![[Feedback]](/icons/banner/feedbackACT.gif){kind=icon}
![[For Subscribers]](/icons/banner/subscriberACT.gif){kind=icon}
![[Archive]](/icons/banner/archiveACT.gif){kind=icon}
![[Search]](/icons/banner/searchACT.gif){kind=icon}
![[Contents]](/icons/banner/tocACT.gif){kind=icon}
|
|
|
|
|
||
Nutritional Epidemiology Branch, Division of Cancer Epidemiology and Genetics, National Cancer Institute, Bethesda, MD 20815
* To whom correspondence should be addressed. E-mail: schatzka{at}mail.nih.gov.
 |
ABSTRACT |
|---|
 TOP ABSTRACT LITERATURE CITED |
|---|
In the context of cancer survivors, a biomarker can be defined as a biological phenomenon that has some relation to a subsequent clinical event such as recurrence or death. A biomarker in survivors may reflect some biological characteristic of the diagnosed tumor or some other biological phenomenon in blood or other nontumor tissue.
A biomarker in survivors can be valuable for 1) clinical management and 2) understanding and preventing adverse outcomes (a form of etiologic research). There is considerable overlap in these 2 functions, and some markers may be better for clinical management, some for etiologic studies, and some for both.
Survivor biomarkers: old and new
A classical survivor biomarker is the tumor, node, metastasis (TNM)3 system, in which T reflects tumor size or depth; N, lymph node spread; and M, metastasis (presence or absence) (1). Disease stages (I–IV) are derived from various combinations of T, N, and M. The TNM system has several advantages: It 1) predicts survival (I is longest, IV shortest); 2) determines choice of initial treatment; 3) provides a rationale for patient stratification in clinical trials; 4) facilitates accurate communications among health care providers; and 5) leads to more uniform reporting of outcomes. The TNM system is relevant to both clinical management and etiologic research: 1) treatment and prognosis depend on stage; and 2) understanding causal factors involved in adverse outcomes requires stage information.
Advances in basic and clinical science have begun to take us beyond the TNM system, revealing both the complexity and opportunity associated with survivor biomarkers. [For an excellent review, see Ludwig and Weinstein (2)]. For example, new biomarkers, including gene expression patterns, show that tumor subsets behave differently one from another. Some "targeted" therapeutic agents may be effective only if certain genetic markers are mutated or adequately expressed. Estrogen receptor (ER) or HER2/NEU (also known as ErbB2) status has implications for both prognosis and therapy and may even turn out to be valuable in etiologic studies (certain exposures, for example, may be causal only for estrogen receptor-positive tumors) (3). Increasing consideration is being given to combining the classical TNM system with newly emerging survivor biomarkers.
Survivor biomarkers: examples
The value of survivor biomarkers is reflected in the following examples.
Anatomic localization. RNA high-throughput and microarray techniques have been used to differentiate ovarian from colonic cancers and head and neck from lung malignancies. This type of marker has both clinical and etiologic relevance.
Grade. Computer-aided diagnostic systems assist in reading of Pap smears in cervical neoplasia.
Stage.
Imaging agents delineate malignant disease or its metabolism, important in discriminating local from metastatic disease. The TNM classification plus serum 
-protein, ß-human chorionic (HCG) gonadotropin, and lactate dehydrogenase (LDH) have been used to determine testicular cancer grade.
Prognosis and treatment selection. For breast cancer, hormone receptor positivity informs decisions on treatment with tamoxifen or aromatase inhibitors; HER2/NEU positivity can determine use of herceptin therapy. Idiosyncratic drug toxicity may be avoided with information on the presence of thiopurine S-methyltransferase (TPMT) gene mutations, which have been linked to white blood cell suppression in leukemia patient (2) .
A number of molecular markers have promise in the (possibly near) future. These include DNA markers (specific gene mutations such as KRAS, p53, and APC); epigenetic markers, including hypermethylated DNA found in saliva, sputum, or serum; RNA markers, especially multigene molecular patterns derived from laser-capture microdissection, which can be used to predict prognosis (4) as well as drug response (5); and protein markers, including protein marker patterns (so-called fingerprints).
It is noteworthy that the number of biomarkers approved is much lower than the number available. In fact, most FDA-approved markers are not used in standard clinical practice.
Examples of biomarkers useful in studies of factors that determine survival include those of dietary intake; BMI, and other anthropometric indices; intermediates, including hormone levels and inflammatory factors; and prostate-specific antigen (PSA).
Functions of survivor biomarkers in cancer prognosis research
How, then, do we go about evaluating these survivor biomarkers? Although I focus here on etiologically relevant survivor biomarkers, the principles presented are applicable to the clinical context as well.
In etiologic research, survivor biomarkers serve 3 potential roles: 1) enhancing the biological plausibility of exposure-survival relations; 2) increasing strength of exposure-survival associations (what can be called "relative risk sharpening"); and 3) serving as surrogate endpoints. I discuss each of these in turn.
Biological plausibility. Biomarkers can reinforce the biological plausibility of exposure-survival relations by clarifying causal pathways. Examples include estrogens as a pathway through which weight control and physical activity operate and inflammatory factors as the intermediate process through which dietary factors act.
Relative risk sharpening. Intake biomarkers (of diet or supplement use) can enhance relative risks that are otherwise seriously diluted by questionnaire-based data. A prominent example is serum selenium levels, a clearer reflection of intake than questionnaire-based measures that are vulnerable to the substantial geographic variability in the selenium composition of grains and other foods. Here, we see an inverse association between serum selenium levels and risk of colorectal adenoma recurrence (6) (see Fig. 1).
|
Surrogate endpoints. Studies with surrogate endpoints can be smaller, faster, and cheaper than those with recurrence or death as endpoints. This holds for intervention studies (clinical trials) as well as observational epidemiologic studies.
Validation of survivor biomarkers
What constitutes the "validation" of survivor biomarkers in etiologic research? For the biological plausibility and relative risk-sharpening functions, validity requires that the biomarker must be truly on (or very close to) the causal pathway(s) to cancer. For surrogacy, the requirement is tougher: the study of exposure or treatment against the putative surrogate marker must give the right answer to the effect of the exposure or treatment on recurrence or death.
Three conditions are needed for survivor biomarker validity. First, the marker must be associated with survival. Standard epidemiologic concepts, such as relative risk and attributable proportion, can be used to gauge the relation of the marker to survival (see Table 1). An excellent example of studies evaluating biomarker validity is the recent prospective cohort study of BMI, blood estrogen levels, and breast cancer (9). This was a study of incident breast cancer, but, again, the principles reflected in this example hold for survival as well. As the accompanying data show, the marker (estrogen levels) is clearly related to breast cancer (see Table 2).
|
|
|
|
|
|
This concept is illustrated below in an idealized pathway diagram of cellular events potentially involved in colorectal cancer survival (see Fig. 2). The idea here is that normal colorectal mucosa, when exposed to a chemopreventive agent or treatment (E), leads to reduced proliferation and enhanced survival. There is, however, an alternative pathway working through other cellular events, such as reduced apoptosis or diminished cellular adhesion factors. These 2 pathways offset one another. If one examines only a marker of proliferation, the offsetting events in the alternative pathway may be neglected. Thus, reduced proliferation may not be necessary for colorectal cancer survival because of the existence of this alternative pathway bypassing proliferation. As a consequence, the effect of an intervention agent on the alternative pathway may counterbalance the effect on the hyperproliferation pathway. Proliferation markers may give the wrong answer about an intervention agent's effect on colorectal cancer survival: 1) an agent that reduces proliferation but at the same time reduces apoptosis would have no effect on survival; or 2) an agent that has no effect on proliferation but increases apoptosis could increase colorectal cancer survival. Either way, the proliferation marker would give the wrong answer with respect to survival.
|
In sum, it is the totality of causal connections that is key. In that regard, a high attributable proportion (derived as above) and mediation of the exposure-survival relation by the marker do provide strong support of the validity of the putative surrogate marker.
Additional issues affecting biomarker validity
A surrogate endpoint that is valid for 1 exposure or intervention in relation to survival is not necessarily valid for a second exposure or intervention. That is because, once again, an alternative pathway to survival may exist, as the following idealized pathway diagram illustrates (see Fig. 3).
|
Survival biomarkers will be measured with some error. This measurement error will tend to attenuate the associations between exposure/treatment and biomarker and between the marker and survival. Moreover, with respect to the mediation criterion, measurement error can lead to underestimation of the extent to which the surrogate mediates the effect of exposure/treatment on survival.
There is a certain irony to surrogate endpoint validation. That is, the large, long, and costly studies needed for validation are precisely the studies surrogate endpoints were designed to replace. There is also a "no free lunch" law that applies to surrogate endpoints in research on factors influencing survival: inferential certainty is directly associated with study costs. (Or, you get what you pay for.)
Survivor biomarkers in etiologic research may be valuable in establishing causality (or lack thereof) for potentially important exposures or treatments (dietary factors, drugs, and so on). As surrogates for recurrence or death, these biomarkers may be particularly valuable in Phase II studies, those evaluating a biologic effect. The savings resulting from use of surrogate endpoints, however, comes at the cost of inferential certainty. In conjunction with other studies (those with biomarker endpoints plus cohort studies of risk factors for survival), survivor biomarkers may enhance the probability of being right.
In conclusion, words from Ludwig and Weinstein are apt: "New high-throughput ‘omic’ technologies in postgenomic biology have yielded many potential biomarkers and biomarker patterns, some of which may prove useful for staging and grading cancers [and etiologic research into survival]...."
"The potential is enormous. Few markers, however, have so far been integrated into clinical practice (or etiologic research). Metaphorically speaking, the water is everywhere, but little is yet ready to drink" (2).
|
FOOTNOTES |
|---|
2 Author Disclosure: No relationships to disclose.
3 Abbreviations used: ER, estrogen receptor; HER2/NEU, HER2/neu (human epidermal growth factor receptor 2); HPV, human papillomavirus; PSA, prostate-specific antigen; RR, relative risk; TNM, tumor, node, metastasis
|
LITERATURE CITED |
|---|
|
TOPABSTRACTLITERATURE CITED |
|---|
1. Greene FL, Page DL, Fleming ID, Frit A. AJCC Cancer Staging Manual, 6th ed. New York: Springer; 2002.
2. Ludwig JA, Weinstein JN. Biomarkers in cancer staging, prognosis and treatment selection. Nat Rev Cancer. 2005;5:845–56.[Medline]
3. Althuis MD, Fergenbaum JH, Garcia-Closas M, Brinton LA, Madigan MP, Sherman ME. Etiology of hormone receptor-defined breast cancer: a systematic review of the literature. Cancer Epidemiol Biomark Prev. 2004;13:1558–68.
4. van de Vijver MJ, He, YD, van't Veer LJ, Dai H, Hart AA, Voskuil DW, Schreiber GJ, Peterse JL, Roberts C, et al. A gene-expression signature as a predictor of survival in breast cancer. N Engl J Med. 2002;347:1999–2009.
5. Olaussen KA, Dunant A, Fouret P, Brambilla E, Andre F, Haddad V, Taranchon E, Filipits M, Pirker R, et al. DNA repair by ERCC1 in non-small-cell lung cancer and cisplatin-based adjuvant chemotherapy. N Engl J Med. 2006;355:983–91.
6. Jacobs ET, Jiang R, Alberts DS, Greenberg ER, Gunter EW, Karagas MR, Lanza E, Ratnasinghe L, Reid ME, et al. Selenium and colorectal adenoma: results of a pooled analysis. J Natl Cancer Inst. 2004;96:1669–75.
7. Chen J, Stampfer MJ, Hough HL, Garcia-Closas M, Willett WC, Hennekens CH, Kelsey KT, Hunter DJ. A prospective study of N-acetyltransferase genotype, red meat intake, and risk of colorectal cancer. Cancer Res. 1998;58:3307–11.
8. Wacholder S, Chanock S, Garcia-Closas M, El Ghormli L, Rothman N. Assessing the probability that a positive report is false: an approach for molecular epidemiology studies. J Natl Cancer Inst. 2004;96:434–42.
9. Key T, Appleby P, Barnes I, Reeves G. Endogenous sex hormones and breast cancer in postmenopausal women: reanalysis of nine prospective studies. J Natl Cancer Inst. 2002;94:606–16.
10. Key TJ, Appleby PN, Reeves GK, Roddam A, Dorgan JF, Longcope C, Stanczyk FZ, Stephenson HE Jr, Falk RT, et al. Body mass index, serum sex hormones, and breast cancer risk in postmenopausal women. J Natl Cancer Inst. 2003;95:1218–26.
11. Schatzkin A, Gail M. The promise and peril of surrogate end points in cancer research. Nat Rev Cancer. 2002;2:19–27.[Medline]
12. Schiffman MH, Bauer HM, Hoover RN Glass AG, Cadell DM, Rush BB, Scott DR, Sherman ME, Kurman RJ et al. Epidemiologic evidence showing that human papillomavirus infection causes most cervical intraepithelial neoplasia. J Natl Cancer Inst. 1993;85:958–64.
13. Fleming TR, DeMets DL. Surrogate end points in clinical trials: are we being misled? Ann Intern Med. 1996;125:605–13.
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
