into account intra-individual variability and exposure from all routes. For biomonitoring to be of reliable use for copper, a biomarker and the analytical ability to measure it with sufficient sensitivity must be identified and this is discussed in a range of matrices. In addition, there needs to be a clear understanding of the dose-response relationship of the biomarker with any health-effect (clinical or sub-clinical) or, between the level of external exposure (by any route) and the level of the copper biomarker in the biological matrix being sampled, together with a knowledge of the half-life in the body to determine accurate sampling times. For many biologically non-essential metals the requirements for reliable biomarkers can be met, however, for 'essential' metals such as copper that are under homeostatic control, the relationship between exposure (short- or long-term) and the level of any copper biomarker in the blood or urine is complex, which may limit the use and interpretation of measured levels. There are a number of types of biomarker guidance values currently in use which are discussed in this paper, but no values have yet been determined for copper (or its inorganic compounds) due to the complexity of its essential nature; the US The American Conference of Governmental Industrial Hygienists (ACGIH) has however indicated that it is considering the development of a biological exposure index for copper and its compounds. In light of this, we present a review of the reliability of current copper biomarkers and their potential use in the occupational context to evaluate whether there is value in carrying out human biomonitoring for copper exposure. Based on the available evidence we have concluded that the reliable use of biomonitoring of occupational exposure to copper and its application in risk assessment is not possible at the present time.

Occupational cancer burden in Great Britain.

A sound knowledge base is required to target resources to reduce workplace exposure to carcinogens. This project aimed to provide an objective estimate of the burden of cancer in Britain due to occupation. This volume presents extensive analyses for all carcinogens and occupational circumstances defined as definite or probable human occupational carcinogens by the International Agency for Research on Cancer. This article outlines the structure of the supplement - two methodological papers (statistical approach and exposure assessment), eight papers presenting the cancer-specific results grouped by broad anatomical site, a paper giving industry sector results and one discussing work-related cancer-prevention strategies. A brief summary of the methods and an overview of the updated overall results are given in this introductory paper. A general discussion of the overall strengths and limitations of the study is also presented. Overall, 8010 (5.3%) total cancer deaths in Britain and 13,598 cancer registrations were attributable to occupation in 2005 and 2004, respectively. The importance of cancer sites such as mesothelioma, sinonasal, lung, nasopharynx, breast, non-melanoma skin cancer, bladder, oesophagus, soft tissue sarcoma and stomach cancers are highlighted, as are carcinogens such as asbestos, mineral oils, solar radiation, silica, diesel engine exhaust, coal tars and pitches, dioxins, environmental tobacco smoke, radon, tetrachloroethylene, arsenic and strong inorganic mists, as well as occupational circumstances such as shift work and occupation as a painter or welder. The methods developed for this project are being adapted by other countries and extended to include social and economic impact evaluation.

Framework for the development and application of environmental biological monitoring guidance values.

Human biomonitoring (HBM) is widely recognised as a useful tool to aid assessment of exposure to chemical substances, but our ability to detect hazardous substances (or their metabolites and health effects) often exceeds our understanding of their biological relevance. There are only a few established frameworks for developing and using occupational and environmental biological guidance values (BGVs), mostly for data-rich substances that have been in use for some time. BGVs for new substances and those with unknown dose-response relationships are difficult to derive. An accepted framework based on current scientific knowledge and best practice is therefore urgently needed to help scientists, regulators, and stakeholders to design appropriate HBM studies, interpret HBM data (both for groups and individuals) understand the limitations and to take appropriate action when required. The development and application of such a tool is described here. We derived a conceptual framework that was refined by consultation with an advisory group and workshop. The resulting framework comprised four levels defined by increasing data, with increasing confidence for human health risk assessment. Available data were used for 12 chemicals with expert judgement to illustrate the utility of the framework.

Can vitamin C induce nucleotide excision repair? Support from in vitro evidence.

Intracellular vitamin C acts to protect cells against oxidative stress by intercepting reactive oxygen species (ROS) and minimising DNA damage. However, rapid increases in intracellular vitamin C may induce ROS with subsequent DNA damage priming DNA repair processes. Herein, we examine the potential of vitamin C and the derivative ascorbate-2-phosphate (2-AP) to induce a nucleotide excision repair (NER) response to DNA damage in a model of peripheral blood mononuclear cells. Exposure of cells to elevated levels of vitamin C induced ROS activity, resulting in increased levels of deoxycytidine glyoxal (gdC) and 8-oxo-2'-deoxyguanosine (8-oxodG) adducts in DNA; a stress response was also induced by 2-AP, but was delayed in comparison to vitamin C. Evidence of gdC repair was also apparent. Measurement of cyclobutane thymine-thymine dimers (T < >T) in DNA and culture supernatant were included as a positive marker for NER activity; this was evidenced by a reduction in DNA and increases in culture supernatant levels of T < >T for vitamin C-treated cells. Genomics analysis fully supported these findings confirming that 2-AP, in particular, induced genes associated with stress response, cell cycle arrest, DNA repair and apoptosis, and additionally provided evidence for the involvement of vitamin C in the mobilisation of intracellular catalytic Fe.

In vivo vitamin C supplementation increases phosphoinositol transfer protein expression in peripheral blood mononuclear cells from healthy individuals.

Ascorbate can act as both a reducing and oxidising agent in vitro depending on its environment. It can modulate the intracellular redox environment of cells and therefore is predicted to modulate thiol-dependent cell signalling and gene expression pathways. Using proteomic analysis of vitamin C-treated T cells in vitro, we have previously reported changes in expression of five functional protein groups associated with signalling, carbohydrate metabolism, apoptosis, transcription and immune function. The increased expression of the signalling molecule phosphatidylinositol transfer protein (PITP) was also confirmed using Western blotting. Herein, we have compared protein changes elicited by ascorbate in vitro, with the effect of ascorbate on plasma potassium levels, on peripheral blood mononuclear cell (PBMC) apoptosis and PITP expression, in patients supplemented with vitamin C (0-2 g/d) for up to 10 weeks to investigate whether in vitro model systems are predictive of in vivo effects. PITP varied in expression widely between subjects at all time-points analysed but was increased by supplementation with 2 g ascorbate/d after 5 and 10 weeks. No effects on plasma potassium levels were observed in supplemented subjects despite a reduction of K+ channel proteins in ascorbate-treated T cells in vitro. Similarly, no effect of vitamin C supplementation on PBMC apoptosis was observed, whilst ascorbate decreased expression of caspase 3 recruitment domain protein in vitro. These data provide one of the first demonstrations that proteomics may be valuable in developing predictive markers of nutrient effects in vivo and may identify novel pathways for studying mechanisms of action in vivo.

An assessment of potential cancer risk following occupational exposure to ethanol.

Recognition of the carcinogenic properties of ethanol has resulted from comprehensive evidence regarding the effect of consumption of alcohol; indeed, ethanol in alcoholic beverages is now considered a Group 1 carcinogen by the International Agency for Research on Cancer. However, there is little information on the effects of ethanol following exposure via the occupationally relevant routes of inhalation and dermal exposure. This review therefore focuses on these exposure routes, to assess potential carcinogenic risk associated with occupational exposure to ethanol. Inhalatory exposure at the current occupational exposure limit (OEL) for the United Kingdom (1000 ppm ethanol over an 8-h shift) was estimated to be equivalent to ingestion of 10 g ethanol (approximately 1 glass of alcohol) per day. However, in the occupational setting the dose-rate delivery of this amount of ethanol is low, allowing for its rapid and effective elimination, for the majority of individuals. Similarly, while dermal absorption in an occupational setting could potentially add to overall body ethanol burden, additional carcinogenic risk of such exposure is considered negligible. Thus, on balance, there appears little cause to suppose occupational exposure at or below the current OEL associates with any appreciable increase in risk of cancer. However, available occupational exposure data to confirm this view are currently limited. It is also suggested that adoption of a more flexible classification regime, considering risk in the context of hazard and exposure (such as that adopted by the German MAK commission), would represent an improvement over traditional occupational risk assessment practices.

Evidence of oligonucleotides containing 8-hydroxy-2'-deoxyguanosine in human urine.

The product of oxidative damage to DNA, 8-hydroxy-2'-deoxyguanosine (8-OHdG), when detected in urine, is considered to be a global, noninvasive biomarker of in vivo oxidative DNA damage. In this paper we describe a novel approach to confirm the presence of oligonucleotides containing 8-OHdG in human urine. Fractions of urine were prepared by gel-filtration chromatography, and the presence of oligonucleotides was confirmed by ELISA using a monoclonal anti-(single-stranded DNA) antibody. Pools of urine fractions were subsequently prepared according to ELISA reactivity, each containing oligonucleotides with a known range of base numbers. The level of 8-OHdG in each pool was subsequently determined using a commercial ELISA kit. Results confirmed that oligonucleotides containing 8-OHdG are present in urine and, most significantly, oligomers of <30-55 bases were found to be associated with 8-OHdG. This finding strongly supports the involvement of nucleotide excision repair (NER) in the removal of 8-OHdG from the cell. The novel approach adopted in this study was validated using cell culture supernatant obtained from an in vitro model comprising CCRF cells exposed to vitamin C; this model has previously been shown to stimulate removal of 8-OHdG from the cell by an NER-dependent process.

Setting evidence-based occupational exposure limits for manganese.

In 2004, a review by the Institute of Environment and Health (IEH) made recommendations on occupational exposure limits (OELs) for manganese and its inorganic compounds for inhalable and respirable fractions respectively. These OELs were based on a detailed comprehensive evaluation of all the scientific data available at that time. Since then, more published studies have become available and a number of occupational standard-setting committees (EU SCOEL, US ACGIH-TLV, and German MAK) have proposed OEL's for manganese and its inorganic compounds that are somewhat lower that those proposed in the 2004 review. Based on current understanding, the key toxicological and human health issues that are likely to influence a health-based recommendation relate to: neurotoxicology; reproductive and developmental toxicology; and mutagenicity/carcinogenicity. Of these, it is generally considered that neurotoxicity presents the most sensitive endpoint. As such, many of the studies that have been reported since the IEH review have sought to use those neurofunctional tests that appear to be particularly sensitive at identifying the subtle neurological changes thought to associate with manganese toxicity. These recent studies have, however, continued to be limited to a significant extent by reliance on cross-sectional designs and also by use of unreliable exposure estimation methods. Consequently the strength of the potential association between manganese exposure and these subtle subclinical cognitive or neuromotor changes is still poorly characterised and the relevance of these minor differences in terms of either their clinical or quality of life consequences remains unknown. Based upon the overall evidence, it is concluded that the 8-h time weighted averages (TWA) for respirable (0.05mg/m3 as Mn) and inhalable (0.2mg/m3 as Mn) fractions as recommended by the SCOEL in 2011 are the most methodologically-sound, as they are based on the best available studies, most suited to the development of health-based OELs for both respirable and inhalable fractions. The dose-response characterisation informed by the examined studies used can be considered to establish a true human NOAEL for all the neurofunctional endpoints examined within the selected studies.

Validation of a novel ELISA for measurement of MDA-LDL in human plasma.

The involvement of oxidatively modified low density lipoprotein (LDL) in the development of CHD is widely described. We have produced two antibodies, recognizing the lipid oxidation product malondialdehyde (MDA) on whole LDL or ApoB-100. The antibodies were utilized in the development of an ELISA for quantitation of MDA-LDL in human plasma. Intra- and inter-assay coefficients of variation (% CV) were measured as 4.8 and 7.7%, respectively, and sensitivity of the assay as 0.04 micro g/ml MDA-LDL. Recovery of standard MDA-LDL from native LDL was 102%, indicating the ELISA to be specific with no interference from other biomolecules. Further validation of the ELISA was carried out against two established methods for measurement of lipid peroxidation products, MDA by HPLC and F(2)-isoprostanes by GC-MS. Results indicated that MDA-LDL is formed at a later stage of oxidation than either MDA or F(2)-isoprostanes. In vivo analysis demonstrated that the ELISA was able to determine steady-state concentrations of plasma MDA-LDL (an end marker of lipid peroxidation). A reference range of 34.3 +/- 8.8 micro g/ml MDA-LDL was established for healthy individuals. Further, the ELISA was used to show significantly increased plasma MDA-LDL levels in subjects with confirmed ischemic heart disease, and could therefore possibly be of benefit as a diagnostic tool for assessing CHD risk.

Deoxycytidine glyoxal: lesion induction and evidence of repair following vitamin C supplementation in vivo.

Oxidative DNA damage is postulated to be involved in carcinogenesis, and as a consequence, dietary antioxidants have received much interest. A recent report indicates that vitamin C facilitates the decomposition of hydroperoxides in vitro, generating reactive aldehydes. We present evidence for the in vivo generation of glyoxal, an established product of lipid peroxidation, glucose/ascorbate autoxidation, or free radical attack of deoxyribose, following supplementation of volunteers with 400 mg/d vitamin C. Utilizing a monoclonal antibody to a deoxycytidine-glyoxal adduct (gdC), we measured DNA lesion levels in peripheral blood mononuclear cells. Supplementation resulted in significant (p =.001) increases in gdC levels at weeks 11, 16, and 21, with corresponding increases in plasma malondialdehyde levels and, coupled with previous findings, is strongly suggestive of a pro-oxidative effect. However, continued supplementation revealed a highly significant (p =.0001) reduction in gdC levels. Simultaneous analysis of cyclobutane thymine dimers revealed no increase upon supplementation but, as with gdC, levels decreased. Although no single mechanism is identified, our data demonstrate a pro-oxidant event in the generation of reactive aldehydes following vitamin C supplementation in vivo. These results are also consistent with our hypothesis for a role of vitamin C in an adaptive/repair response and indicate that nucleotide excision repair specifically may be affected.

Inter- and intra-individual variation in urinary biomarker concentrations over a 6-day sampling period. Part 1: metals.

The aim of the current HBM-study is to further the understanding of the impact of inter- and intra-individual variability in HBM surveys as it may have implications for the design and interpretation of the study outcomes. As spot samples only provide a snapshot in time of the concentrations of chemicals in an individual, it remains unclear to what extent intra-individual variability plays a role in the overall variability of population-wide HBM surveys. The current paper describes the results of an intensive biomonitoring study, in which all individual urine samples of 8 individuals were collected over a 6-day sampling period (a total of 352 unique samples). By analyzing different metals (As, Cd, Mn, Ni) in each individual sample, inter- and intra-individual variability for these four metals could be determined, and the relationships between exposure, internal dose, and sampling protocol assessed. Although the range of biomarker values for different metals was well within the normal range reported in large-scale population surveys, large intra-individual differences over a 6-day period could also be observed. Typically, measured biomarker values span at least an order of magnitude within an individual, and more if specific exposure episodes could be identified. Fish consumption for example caused a twenty- to thirty-fold increase in urinary As-levels over a period of 2-6h. Intra-class correlation coefficients (ICC) were typically low for uncorrected biomarker values (between 0.104 and 0.460 for the 4 metals), but improved when corrected for creatinine or specific gravity (SG). The results show that even though urine is a preferred matrix for HBM studies, there are certain methodological issues that need to be taken into account in the interpretation of urinary biomarker data, related to the intrinsic variability of the urination process itself, the relationship between exposure events and biomarker quantification, and the timing of sampling. When setting up HBM-projects, this expected relationship between individual exposure episode and urinary biomarker concentration needs to be taken into account.

Inter- and intra-individual variation in urinary biomarker concentrations over a 6-day sampling period. Part 2: personal care product ingredients.

An intensive study was conducted to provide data on intra- and inter-individual variation in urinary excretion of a series of ingredients in personal care products (parabens, triclosan, benzophenones) and bisphenol A (BPA, not expected to be an ingredient) in 8 volunteers over 6 days. Exposure diaries recorded use of personal care products with identified target analytes as ingredients. Participants' usual products were replaced with products without the target analytes for 2 of the 6 days. Urine void volumes and times were recorded. Methyl, ethyl, and n-propylparabens, triclosan, benzophenone-3, and BPA were frequently detected (≥70% of samples). Urinary concentrations of the parabens and triclosan were lower on product replacement days. First morning void concentrations correlated moderately to highly with 24-h composite concentrations for all analytes. Intraclass correlation coefficients (ICCs) for spot samples collected on days with usual product use were low for BPA (0.15), moderate for n-propylparaben and methylparaben (0.39 and 0.56, respectively), and high for ethylparaben, benzophenone-3, and triclosan (0.76, 0.81, and 0.934, respectively); ICCs were consistently higher on the basis of cr-adjusted concentrations. Hydration status adjustment methods were assessed by comparing unadjusted and adjusted concentrations to urinary excretion rates (ER, ng/kg-h) for all analytes and samples. Specific gravity-adjusted concentrations correlated slightly better with ER than creatinine-adjusted concentrations. Within-individual variation in biomarker concentrations was highest for methyl and ethylparabens (2 orders of magnitude variation in spot sample concentrations) and lower for the other analytes (1-1.5 orders of magnitude). This dataset provides insight into the design and interpretation of urinary biomonitoring studies for non-persistent chemicals.

Determination of ethylenethiourea in urine by liquid chromatography-atmospheric pressure chemical ionisation-mass spectrometry for monitoring background levels in the general population.

This study reports a sensitive analytical method suitable for the quantitative analysis of ethylenethiourea (ETU) in human urine and its application to samples from the general population. Sample preparation involved the use of diatomaceous earth extraction columns to remove matrix interferences. Quantification was achieved by liquid chromatography-mass spectrometry using positive ion atmospheric pressure chemical ionisation. Within-day and between-day variability of 14% (n=10) and 11% (n=6), respectively, were obtained at 98 nmol/l (10 µg l(-1)). The assay was linear over the investigated range 2.5-245 nmol/l, with a limit of detection of 2.5 nmol/l. The method was applied to monitoring background levels of ETU in urine samples from the general population in the UK. Results obtained from 361 spot samples contained ETU levels ranging from less than the detection limit (54% of samples) to a maximum of 15.8 µmol/mol creatinine (14.3 µg/g creatinine). The 95th percentile was 5.7 µmol/mol creatinine (5.2 µg/g creatinine).

Antiserum detection of reactive carbonyl species-modified DNA in human colonocytes.

Polyunsaturated fats have been linked to occurrences of sporadic colon cancer. One possible cause may be degradation of polyunsaturated fats during cooking, resulting in multiple reactive carbonyl species (RCS) that can damage nuclear DNA and proteins, particularly in rapidly dividing colon crypt cells. This study describes a novel antiserum against RCS-modified DNA, with apparent order of reactivity to DNA modified with 4-hydroxy-trans-2-nonenal > glyoxal > acrolein > crotonaldehyde > malondialdehyde; some reactivity was also observed against conjugated Schiff base-type structures. Anti-(RCS-DNA) antiserum was successfully utilised to demonstrate formation of RCS-DNA in a human colon cell model, exposed to RCS insult derived from endogenous and exogenous lipid peroxidation sources. Further utilisation of the antiserum for immunohistochemical analysis confirmed RCS-modified DNA in crypt areas of 'normal' colon tissue. These results fully support a potential role for dietary lipid peroxidation products in the development of sporadic colon cancer.