Biomarkers in Cancer Staging, Prognosis and Treatment Selection

Joseph A. Ludwig; John N. Weinstein

In This Article

Biomarker Use at Diagnosis


Classification of a malignancy by tissue of origin is the first step towards predicting survival and choosing therapy. Because a tumour's anatomical location usually indicates its tissue of origin, molecular markers are rarely required. Histological examination generally confirms the diagnosis and identifies the tumour subtype. However, new molecular markers might sometimes be helpful in the differential diagnosis. For example, as shown schematically in Fig. 2, we recently used a combination of high-throughput RNA, protein and tissue microarray technologies to identify markers potentially useful for distinguishing colon and ovarian abdominal carcinomas from an unknown primary location.[6] Similarly, biomarkers have been reported to distinguish primary head and neck squamous cell carcinoma (HNSCC) from metastatic lung squamous cell carcinoma (SCC),[7] to determine the site of origin for HNSCC of unknown primary location[8] and to track genetic mutations that occur with the progression of that tumour.[9]

Table 1.


US Food and Drug Administration-Approved Cancer Biomarkers


Each anatomical site has its own histological grading system, designed to classify malignancies by degree of differentiation. Low-grade, well-differentiated tumours are usually less aggressive and more favourable in prognosis than high-grade tumours, which tend to grow faster and metastasize earlier. However, tumour grade is included in formal TNM staging only when intimately linked to prognosis, as it is for soft-tissue sarcomas, prostate cancer and primary brain malignancies. Assignment of grade is inherently subjective and dependent on the skill and experience of the reviewing pathologist, but several reports indicate that biomarker patterns can correctly score tumours according to their pathologist-assigned grades.[10] COMPUTER-AIDED DIAGNOSTIC SYSTEMS (CAD systems) have been approved by the FDA for preliminary grading of cervical smears (that is, Pap smears)[11] and for assisted interpretation of radiological images such as screening mammograms,[12] computerized tomography (CT) scans[13] and standard X-ray films.[14] CADs are generally designed to make routine distinctions, giving the pathologist time to focus on difficult diagnostic problems. The acceptance of CADs has been accelerated by the fact that there had previously been extensive and rigorous standardization and quality control of the underlying imaging technologies (for example, mammograms and chest X-rays). The absence of analogous standardization of biomarker platforms is an important practical problem.

If the problems can be overcome, however, the addition of either individual or PATTERN-BASED BIOMARKERS in the assessment of histological grade could increase the utility of grading for predicting response to therapy. That would be a natural extension of current practice as pathologists already have considerable experience using at least a few markers (for example, ER, the progesterone receptor (PR) and HER2/NEU) in related contexts.


The AJCC, in collaboration with the TNM Committee of the International Union Against Cancer (UICC), has defined staging criteria for most anatomical sites.[15] T, N and M are determined separately and then grouped, usually to classify the cancer into one of four main stages (stages I-IV) and subdivisions thereof. Breast cancer staging, for example, distils 30 possible TNM combinations into 5 main prognostic stages.[15] Clinical staging, which is primarily used to guide initial therapy, integrates information from physical examination with data such as those from standard X-ray, CT, MRI, PET, endoscopic examination, biopsy, and surgical exploration. Pathological staging on the basis of surgical specimens, if acquired, complements clinical staging with a precise determination of the extent of disease and additional histological information.

Increasingly, imaging agents targeted at biomarkers are being used for anatomical localization. The most common are radioisotopes, detected by standard nuclear medicine imaging, by SINGLE-PHOTON EMISSION COMPUTED TOMOGRAPHY (SPECT) or by PET. Also under study are fluorescent molecules, which are detected by optical imaging, and paramagnetic particles for enhancing MRI. The target can be any marker that delineates the cancer or its metabolism. Some tumours (for example, carcinoid, phaeochromocytoma, and cancers of the prostate, thyroid and colon) can be targeted by specific radiolabelled ligands. Carcinoid tumours, for example, are often localized using a radiolabelled analogue of octreotide (111-indium pentetreotide), which avidly binds to the somatostatin receptor, a protein commonly overexpressed in those tumours. Nuclear medicine-based imaging modalities are also clinically useful for evaluating tumour-related phenomena including angiogenesis,[16] apoptosis,[17,18] proliferation,[19] metabolism,[20,21] hypoxia[22] and drug resistance (such as P-glycoprotein function).[23] Molecularly targeted functional imaging has enormous potential for staging, as it does for other aspects of cancer diagnosis and management, and it might also be easier than serum biomarkers to integrate into clinical TNM staging, given its anatomical basis.

For most solid tumours, the primary purpose of anatomy-based staging is to discern the probability of localized, as opposed to metastatic, disease. That crucial distinction, in considerable part, predicts survival and guides the choice of initial therapy. Anatomy-based staging, however, provides only part of the answer. A more precise picture can often be obtained by incorporating tumour grade and histological subtype. The role that biomarkers can play is exemplified by HER2/NEU, which is associated with an aggressive phenotype, decreased patient survival[24,25] and response to trastuzumab.

Acknowledging the potential importance of serum-derived biomarkers for staging, the AJCC incorporated the first such markers — serum α-fetoprotein (AFP), human chorionic gonadotropin-β (β-HCG, also known as CGB) and lactate dehydrogenase (LDH) for testicular cancer — into the TNM system. That necessarily cautious shift in the staging guidelines reflects the increased scientific evidence, at least for several cancer types, of more accurate prognosis with the addition of factors independent of anatomy and histological grade. However, as stated previously, most cancers are still exclusively staged by anatomic criteria, with a few exceptions (for example, sarcomas and malignancies of the thyroid, prostate, brain and testicle).

The AJCC sometimes informally recommends supplementation of TNM staging with information about tumour grade, histological subtype or relevant immunohistochemical (IHC) markers when they have prognostic or therapeutic value. Suggested supplementary parameters for breast cancer, for example, include those with proven value in predicting response to therapy (ER, PR and HER2/NEU receptor status). IHC or reverse transcription-PCR (RT-PCR) evaluation of sentinel lymph nodes is occasionally performed in clinical trials when microscopic examination is negative and, if obtained, is included as supplemental information in AJCC breast cancer staging. However, most supplementary markers have not been clinically validated, and the significance of lymph nodes that are negative by standard pathological staining with haematoxylin and eosin but positive by IHC or PCR remains unclear. The TNM system classifies nodes with cancer cell clumps less than 0.2 mm in diameter as node-negative, even when RT-PCR detects tumour cells. Such small clumps of cancer cells usually lack markers of proliferation and rarely induce a stromal reaction that indicates tumour implantation or growth.[15,26]

Any updating of a system like TNM can have advantages but also imposes a price. Ideally, one would like to incorporate the latest supportable medical science, including new biomarkers. However, the relatively static, stable character of the system enhances its utility for stratifying patients in clinical trials, for communication among physicians and for standardizing the classification of tumours among institutions and nations. If the staging criteria were changing more often, it would be difficult, for example, to determine whether earlier clinical research was pertinent to current or future patients who are, or will be, staged differently. Furthermore, information supplemental to TNM is often incomplete or sporadically noted in medical records because it is not formally part of the staging process. Adoption of a formal 'augmented' staging system that included both a relatively static TNM component (updated occasionally) and a dynamic supplemental component (revised at the pace of scientific discovery) could perhaps resolve that paradox and allow new markers to be evaluated formally without undermining the value of anatomical staging.

Prognosis and Treatment Selection

Tumour classification, stage and sometimes grade are used to assess prognosis. However, as noted above, there would be a cost if formal cancer staging incorporated every other parameter able to improve prognosis. Furthermore, stratification in clinical trials using all possible TNM combinations would be impractical, given limitations in patient participation and resources. Addition of markers could similarly fragment the staging process, thereby limiting its utility. More information is generally better than less information, but the advantages must be weighed against those of a stable classification with relatively few categories.

Biomarker expression often supplants or complements tumour classification, stage and grade when biologically targeted therapeutics are under consideration. Prominent examples include CD20 positivity for treatment of lymphomas with rituximab, HER2/NEU positivity for treatment of breast cancer with trastuzumab,[27] BCR-ABL translocation for treatment of chronic myelogenous leukaemia (CML) with imatinib, and KIT or platelet-derived growth factor receptor-α (PDGFRA) positivity for treatment of gastrointestinal stromal tumours (GIST) with imatinib.[28] As previously discussed, ER positivity or PR positivity is a prerequisite for treatment with tamoxifen or aromatase inhibitors. Similarly, somatic mutations in the tyrosine-kinase domain of the epidermal growth factor receptor (EGFR) have recently been shown to predict a greater efficacy of gefitinib in patients with non-small-cell lung cancer (NSCLC).[29,30] Some of those markers are FDA-approved and in widespread clinical use (see Table 1 ); others have been assessed only in the research setting. For example, ER, PR and HER2/NEU status are routinely determined for breast cancer, whereas EGFR mutations are usually assessed only in clinical trials. Outside of such trials, patients with NSCLC are often given EGFR-antagonists, such as gefitinib, as salvage therapy on an empirical basis without marker studies, especially if they are more likely to have the mutation (that is, patients who are female, never-smokers, diagnosed with adenocarcinoma, or Asian).[31]

Both prognosis and prediction of response are necessary for the selection of neoadjuvant or adjuvant chemotherapy. Tissue classification, TNM staging, molecular biomarkers, grade and other factors might be used in combination for that purpose. The combinations of variables might not be easy to analyse manually, but computer DECISION SUPPORT SYSTEMS (DSS) can make the assessments automatically. For example, Adjuvant Online (see Online links box), a DSS used for breast cancer, estimates 10-year cancer recurrence and survival for women, taking into account their predicted response to adjuvant chemotherapy.[32] Markers can also be used to avoid idiosyncratic drug toxicity such as the sustained, life-threatening leukocyte suppression seen when mercaptopurine is given to leukaemia patients with homozygous mutations of the thiopurine methyl-transferase (TPMT) gene.[33,34]

Biomarkers traditionally used for risk assessment and screening are also available to enhance cancer staging, refine prognosis and estimate response to biological therapy, as summarized in Box 1 and Fig. 3. An important point: characteristics that suit a molecular marker for one application might not do so for another. For example, a marker to be used in screening the general population must have an extremely high specificity to minimize false positives that necessitate costly or invasive follow-up studies and scare patients and their families needlessly. The same marker need not be so specific if used for high-risk populations and can be even less so once a cancer has been detected. The arguments about use of PSA for screening continue, but its value in monitoring diagnosed prostate cancer or its treatment would be hard to dispute.

Figure 1.

Numbers of publications on biomarkers and FDA approval of biomarkers.
Despite the increasing rates of publications on biomarkers, the number of US Food and Drug Administration (FDA)-approved plasma-protein tests is decreasing. Triangles and the associated trend line (green) represent the number of FDA-approved plasma-protein markers per year (data taken from Ref. [5]). Red squares and circles indicate publications under the Medline medical subject heading 'biomarker' and text word 'biomarker', respectively.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: