Overview Of Biomarkers

As adapted from a report by the Committee on Biological Markers of the National Research Council (6), the development of disease that results from exposure to an environmental agent or other toxicant is multistage, starting with exposure, progressing to internal dose (e.g., deposited body dose), to biologically effective dose (e.g., dose at the site of toxic action), to early biological effect (e.g., at the subcellular level), to altered structure or function (e.g., subclinical changes), and finally, to frank, clinical disease. Any step in this process may be modified by host-susceptibility factors, including genetic traits and effect modifiers, such as diet or other environmental exposures. Therefore, biomarkers are indicators of events for physiological, cellular, subcellular, and molecular alterations in the multistage development of specific diseases (6).

Molecular biomarkers are typically used as indicators of exposure, effect, or susceptibility (6, 7). A biomarker of exposure refers to measurement of the specific agent of interest, its metabolite(s), or its specific interactive products in a body compartment or fluid, which indicates the presence (and magnitude) of current and past exposure. A biomarker of dose bears a quantitative/qualitative relationship with exposure. Such a biomarker may be an exogenous substance, an interactive product (e.g., between a xenobiotic compound and endogenous components), or other indicator. A biomarker of effect indicates the presence (and magnitude) of a biological response to exposure to an environmental agent. Such a biomarker may be an endogenous component, a measure of the functional capacity of the system, or an altered state recognized as impairment or disease. A biomarker of susceptibility is an indicator or a measure of an inherent or acquired ability of an organism to respond to the challenge of exposure to a specific xeno-biotic substance or other toxicant. Such a biomarker may be the unusual presence or absence of an endogenous component or an abnormal functional response to an administered challenge. Molecular epidemiology and molecular dosimetry thus have great utility in addressing the relationships between exposure to environmental agents and development of clinical diseases and in identifying those individuals at high risk for the disease (1, 8).

The validation and application of molecular biomarkers for environmental agents should be based on specific knowledge of metabolism, interactive product formation, and general mechanisms of action (9, 10). Examples are studies on the relationships between tobacco smoking and lung cancer (11) and between afla-toxin exposure and liver cancer (12). A specific application of biomarker technology to human cancer is the study of the variation in response among individuals after exposures to tobacco products. For example, even in heavy tobacco smokers, less than 15% of these people develop lung cancer (13); thus, intrinsic susceptibility factors must affect the time course of disease development and eventual outcome. The identification of those at highest risk for developing cancers will be facilitated by biomarker studies. Extensive efforts have been made in the identification of these high-risk individuals through the use of various genetic and metabolic susceptibility markers, e.g., measurement of polymorphism of genotype and phenotype of various enzymes involved in phase I and phase II metabolic reactions of certain known carcinogens (14-17). This strategy has not yet proven to be broadly applicable to many other human diseases, although progress is being made for many types of cancers (2).

The validation of any biomarker-effect link requires parallel experimental and human studies (12). Ideally, an appropriate animal model is used to determine the associative or causal role of the biomarker in the disease or effect pathway, and to establish relations between dose and response. The putative biomarker can be validated in pilot human studies, where sensitivity, specificity, accuracy, and reliability parameters can be established. Data obtained in these studies can then be used to assess intra- or interindividual variability, background levels, relationship of the biomarker to external dose or to disease status, as well as feasibility for use in larger population-based studies. It is important to establish a connection between the biomarker and exposure, effect, or susceptibility. To fully interpret the information that the biomarker provides, prospective epidemiologi-cal studies may be necessary to demonstrate the role that the biomarker plays in the overall pathogenesis of the disease or effect. To date, few biomarkers have been rigorously validated using this entire process.

A. Biomarkers of Exposure

Although biomarkers of exposure can refer to any biomarker used to estimate current or past exposure to a specific environmental agent, the traditional definition of an exposure biomarker involves measurement of a xenobiotic, its metabolite, or its interactive products found in body tissue, fluids, and excreta, such as blood, urine, feces, or milk (18). These measures provide information about the actual concentration or internal dose of a specific agent that has been absorbed and distributed in the body. Measurement of the xenobiotic itself or its metabolites has been incorporated into a number of human epidemiological studies. For example, excretion of aflatoxin M1, one of the major metabolites of aflatoxin B1, has been used as a biomarker for the evaluation of human exposure to aflatoxin, and this marker was found to be associated with the risk of liver cancer (19, 20). Specific metabolites of one of the tobacco-specific nitrosamines, 4-(methyl-nitrosamino)-1-(3-pyridyl)-1-butanone (NNK), a potent chemical carcinogen, have been detected and quantified in the urine of smokers, but these metabolites were not found in the urine of nonsmokers (21). Intraindividual and interindividual variations in these metabolites of NNK in smokers' urine were noted, and this might be important in disease risk (22). Other examples include the measurement of blood and serum levels of heavy metals and pesticides (23, 24), such as DDE (1,1-dichloro-2,2-bis(p-chlorophenyl)-ethylene), the major metabolite of DDT (2,2-bis[p-chlorophenyl]-1,1,1-trichloroethane), which has been used as a biomarker in breast cancer studies in women (25, 26).

The metabolically activated ultimate forms of environmental carcinogens can covalently interact with cellular macromolecules such as DNA and proteins (27-30). These carcinogen-macromolecular adducts have an important role in human biomonitoring and molecular epidemiological studies (2, 9). They are specific biomarkers that provide a way to measure human exposure to these chemical carcinogens and provide information about specific dose to a carcinogen target site (DNA or protein). Moreover, it may be possible to establish a correlation between tumor incidence and exposure by measuring these adducts' level (5). Many different analytical methods have been developed to identify and measure carcinogen-macromolecular adducts, including enzyme-linked immunosorbent assay (ELISA), radioimmunoassay (RIA), and immunohistochemical staining assay (IHC), radiometric postlabeling methods, high performance liquid chromatography (HPLC), gas chromatography/mass spectrometry (GC/MS), liquid chromatography/mass spectrometry (LC/MS), and fluorescence spectroscopy (FS) (2, 8, 31, 32). These techniques have been used to measure composite and specific DNA adducts in cellular DNA isolated from peripheral lymphocytes, bladder, breast, lung, and colonic tissues, as well as excreted DNA adducts in urine. These methods are also used to measure hemoglobin (Hb) and albumin adducts in blood samples of people exposed to environmental carcinogens and xenobiotics. In addition, these techniques have been applied in the clinical setting to examine carcinogen-macromolecular adducts of people undergoing chemotherapy with alkylating agents, in an attempt to associate adduct levels with clinical outcome (33, 34). Recently, these methods have also been applied to human clinical trials to validate various intervention tools for the assessment of chemo-preventive agents in modulating various intermediate biomarkers (12).

In addition to the use of biomarkers for exogenous exposure, it is desirable to have a biomarker of endogenous oxidative damage, as many studies have found that endogenous oxidative DNA damage correlates with the formation of chronic degenerative diseases, including cancer (35). Among the many oxida-tively damaged DNA bases formed, 8-oxo-2-deoxyguanosine is a lesion that can be sensitively measured. Several techniques, including HPLC-electrochemical detection (HPLC-EC), have been developed and applied to detect this damage product in biofluids and tissue samples from animals and humans (36). Immu-noaffinity column methods have also been described for the analysis of oxidative damage products of nucleic acids excreted in urine (37). Quantitative analysis of these adducts in urine and tissues may eventually be used to assess the risk of disease from oxidative damage in people.

B. Biomarkers of Effect

A biomarker of effect has been defined by The International Programme on Chemical Safety as a measurable biochemical, physiological, behavioral, or other alteration within an organism that, depending on the magnitude, can be recognized as associated with an established or possible health impairment or disease (38). Although this broad definition covers the wide spectrum of functional alterations, in practice, biomarkers of effect represent changes at the subcellular level, particularly at the chromosomal and molecular levels, such as cytogenetic alterations and gene mutations (18).

Cytogenetic biomarkers currently applied in molecular epidemiological studies include chromosome aberrations (CAs), sister chromatid exchanges (SCEs), and micronuclei (MN). Chromosome aberrations are structural alterations, breaks, and rearrangements in chromosomes. Recently developed analytic methods using fluorescent in situ hybridization (FISH) and polymerase chain reaction (PCR) provide detection of new types of rearrangements and translocations of specific regions in certain chromosomes (39, 40). Exposure to ionizing radiation, alkylating cytostatics, tobacco smoking, benzene, and styrene has been found to induce CAs in humans (18). Sister chromatid exchange represent symmetrical exchanges of DNA segments between the sister chromatids of a duplicated metaphase chromosome. Tobacco smoking, alkylating cytostatics, and ethylene oxide can induce SCEs in human lymphocytes (41). Micronuclei are small additional nuclei observable in interphase cells. Increased MN frequencies in human lymphocytes have been found after exposure to ionizing radiation and formaldehyde (42, 43).

Biomarkers of gene mutations include somatic mutations in surrogate tissues and gene mutations in target tissues. The hypoxanthine phosphoribosyl-transferase (HPRT) gene mutation and the glycophorin A (GPA) assay are two somatic gene mutation assays currently applied in molecular epidemiological studies for human risk monitoring (18). Gene mutations, particularly in critical target genes, are important biomarkers of biological effects, altered function, and preclinical disease (44, 45). Activation of protooncogenes and inactivation of tumor suppressor genes caused by mutations are critical genetic changes linked to eventual cancer formation. For example, the ras protooncogenes are targets for many genotoxic carcinogens. Activation of ras is an early event, possibly the initiating step, in the development of many chemical carcinogen-induced rodent tumors (46). The ras oncogene is also observed in different types of human tumors and at a higher frequency than any other oncogene. Both the activation of ras oncogenes and the inactivation of several suppressor genes, including p53, have been found in the development of human colon and lung cancers (47).

The tumor suppressor gene p53, the most commonly mutated gene detected in human cancers, has been used as a biomarker for molecular carcinogenesis, molecular epidemiology, and cancer risk assessment (48, 49). The number and type of mutations in this gene are not equally distributed but occur in specific hotspots that vary with tumor type (50). The differences in mutation patterns between tumors are consistent with different etiologies for the specific tumor types. One striking case in this research field is the studies on the relationship between aflatoxin exposure and development of human hepatocellular carcinoma (HCC), as summarized below.

The initial results from three independent studies of p53 mutations in HCCs occurring in populations exposed to high levels of dietary aflatoxin showed high frequencies of G to T transversions, with clustering at codon 249 (51-53). On the other hand, studies of p53 mutations in HCCs from Japan and other areas where there is little exposure to aflatoxin revealed no mutations at codon 249 (54). These studies provided the circumstantial, but as yet unproven, linkage between this signature mutation of aflatoxin exposure and the events detected in p53 in liver tumors from China and Southern Africa.

Fujimoto et al (55) further examined the hypothesis that exposure to aflatoxin B1 (AFB1), either alone or coincident with other environmental carcinogens, might be associated with occurrence of allelic losses during the development of HCC in China. The HCC tissues were obtained from two different areas in China: Qidong, where exposure to hepatitis B virus (HBV) and AFB1 is high, and Beijing, where exposure to HBV is high but to AFB1 is low. They analyzed the tumors for mutations in the p53 gene and loss of heterozygosity for the p53, Rb, and APC genes. The frequencies of mutation, loss, and aberration

(either mutation or loss) of the p53 gene in 25 HCC specimens from Qidong were 60%, 58%, and 80%, respectively. The frequencies in nine HCC specimens from Beijing were 56%, 57%, and 78%; however, the frequency of a G to T transversion at codon 249 in HCCs from Qidong and Beijing were 52% and 0%, respectively. These data indicate that mutation or loss of heterozygosity in the p53 gene, independent of the codon 249 mutation, plays a critical role in the development of HBV-associated HCCs in China. These results also show distinct differences in the pattern of allelic losses between HCCs in Qidong and Beijing, and suggest that AFB1 and other environmental carcinogens may contribute to this difference.

The observation of the codon 249 mutation in p53 with aflatoxin exposure is not limited only to China and Southern Africa. Senegal, a country where liver cancer incidence is one of the highest in the world, has high exposure to afla-toxins. Fifteen liver cancer samples were examined for mutation at codon 249 of the p53 gene (56). Nontumoral DNA from the patients showed a wild type genotype. Mutation at codon 249 of the p53 gene was detected in 10 of the 15 tumor tissues tested (67%). This frequency of mutation in codon 249 of the p53 gene is the highest described to date in the literature, and these results confirm that there is an association between countries of high aflatoxin intake and a high frequency of mutation in codon 249 of the p53 gene, and that HBV alone does not contribute to these base changes. Aguilar et al (57) re-examined the role of AFB1 and p53 mutations in HCCs and in normal liver samples from the United States, Thailand, and Qidong, where AFB1 exposures are negligible, low, and high, respectively. The frequency of the AGG to AGT mutation at codon 249 paralleled the level of AFB 1 exposure, which also supports the hypothesis that this mycotoxin has a causative and probably early role in hepatocarcinogenesis.

Results from experimental studies have also linked aflatoxin as a causative agent in the described p53 mutations (58). Previous work had shown that AFB1 exposure causes almost exclusively G to T transversions in bacteria (59), and that aflatoxin-epoxide can bind to codon 249 to p53 in a plasmid in vitro, providing further indirect evidence for a putative role of aflatoxin exposure in p53 mutagen-esis (60). Aguilar et al (61) studied the mutagenesis of codons 247-250 of p53 by rat liver microsome-activated AFB1 in human HCC cells HepG2 using a RFLP/PCR genotypic strategy. They found that AFB1 preferentially induced the transversion of G to T in the third position of codon 249; however, AFB1 also induced G to T and C to A transversions into adjacent codons, although at lower frequencies. Cerutti et al (58) studied the mutability of codons 247-250 of p53 with AFB1 in human hepatocytes using a similar technique. Aflatoxin B1, preferentially induced the transversion of G to T in the third position of codon 249, generating the same mutation found in a large fraction of HCCs from regions of the world with AFB1-contaminated food. These experimental results support AFB1 as an etiological factor for HCCs in AFB1-contaminated areas (45).

C. Biomarkers of Susceptibility

Biomarkers of susceptibility are mainly concerned with factors in kinetics and dynamics of uptake and metabolism of exogenous chemicals. The enzymes involved in activation and detoxification of these xenobiotics are divided into two categories: phase I enzymes, mainly the superfamily of cytochrome P450 mixed function oxidase enzymes, and phase II enzymes, which act on an oxidized substrate to conjugate them with various moieties, such as glucuronic acid, glutathione, and sulfate (62-64). Genetic differences in the expression of these metabolic enzymes could be a major source of interindividual variation in susceptibility to disease (65-67). Therefore, the determination of genotype and phe-notype of these metabolic enzymes in different populations is being studied to determine if an association exists between exogenous exposure and formation of specific disease in specific metabolic genotype subsets. Many studies over the past several years have found that genes involved in xenobiotic metabolisms, including the cytochrome P450 enzymes CYP1A1, CYP1A2, CYP2A6, CYP2D6, CYP2E1, N-acetyltransferase 1 and 2 (NAT1 and NAT2), and glutathione S-transferase and theta (GSTM1 and GSTT1), are polymorphic in human populations, and in some cases specific alleles are associated with increased risk of a variety of different cancers. Further, studies on the genotypes for human cyto-chrome P450 enzymes in diverse populations have also found ethnic differences in transcription and translation of these enzymes (68-70).

The acetylator biomarkers are another group based on metabolic susceptibility genes that could be important in aromatic and heterocyclic amine exposures (71, 72). Acetyltransferases catalyze both N- and O-acetylation reaction. N-acetylation is a detoxification step for arylamines like 2-naphthylamine and 4-aminobiphenyl (4-ABP), but O-acetylation of N-hydroxy arylamine can form esters that are highly reactive, inducing DNA damage (73-75). Through the use of various biomarkers, such as caffeine metabolites, as indices of acetylator phe-notype, several epidemiological studies have found an association between the slow acetylation phenotype and risk for developing bladder cancer, particularly for those people who are occupationally exposed to aromatic amines (76, 77). Additionally, several studies suggest a relationship between the rapid acetylator phenotype and colon cancer (78, 79).

N-acetyltransferases are coded by two distinct genes located in humans on chromosome 8 and designated as NAT1 and NAT2 (80). Polymorphism in NAT2 results from point mutations in the coding regions of this intronless gene, which phenotypically leads to slow and rapid metabolism (81). Slow acetylators are homozygous for the slow acetylator gene, which was found in 5% of Canadian Eskimos, in 10% to 20% of Japanese, in 50% to 60% of whites, and in 90% of Northern Africans (82), whereas rapid acetylators are either heterozygous or ho-mozygous for the rapid acetylator gene. N-acetyltransferase 1, which codes for the acetyl transferase activity originally thought to be monomorphic, has also been shown to be a polymorphic gene (83, 84). A gene-gene-environmental exposure three-way interaction was recently found in N-acetylation polymorphisms in smoking-associated bladder cancer (85).

Among the most studied cytochrome P450s with respect to polymorphism is CYP1A2, which has also been found to be associated with increased risk to human colorectal and bladder cancer (86). CYP1A2 catalyzes the N-oxidation of several aromatic and heterocyclic amines to DNA reactive species. Although no polymorphic sequences in the structural CYP1A2 gene have been found, the metabolic phenotype of the enzyme has been evaluated by using caffeine metabolism, phenacetin O-deethylation, and theophylline 1-demethylation (87). The use of urinary caffeine metabolites as a biomarker for the activity of the enzyme has been validated in several epidemiological studies. The rationale for developing this method was that the initial step in the biotransformation of caffeine (caffeine 3-demethylation) is catalyzed by CYP1A2. The ratio of either [1,7-dimethylxan-thine (17X) + 1,7-dimethylurate (17U)]/[caffeine (137X)], examined in urine 4 to 5 hours after caffeine ingestion (88), or [5-acetylamino-6-amino-3-methyluracil (AAMU) + 1 -methylxanthene (1X) + 1-methylurate (1U)]/17U, measured 24 hours after caffeine ingestion (89), has been used as a marker of CYP1A2 pheno-type in human studies. Except in a Japanese population, the CYP1A2 phenotype distribution appears to be trimodal (slow, intermediate, and rapid metabolizers).

The simultaneous determination of NAT2 and CYP1A2 phenotypes has also been accomplished by measuring different caffeine metabolites as markers of hepatic CYP1A2 and NAT2. The NAT2 phenotype can be determined using the ratio of 5-acetylamino-6-formylamino-3-methyluracil (AFMU)/1X in urine (87, 88). Polymorphisms in both NAT2 and CYP1A2 are associated with susceptibility for various types of human cancers (86). For example, in a case-control study involving patients with a history of colorectal cancer or polyps, a trend toward an increased proportion of rapid acetylators for NAT2 was observed. In addition, a significantly greater percentage of rapid metabolizers for CYP1A2 was found in the cancer and polyp cases than in the controls (90). When comparing the prevalence of individuals who were both rapid acetylators and rapid metabo-lizers for CYP1A2, 33% of the patients with colorectal cancer or polyps possessed the rapid/rapid phenotype, compared with only 13% of the controls (72).

Human cytosolic glutathione S-transferases (GSTs) belong to a supergene family of enzymes consisting of at least four distinct classes, named alpha (a), mu (p.), pi (n), and theta (9) (91). The GSTs are regulated by at least seven gene loci, and genetic polymorphisms have been found in the p and 9 class isoenzymes. The p class isoenzymes include products of the GSTM1 locus that were found to be deleted in approximately half of populations of diverse ethnic origins (92, 93). Because this class of enzymes catalyzes the conjugation of reduced glu-tathione to reactive electrophilic substrates, including metabolic activating prod ucts of benzo(a)pyrene (94), individuals genetically lacking GSTM1 activity could be at enhanced risk for various carcinogen-related cancers. Several human studies using phenotype/genotype determinations have found that lack of GSTM1 activity or its gene is, at least in part, related to the genetic susceptibility to tobacco-related lung cancers, particularly those resulting in adenocarcinomas (95-97). Weak associations with increased cancer risk have also been found for other carcinogen-exposed cohorts. A recent study found that the GSTM1 null genotype is associated with susceptibility to gastric adenocarcinoma and distal colorectal adenocarcinoma in a Japanese population (98). Moreover, the GSTM1 null genotype was found to be associated with a high inducibility of CYP1A1 gene transcription (99). A similar genetic polymorphism has been reported for the GST 9 GSTT1) gene locus (100, 101). The GSTT1 null genotype has been linked to induced genetic damage by 1,3-butadiene and alkylhalide exposure (102). A recent case-control study found that the frequency of the GSTT1 null type was significantly increased in colorectal cancer cases and was also associated with an increased susceptibility to total ulcerative colitis (103).

In addition to NAT, CYP1A2, and GST, polymorphisms in CYP1A1 are associated with higher risk for human lung (104-107), laryngeal (108), and colon (109) cancers in specific population. Specific CYP2D6 genotypes and pheno-types were noted to be associated with higher cancer risks in lung and larynx (110-112). Different polymorphisms in CYP2E1 were also linked with gastrointestinal, lung, and nasopharyngeal cancers (16, 113-116). The molecular bases for most of these polymorphisms are mutations in the structural genes that result in unstable enzymes or enzymes with altered catalytic activities. However, the polymorphism may also be caused by other regulatory factors that control gene expression.

Another type of biomarker of susceptibility is DNA repair capacity (DRC). The recognition that genetically determined individual DRC may influence the rate of removal of DNA damage and of fixation of mutations prompts the development of methods to measure this important host-susceptibility factor. Several assays, including the unscheduled DNA synthesis (UDS) and measurement of DNA adduct persistence, have been applied previously to detect carcinogen-induced DNA repair or the capacity of global DNA repair (117). A host cell reactivation assay for examining the proficiency of DNA repair in human lymphocytes has been developed and evaluated in epidemiological studies (118). In this assay, a chemically damaged plasmid DNA containing a reporter gene, chloram-phenicol acetyltransferase (CAT), is transfected into human lymphocyte samples obtained from an individual. After an incubation period to allow for repair and expression of the reporter gene, DNA excision-repair capacity is scored by determining the amount of reactivated CAT enzyme activity. Human cells with decreased ability to repair the damaged DNA will have lower expression of the reporter gene and, hence, lower enzymatic activity. This technique has been used to compare the DNA-repair capacities of basal cell carcinoma (BCC) in skin cancer patients and controls (119). An age-related decline in DNA-repair capacity was detected, and reduced repair capacity was a particularly important risk factor for young individuals with BCC and for those with a family history of skin cancer. Younger individuals with BCC repaired DNA damage poorly when compared with controls. With increasing age, however, differences between cases and controls gradually disappeared (120). The normal decline in DNA repair observed with increasing age may account for the increased risk of skin cancer that begins in middle age, which suggests that the occurrence of skin cancer in the young may represent a biochemical manifestation of decreased repair capacity. The same technique has been applied in another case-control study (121) of 51 newly diagnosed lung cancer patients and 56 age-, sex-, and ethnicity-matched controls. The mean level of DRC in cases (3.3%) was significantly lower than that in controls (5.1%) (P < 0.01). Only nine cases (18%) had DNA repair levels higher than the controls. Using the median level of DRC in controls as the cutoff value for calculating the odds ratio (OR), the cases were almost six times more likely than the controls to have reduced DRC after adjustment for age, sex, ethnicity, and smoking status.

In addition, a study (122) of 21 lung cancer patients and 41 healthy controls evaluated induced DNA adducts profiles in peripheral lymphocytes exposed to benzo[a]pyrene-7,8-diol-9,10-epoxide (BPDE), the ultimate carcinogenic metabolite of benzo[a]pyrene. The peripheral lymphocytes of cancer patients tended to accumulate higher levels of BPDE-DNA adducts than those of controls. Using the relative adduct labeling value of controls (10 adducts/107 nucleotides) as the cutoff point, 18 of 21 cases and 23 of 41 controls distributed above this level (OR 4.7). In a logistic regression analysis, the level of induced adduct was an independent risk factor (OR 6.4) after adjustment for potential confounding factors, that is, age, sex, ethnicity, and smoking. The significant association between accumulated BPDE-induced DNA adducts and risk for lung cancer suggests that this assay may also be applicable to other carcinogen-related cancer studies.

Mutagen sensitivity as a susceptibility biomarker has also been reported in triple primary cancers (123) and in HCCs (124). A reported study of 28 patients with HCC and 110 healthy controls found that the mean numbers of bleomycin-induced breaks per cell for cases and controls were 0.92 and 0.55, respectively (P < 0.001). For BPDE sensitivity, the values were 0.90 for cases and 0.46 for controls (P < 0.001). Nearly 68% of the cases, but only 27% of the controls, exhibited bleomycin sensitivity. Eighty percent of the case group, but only 22% of the control group, exhibited BPDE sensitivity. Multivariate analyses found that both bleomycin sensitivity and BPDE sensitivity were associated with significantly elevated risks for HCC, with ORs (95% CI) of 5.63 (2.30, 13.81) and 14.13 (3.52, 56.68) respectively. For individuals who were sensitive to both assays, the risk was 35.88 (123).

5 Easy Ways To Stop Smoking

5 Easy Ways To Stop Smoking

Your first day without cigarettes can be difficult, but having a plan will make it easier! Learn what steps to take on the day you quit smoking.

Get My Free Ebook

Post a comment