Biomarkers for Assessing Risk of Cancer

Biomarkers of Exposure

Biomarkers of exposure can provide the most direct evidence of human exposure to a carcinogen as well as internal dose and biologically effective dose, affirm exposure-cancer associations established from traditional epidemiological approaches, and provide biological plausibility for the observed exposure-cancer association.

Biomarkers of Effect

Biomarkers of effect measure early biological alterations that occur in the time frame between exposure and cancer development and are also known as biomarkers of intermediate endpoints or intermediate biomarkers. Historically, the most commonly studied biomarkers of effect in surrogate tissues are those related to genotoxicity endpoints, such as chromosome aberrations (CAs) and micronucleus (MN) formation in peripheral blood lymphocytes (PBLs). A number of prospective epidemiological studies have reported positive associations between elevated chromosomal aberrations in PBLs and increased cancer risk. Importantly, a joint nested case-control study using subjects from two prospective cohort studies found that CA level in PBLs could predict future cancer development independent of environmental exposures (smoking and occupational exposure), supporting the notion that chromosomal instability markers could serve as both biomarkers of effect and biomarkers of susceptibility. A few prospective studies showed that individuals with higher frequencies of MN in PBLs had 1.5- to 2-fold increased relative risk of cancer compared with the low-frequency subjects.

Among various chromosome aberrations, chromosome translocation is one of the most well-established biomarkers of exposure and effect. Translocations have been observed in nearly every cancer type, and a single chromosome translocation can cause cancer—for example, Philadelphia translocation causes chronic myelogenous leukemia. Chromosome translocation can survive mitosis and is the most persistent of all the different types of chromosome exchanges. PBLs with radiation-induced translocations persisted for several decades in Japanese atomic bomb survivors. In contrast, other types of chromosome exchanges, for instance, dicentrics and acentric fragments, are unstable because they encounter difficulties during mitosis, which results in the affected cells being killed and eventually removed from the population, making them undetectable in long-term atomic bomb survivors.

The mutational analyses of cancers with distinct environmental exposure provide elegant examples of biomarkers of effect. Exposure-specific mutational fingerprints in tumor suppressor genes have been known for many years. The TP53 gene is the most frequently mutated gene in human cancers, and its mutational spectrum varies substantially by tumor site, which is at least partially due to distinct environmental exposure. For example, sunlight (UV exposure)-induced TP53 mutations in skin cancer occur exclusively at dipyrimidine sites, including a high frequency of C-to-T transitions and unique CC-to-TT double base changes that do not happen in other malignancies. In lung cancer, TP53 mutational patterns are different between smokers and nonsmokers, with an excess of G-to-T transversions in smoking-associated cancers. In liver cancer, a unique mutation (Arg249Ser) in TP53 was linked to aflatoxin B1 exposure. Moving beyond sequencing of candidate genes, the recent explosion of whole-genome sequencing (WGS) of cancer provides further evidence for profound effects of environmental exposure on the cancer genome. In the first WGS of a solid tumor genome, Pleasance and colleagues sequenced a malignant melanoma and a lymphoblastoid cell line from the same person. Consistent with observations in the TP53 gene, of the more than 30,000 base substitutions found in the tumor genome compared to the lymphoblastoid genome, about 70% were C-to-T transitions. Of these, 92% occurred at the 3′ base of dipyrimidine sites, much higher than expected for chance occurrence. These mutations are characteristic of UVB-induced DNA lesions. In the WGS of a small-cell lung cancer cell line and a lung adenocarcinoma, the mutational pattern (an excess of G-to-T mutations, an enrichment of CpG dinucleotides in the G-to-T mutation sets, etc.) is also consistent with that of TP53 and reflects the influences of tobacco carcinogens.

With the breathtaking pace of technological advancements in the biomedical field, and with all the high-throughput “omics” technology, the list of potential biomarkers of effect is long and growing, encompassing molecular and functional changes in the epigenome, transcriptome, proteome, metabolome, and so on. There have been numerous investigations in exploring these technologies in epidemiological studies to identify biomarkers of effect in relation to cancer etiology; however, most of these studies are not validated, and the roles of the potential biomarkers in cancer causation are not clear. For biomarkers of effect (intermediate biomarkers), prospective studies are always preferred over retrospective case-control studies to avoid “reverse causation.” Longitudinal evaluation of sequential samples should be the gold standard compared to cross-sectional and single time point analysis. There are many other issues in terms of study design, biospecimen collection and handling, assay reliability, and data reporting that are particularly important in the study of intermediate biomarkers to avoid spurious results.

Biomarkers of Susceptibility

A major focus of cancer etiological study in recent years has been the determination of cancer susceptibility based on genetic variability. The earliest evidence for genetic susceptibility to cancer came from epidemiological observations of increased cancer risk among relatives of cancer patients. The existence of many rare inherited syndromes that predispose patients to increased risks of certain cancers provides other evidence for genetic susceptibility to cancer. A large classical twin study estimated the genetic heritability of most common cancers to be between 20% and 40%. The identification of a large number of genetic susceptibility loci to common cancers by genome-wide association studies (GWAS) provides the strongest evidence for genetic susceptibility to common cancers. Biomarkers of susceptibility have provided significant biological insight into cancer etiology and may become potential targets for preventive and therapeutic interventions. Biomarkers of susceptibility can also improve risk prediction and identify high-risk populations for targeted surveillance, screening, and prevention. Further investigation of gene-environment interactions is critical to advance the understanding of human carcinogenesis and improve the accuracy of cancer risk prediction.

Earlier analyses in cancer susceptibility have mos舉tly focused on family-based linkage studies to identify high-penetrance genes whose mutations (mutation rate typically less than 0.1% in the general population) cause Mendelian cancer-predisposing syndromes. More recent efforts have mainly focused on association studies that compare variant allele frequencies of common single-nucleotide polymorphisms (SNPs) between a large number of cancer cases and unrelated controls. The underlying hypothesis for such an approach is the “common disease–common variant” (often abbreviated CD-CV) hypothesis, which postulates that genetic susceptibility to common diseases such as cancer is largely due to many common variants (typically having frequencies greater than 5% in the general population) with only modest effect conferred by each allele. The competing hypothesis is the “common disease–rare variant” (CD-RV) hypothesis, which suggests that multiple rare variants (typically having frequencies that lie between approximately 0.1% and 1%, the upper limit for Mendelian mutations and the lower limit of SNPs, respectively) of larger effect cause common diseases. However, recent evidence has suggested that these two hypotheses should not be viewed as an “either/or” choice, but rather that both common and rare variants may contribute to common diseases, and the degree of contribution by either variant depends on the particular disease phenotype. In cancer, for example, and as described later, several rare susceptibility variants for breast cancer have been identified by candidate gene sequencing and genotyping. Furthermore, next-generation sequencing studies have identified novel rare susceptibility variants for other cancers.

Depending on the population frequency of risk alleles and effect size, cancer genetic susceptibility markers can be roughly grouped into three classes: rare high-penetrance mutations, rare low- to moderate-penetrance disease-causing variants, and common low-penetrance SNPs ( Table 21-1 ).