Tumor markers provide a minimally invasive, cost-effective source of data valuable for monitoring disease course, determining prognosis, and aiding in treatment planning. An understanding of the individual test characteristics and limitations, however, is important for optimal use and accurate interpretation of results. A variety of types of markers are available in the routine clinical laboratory including enzymes, oncofetal antigens, oncogene products, monoclonal antibody-defined epitopes, and hormones. The clinical applications and limitations of some that are currently in widespread use will be discussed.
Cancer remains the second leading cause of death in the United States, behind heart disease, with an estimated 1,437,180 new cases and 565,650 deaths in 2008 alone. It is expected that malignancies of prostate and breast origin will be the most common causes of new diagnoses in men and women, respectively, with tumors of the lung and bronchus the leading cause of cancer-related death in both genders.1 Annually, costs associated with diagnosis and treatment together constitute a considerable proportion of the total national economic burden attributable to cancer care (estimated at $72.1 billion of $190 billion in 2004).2,3 Relatively noninvasive laboratory testing that is simple and cost-effective to perform, yet facilitates earlier diagnoses and improved treatment outcomes, is the optimal role of any tumor marker. The number of potentially viable markers is rapidly expanding and promises further improvement in sensitivity and specificity. A variety of markers are presently available in the routine clinical laboratory (Table 1). Clinical applications of some that are currently in widespread use will be discussed.
Urinary catecholamines, vanillylmandelic acid, homovanillic acid, metanephrines, chromogranin A
FSH, LH, TSH, prolactin,
Testicular germ cell
hCG, AFP, LDH
Ovarian germ cell
Gestational trohpoblastic disease
β2-microglobulin, serum free light chains, LDH, M-protein (serum or urine)
The definition of a tumor marker is broad. It consists of any product of either the tumor itself or the host in reaction to the tumor’s presence, that distinguishes malignant tissues from benign and is measurable in body fluids or tissues.4 Ideally, they reflect the body’s tumor burden, and subsequently increase with progressive or recurrent disease, decrease with response to treatment, and normalize with remission. Historically, sources of markers have been equally as broad, including enzymes and isoenzymes, oncofetal antigens, oncogene products, monoclonal antibody-defined epitopes, abnormal hormone concentrations, and genomic alterations.4 Clinical applications include screening in asymptomatic individuals, confirming a suspected diagnosis, assisting in tumor classification and staging, estimating prognosis, monitoring treatment response, surveillance for residual disease, and early detection of recurrent disease. Due to the relatively low prevalence of individual cancer types, use of these markers for screening asymptomatic individuals, to date, has generally not been an effective strategy.5 A notable exception, perhaps, is evaluating prostate-specific antigen (PSA) concentrations in conjunction with a physical exam.
Factors Affecting Clinical Utility
Several factors influence the clinical utility of any potential marker. Those attributable to the type of tumor include its prevalence among the population under investigation in addition to the availability of an effective treatment regimen.5 The greatest benefit is obtained when there is both a high prevalence of the disease in the population and an efficacious treatment exists. Factors attributable to the test include its sensitivity (ability to detect individuals with the disease) and specificity (ability to discern individuals with the disease from those that are either normal or have some other condition). These can be calculated using the following formulas:6,7
A receiver operator characteristic (ROC) analysis demonstrating the tradeoff between specificity and sensitivity at various concentrations (graphical plot of 1-specificity on the X axis and sensitivity on the Y axis) is useful in determining an appropriate cutoff value.7 Values that allow maximization of sensitivity, specificity, or the best tradeoff between the two can be selected depending on the anticipated clinical use of the test. Optimum sensitivity is desirable for screening applications, while maximizing specificity is crucial for confirmatory purposes.7
There are some well-described limitations, however, to the use of tumor markers. Most importantly, the majority of those currently available can be measured to some extent in individuals who are healthy, have benign conditions, or have malignancies other than the one of interest. Similarly, most of these do not reliably detect early stage disease and may even fail to detect recurrent or advanced disease in some patients.5 As such, results should always be interpreted in conjunction with other clinical findings. Evaluating serial values requires an understanding of the test’s in vivo performance including normal biologic variability and expected biologic half-life in order to accurately interpret a change in levels.6 As illustrated in Figure 1, markers frequently show inter-individual variability in their ability to reflect a patient’s disease status. Care must also be taken to avoid misinterpreting transitory increases (“tumor marker spike”) that can be seen in some patients who are actually responding to treatment. Analytical factors such as heterophile or other antibody or substance interference can adversely affect the methodology used and must be ruled out when results are clinically inconsistent with other findings. Additionally, due to variability between laboratories, instruments, and reagent kits, serial results should be compared only when obtained from the same laboratory using the same methodology. Nonetheless, a marker profile seen at disease recurrence may differ from that determined at the time of diagnosis due to additional genetic alterations accumulated by the malignant cells.8 Because of this, it has been suggested that markers still be measured during follow-up evaluations, regardless of the levels at baseline, and early detection of recurrence should not depend solely on changes in marker levels, even if they were initially abnormal at baseline.8
Four serial measurements of 3 tumor markers obtained from a patient with metastatic breast cancer illustrate the variability of marker response to disease status commonly seen in individual patients. While levels of 2 markers remained relatively unchanged (CA 27.29 and CEA), values of the third (circulating tumor cells [CTC]) initially declined with treatment but later increased with the development of brain metastases. Results were normalized by dividing by the upper limit of normal.
CA 15-3 and CA 27.29
Mucins are large glycoproteins normally found on a variety of epithelial cell types, including breast, which are divided into 7 structurally identifiable families (MUC1 to MUC7).9,10 The MUC1 gene product is a polymorphic transmembrane glycoprotein frequently overexpressed on malignant glandular cell surfaces, resulting in increased levels shed into the blood of some patients with carcinomas.11 Both CA 15-3 and CA 27.29 are monoclonal antibody-defined markers. Antibodies designated DF3 and 115D8 bind CA 15-3, while CA 27.29 is the molecule that competes with antibody CA 27.29 for binding to an antigen obtained from a breast cancer cell line (ZR-75-1). Good correlation has generally been found between these two markers.9,12 Both are used in conjunction with other assessments to monitor treatment response in patients with meta-static breast cancer as well as those who have been previously treated for stage II or stage III disease. These markers have shown improved sensitivity and specificity over the use of carcinoembryonic antigen (CEA) in the assessment of these patients.13,14
Carcinoembryonic Antigen (CEA)
Carcinoembryonic antigen is one of the earliest described carcinoembryoinc proteins. These are proteins normally abundant during fetal development, but not detectable (or in very low concentrations) in the healthy adult, that reappear in the circulation of patients with some malignancies. Carcinoembryonic antigen is a nonspecific marker, and increased values can be seen with a variety of tumor types including those of breast, gastrointestinal (GI), and lung origin. The normal range for smokers is slightly higher than that for nonsmokers. Carcinoembryonic antigen concentrations can be most informative if they are increased while other markers are still within normal levels.
Human Epidermal Growth Factor Receptor 2 (HER-2)
The HER-2 gene product is a transmembrane protein normally involved in cell growth and differentiation through its interaction with circulating growth factors. Identified as an oncogene, amplification results in protein overexpression that supports rapid cellular proliferation. Seen in approximately 25% of breast cancer patients, overexpression (HER-2-positive tumors) is associated with a more aggressive clinical course and worse overall prognosis.15 HER-2 status has also been shown to be somewhat predictive of responsiveness to various chemo-therapeutic agents.16 As such, test results can greatly impact treatment planning by allowing selection of drug regimens that maximize expected benefit compared with the known toxicities.16 In particular, targeted therapy with trastuzumab (Herceptin) can significantly improve outcome in these patients; consequently, the requirement for accurate evaluation of tumor HER-2 status by the laboratory has been greatly emphasized.17 Given the significant prognostic and therapeutic clinical implications, it has been recommended that assessments for HER-2 overexpression should be obtained routinely on patients presenting with invasive breast cancer.16 A variety of assays, which have not been standardized, have been in use, and evidence suggests that up to 20% of obtained results may not be accurate, potentially negatively affecting patient outcome.16
Estrogen/Progesterone Receptors (ER/PR)
These steroid receptors can play a role in breast carcinogenesis, and similar to HER-2, assessment of ER/PR status is crucial for optimal treatment planning. Approximately one-third of breast cancers show a response to antiestrogen agents such as tamoxifen (Nolvadex), a selective estrogen receptor modulator (SERM). This response is predicted by receptor status.18 Substantially decreased recurrence rates and disease-related mortality have been demonstrated with tamoxifen for ER+ cancers, which are not seen when the tumors are ER–.18,19 Receptor status also has significant prognostic implications, with best to poorest survival ranging from ER+/PR+ to ER−/PR–.20
Prostate-Specific Antigen (PSA)
Prostate-specific antigen is a serine proteinase secreted by prostatic and periurethral gland epithelium which promotes liquefaction of semen.21 The majority circulates irreversibly complexed to endogenous protease inhibitors, although a variable fraction remains free (unbound).23,24 The American Cancer Society and the American Urologic Association recommend that annual screening of PSA levels combined with a digital rectal exam (DRE) be offered to men 50 years of age or older who have at least a 10-year life expectancy, or by 45 years of age to men at increased risk of prostate cancer (African-American or a family history of the disease in a first-degree relative at an early age). PSA, however, is neither prostate- nor cancer-specific (despite the name) as minute amounts are also identifiable in some other male and female tissues22 and increased concentrations can also be found in benign prostatic hyperplasia (BPH), prostatitis, and other nonneoplastic conditions. In circulation, the half-life of measured total PSA is approximately 3 days, and, consequently, several weeks may be required before levels decline to a nadir after treatment.22,25 Although the usual normal range for total PSA is 0.0 to 4.0 ng/mL, approximately 20% of men with PSA levels between 2.6 to 4.0 ng/mL and a normal DRE may actually have prostate cancer.26 Similarly, levels obtained from patients with BPH and those with cancer more commonly found at concentrations greater than 10.0 ng/mL. Consequently, a clinically challenging “grey zone” exists between 4.0 to 10.0 ng/mL in which cancer is identified in only one-quarter of patients tested.27 A number of approaches have been taken in an attempt to improve the sensitivity and specificity of PSA including age-adjusted normal ranges,28 calculating PSA density (total PSA divided by prostate volume),23,29 determining PSA velocity (change in levels over time),23,30 and measuring PSA forms (free and complexed).23,27,31
Although relatively nonspecific as previously mentioned, CEA is the primary serum tumor marker used to monitor treatment response and to aid in detecting recurrence in patients with colorectal cancer. Increased preoperative concentrations in conjunction with other clinical factors, have been associated with a worse prognosis.32 As a marker of overall tumor burden, levels are expected to decline with successful treatment. Subsequent rising values, particularly when progressive, are typically indicative of relapsing disease and may precede other clinical indicators by several months.5
A monoclonal antibody-defined antigen, CA 19-9 is a sialylated Lewis (Lea) blood group antigen recognized by antibody 1116 NS 19-9.33 Although derived using a colorectal carcinoma cell line, substantially increased serum concentrations are found in the majority of patients presenting with locally advanced or metastatic pancreatic cancer, and are less frequently seen with other malignancies of the gastrointestinal tract or with pancreatitis and other benign conditions.5,33 Individuals with the Lea-b- phenotype, however, do not express CA 19-9.34 In addition to disease stage, both pretreatment and posttreatment concentrations can provide prognostic information, with higher values associated with a shorter survival.33 Similarly, evidence indicates that the degree of marker decline from pretreatment levels in patients receiving systemic therapy do correlate with both overall survival and time to treatment failure. Additional studies are needed, however, to better clarify the utility of CA 19-9 measurements in the evaluation of potentially new treatment modalities.33,35
CA 125 is a monoclonal antibody-defined antigen that can be found in increased concentrations in the serum of patients with epithelial ovarian cancers. Increased values can also be found in patients with other epithelial cancers as well as in benign conditions, particularly if there is involvement of a serosal lining. Located on a heterogeneous high-molecular-weight glycoprotein, it is recognized by a murine antibody OC125 derived from an ovarian cancer cell line (OVCA 433).36,37 Used in the assessment of treatment response, elevated levels after therapy are highly suggestive of residual disease, even if not grossly apparent, and increasing concentrations suggest relapse.38 Residual tumor can still be present, however, despite levels falling within the normal range.36 Levels of nadir concentrations, even when within normal limits, may be predictive of both time to post-treatment disease progression and overall survival.39
The markers discussed above have all been granted Food and Drug Administration (FDA) approval to be used as tumor markers within specified clinical indications. These also have assigned CPT codes, such as 86300, 86301, 86304—immunoassay for tumor antigen, quantitative, CA 15-3 (CA 27.29), CA 19-9, CA 125, and 84153—prostate-specific antigen; total. There are numerous additional analytes with long histories of established use including hormones, enzymes, immunoglobulins, and metabolic byproducts, normally measurable in body fluids that are found in abnormal concentrations when reflecting the volume of a tumor mass. Although typically quite useful for assisting in monitoring disease course, tumor classification, and even disease staging, as is seen with lactate dehydrogenase (LDH), α-fetoprotein (AFP), human chorionic gonadotropin (hCG), and others, aberrant levels of the same marker can be characteristic of tumors originating from multiple organs and can only be interpreted in light of the clinical setting. Despite the information provided, some have not been FDA approved for specific use as tumor markers.
Tumor markers provide a minimally invasive, cost effective, and usually easily obtainable source of data that has proven to be valuable in monitoring disease course, determining prognosis, and aiding in treatment planning. Knowledge of the in vivo performance of the marker used, however, remains critically important for accurate interpretation of the information provided. Published recommendations regarding the use of individual markers frequently vary, and large scale studies confirming the criteria for their application would help to improve consistency in their use and interpretation.13 With the exception of PSA in conjunction with a DRE, none have been proven to be suitable for screening purposes. The future is promising for new markers, the discovery of which is greatly enhanced by the availability of molecular-based techniques. Genomic analysis, gene expression profiling, investigation of epigenetic changes, proteomic-focused studies, and isolation/analysis of circulating tumor cells all offer new opportunities for biomarker discovery.40 Optimally, markers suitable for widespread screening, early detection, and individualized selection of anticancer therapy may eventually augment our current armamentarium.
After reading this paper, readers should be able to describe the expanding range of viable tumor markers, tumor markers that are presently available in the routine clinical laboratory, and clinical applications of markers that are currently in widespread use.
Chemistry 20902 questions and corresponding answer form are located after this CE Update article on page 104.
. Use of the percentage of free prostate-specific antigen to enhance differentiation of prostate cancer from benign prostatic disease: A prospective multicenter clinical trial. JAMA. 1998;279:1542–1547.