antigen (PSA) test has been widely adopted in clinical practice to screen for prostate cancer. Elevated and, especially, rising PSA indicates cell prolifera-tion in prostatic tissue and, thus, is a marker of disease progression. Simi-larly, a decrease in the level of this marker can be used as an endpoint for interventions. Unfortunately, no comparable markers have been identified for other common cancers.
Molecular genetics may provide new sensitive detection methods in the coming years. Recent studies have shown that circulating tumor DNA (‘‘cell-free DNA’’) can be isolated in peripheral blood and used to detect selected mutations that, individually or as a group, can be highly specific of the solid tumor from which the DNA originated (12). Tumor DNA with signature mutations have also been detected in sputum, breast fluid, and feces. However, it is still unclear how early in the disease process these tests could be used. Another emerging approach, namely proteomics, allows for the screening of hundreds or thousands of proteins in peripheral blood to detect patterns that may be highly specific for certain diseases, including cancers (13).
2.6. Improving Risk Assessment
More homogeneous disease groupings, improved measurement of study variables, and accounting for effect modification due to genetic or acquired susceptibility result in reduced misclassification in epidemiological studies, allowing for risk estimates that are more precise and that pertain to more refined subgroups of the population. Since epidemiological data are used in risk assessment to formulate individual risk functions, molecular epide-miologic studies can increase the validity and specificity of these functions.
Table1ExamplesofMarkersofInternalDose BiomarkersSpecimensExposureHalf-lifeTargetorganReference AflatoxinB1andmetabolitesUrineAflatoxinindietLiver14 ArsenicUrineCoppersmelting1weekLung15 CotinineUrineTobacco3daysLung16 DDESerumDDTSeveral months=yearsBreast17 AscorbicacidPlasmaDietaryvit.CFewhoursAll18 FolateRedblood cellDietaryfolate120daysAll19 SeleniumToenailsDietaryselenium>1yearProstate20 EstradiolPlasmaEndogenous estrogensBreast Source:ModifiedfromRefs.8and21.
eliminated or if the exposure is episodic, multiple measurements may be required in order to reduce the ratio of within- to between-individual vari-ance (see Sec. 4) and estimate a ‘‘usual’’ level. Biomarkers of internal dose are particularly useful in prospective studies because the problem of reverse causality (i.e., disease status affecting the level of the biomarker) that can plague case–control studies is minimized or eliminated. However, markers of internal dose can sometimes be used in case–control studies when it is unlikely that either the exposure or the markers have been affected by the disease process or treatment (i.e., when studying a precursor lesion or early disease stage), or when the marker reflects past (presumably, prediagnostic) exposure (e.g., DDT metabolites in adipose tissue) and is not affected by disease status.
3.2. Markers of Biologically Effective Dose
In contrast to markers of internal dose, which measure the internal level of a compound or its metabolites, markers of biologically effective dose assess the amount of this compound that interacts with critical subcellular or cel-lular targets. Thus, these markers have the advantage of integrating the effects of both exposure and host susceptibility. For example, certain chemi-cals can bind covalently to proteins in the cell to form an adduct (Table 2).
DNA adduct formation often occurs after metabolic activation of a carcino-gen and can be followed by DNA repair. Thus, measurement of adducts can assess both exposure to a specific carcinogen and the individual’s capacity to activate this carcinogen and repair DNA, as well as other possible host fac-tors. Since formation of chemical–DNA adducts are thought to be impor-tant in carcinogenesis, individuals with the highest levels of DNA adducts are expected to be at greater cancer risk. A frequent limitation of adduct studies is that samples of target tissues are often not available and that surrogate tissue needs to be used. DNA adducts have limited applications in case–control studies due to the relatively short life of most adducts evalu-ated to date. However, protein adducts (hemoglobin or albumin adducts) have a longer half-life and, thus, their use may be possible in retrospective studies of early-stage cancers.
3.3. Markers of Susceptibility/Resistance
Cancer families have been noted for centuries. Linkage studies followed by positional cloning have allowed the identification of the genes responsible for a number of familial cancer syndromes, such as retinoblastoma, Wilm’s tumor, Li–Fraumeni syndrome, Von Hipple–Lindau disease, familial ade-nomatous polyposis (FAP), hereditary nonpolyposis colorectal cancer (HNPCC), and familial breast–ovary cancers (44). These mutations are typi-cally rare in populations but carry a high disease risk (high penetrance).
Table2ExamplesofMarkersofBiologicallyEffectiveDose Adductsa ExposureBiospecimenb PopulationReference AlkylatedHbPropyleneoxideRBCWorkers22 4-Aminobiphenyl–HbCigarettesmokingRBCSmokers23 AFB1–guanineDietUrineChinese24 AFB1–DNADietLivertissueTaiwanese25 PAH–DNAPAHincigarettesmoke= environmentWBCLungcancerpatients, smokers,workers26 NNK–HbCigarettesmokeRBCSmokers27 aAFB1,aflatoxinBI;PAH,polycyclicaromatichydrocarbon;NNK,4-(methylnitrosamino)-1-(3-pyridyl)-1-butanone. bRBC,redbloodcells;WBC,whitebloodcells. Source:ModifiedfromRef.8.
They can be used for predictive genetic testing in situations where preventive interventions are possible.
Potentially more relevant to public health are low-penetrance but com-mon susceptibility genes, such as those listed in Table 3. Examples include genetic polymorphisms associated with interindividual differences in the metabolism of xenobiotics, DNA repair, or metabolism of hormones or key nutrients. Markers of susceptibility to behavioral exposures include genetic polymorphisms controlling metabolic processes affecting one’s adoption of healthy or unhealthy lifestyle habits (e.g., toxic reaction to alco-hol curtailing ethanol consumption; biological basis for susceptibility to nicotine or ethanol addiction).
Measuring enzymes or hormones that are thought to be the basis for the susceptibility, directly by assessing their levels in plasma or tissue, or
Table 3 Examples of Low Penetrance Susceptibilty Genes
Mechanism Gene examples Cancer Reference
Dominant oncogene
ras, myc Lung 28
Tumor suppressor genes
p53, rb Lung, bladder 29
Carcinogen activation
CYP1A1 Lung 30
CYP1A2 Bladder, colon 31
CYP2E1 Lung, NPC 30, 32, 33
NAT2 Colon 31
Carcinogen detoxification
GSTM1 Lung, bladder 34
NAT2 Bladder 35
Hormone metabolism
CYP17 Breast 36
Hormone receptor
Androgen receptor Prostate 37
Vitamin metabolism
Vitamin D receptor Prostate 38
MTHFR Colon 39
Alcohol metabolism
ADH, ALDH Oral 40
Addiction Dopamine receptors Smoking related cancers
41
DNA repair XP, AT hOGGI Skin, Burkitt
lymphoma, lung
43
Source: Modified from Ref. 44.
indirectly by using a pharmacological probe (e.g., caffeine for NAT2 or CYP1A2 activity, chlorzoxazone for CYP2E1), is often possible in order to characterize the phenotype of interest. Such phenotyping assays are often difficult to use in case–control studies as the disease or its treatment may affect these measurements. In contrast, genotyping using genomic DNA has often been favored in case–control studies since the genotype of an indi-vidual is determined at birth and remains unaffected by disease or treat-ment. However, even with genotyping, the prospective design remains optimal when studying rapidly lethal diseases because of biases resulting from potential differential survival (e.g., due to differences in response to therapy by genotypes (45)) that may plague case–control studies. Moreover, the genes studied may convey an increased or decreased risk only in indivi-duals who have been exposed (gene–environment interaction); hence, the importance of carefully assessing exposure. Compared to case–control stu-dies, prospective studies offer the advantage of generating exposure data that are devoid of recall bias since this information is collected before the disease develops.
3.4. Markers of Early Biological Effects
Markers of early biological effects represent processes that are intermediate on the etiological pathway between exposure and clinically detectable disease.
Examples of such markers are given in Table 4. These markers may help iden-tify a mechanistic link between exposure and disease. They may also be used as disease surrogates for intervention studies or as screening tools for primary
Table 4 Examples of Markers of Early Biological Effect
Marker Exposure Biospecimen Reference
Sister chromatid exchange
Industrial, radiation
WBC 46
Micronuclei Cigarette smoke WBC 47
Betal quid Buccal cells 48
Chromosomal aberrations
Industrial, radiation
WBC 49
Mutations in tumor suppressor genes
Codon 249serp53 Dietary AFB1 Liver cells 50 p53 hot spot
mutations at codons 157, 248 and 273
Benzo[a]pyrene Bronchial epithelial cells
51
CC to TT mutation in p53
UV Skin 52
and secondary prevention. For example, somatic mutations may be identified in target genes (e.g., p53), providing evidence of irreversible genetic damage.
In some cases, specific mutations in the gene may indicate exposure to specific agents (DNA fingerprints) or may point to specific mechanisms. They may also be used clinically as prognostic factors or as endpoints in intervention studies evaluating genetic responses to various exposures.
Chromosomal aberrations are less specific markers but they are also thought to be intermediate in the etiological pathway to cancer. They have been used as markers of exposure or to evaluate individual sensitivity to mutagens or carcinogens.