Abeloff's Clinical Oncology, 4th Edition

Part I – Science of Clinical Oncology

Section C – Diagnosing Cancer: Pathology and Laboratory Medicine

Chapter 20 – Biomarkers for Cancer Diagnostics

Lori J. Sokoll,Daniel W. Chan

SUMMARY OF KEY POINTS

Serum Tumor Markers

Key Methods

  

   

Radioimmunoassay and, more recently, enzyme and other immunoassays are cornerstones of cancer diagnostic methods.

  

   

Monoclonal antibodies have been produced against a wide variety of secreted antigens.

Applications

Screening and Early Detection

  

   

In most tumor systems, biomarkers do not have sufficient sensitivity and specificity for routine population screening, although combining them with other approaches can improve utility.

  

   

In some high-prevalence situations tumor marker screening is appropriate (e.g., prostate-specific antigen [PSA] for prostate cancer).

Diagnosis

  

   

Levels of particular markers are useful adjuncts to diagnosis in many tumor systems, including ovarian cancer, germ cell tumors, or neuroendocrine tumors.

  

   

In some tumor systems, serum markers contribute to staging information and help direct therapy.

Monitoring

  

   

Most serum tumor markers are useful to monitor treatment or progression of cancer.

  

   

A decrease in tumor marker levels can be used as a means to assess the success of surgery or chemotherapy.

  

   

An increase in a tumor marker can be a harbinger of relapse or disease progression.

INTRODUCTION

Tumor markers, also called cancer markers, biomarkers, and, in some instances, cancer-associated antigens, are substances present in or produced by a tumor itself or produced by the host in response to a tumor that can be measured in the blood or secretions and used to differentiate a tumor from normal tissue or to determine the presence of a tumor. Such substances can be found in cells, tissues, or body fluids. They can be measured qualitatively or quantitatively by chemical, immunologic, or molecular biologic methods to identify the presence of a cancer. A wide spectrum of molecules may be classified as tumor markers, including enzymes, hormones, oncofetal antigens, carbohydrate markers, blood group antigens, proteins, receptors, and genes or gene products.[1] Although, strictly speaking, tumor markers can be recovered from any tissue or fluid, this chapter is devoted largely to those that are secreted or shed and detectable in serum.

The first recognized tumor marker was the Bence-Jones protein, discovered in 1847 by precipitation of a protein in acidified boiled urine. This protein, the monoclonal light chain of immunoglobulins secreted by tumor plasma cells, still is used in the diagnosis of multiple myeloma. The first half of the twentieth century included the discovery of hormones, for example, human chorionic gonadotropin (hCG); enzymes, for example, acid phosphatase (for prostate cancer); and isoenzymes and proteins whose concentrations are altered in biologic fluids in malignancy. The discoveries of α-fetoprotein (AFP) and carcinoembryonic antigen (CEA) in the 1960s led to the use of tumor markers for monitoring and to the use of the term oncodevelopmental markers, coined because these markers were produced both in fetal development and in tumors.

An ideal tumor marker should be specific for a given type of cancer and undetectable in healthy people or in those with benign disease, as well as sensitive enough to detect small tumors for early diagnosis or during screening. Unfortunately, most known tumor markers are neither specific nor sensitive enough for these purposes. Other desirable characteristics include concentrations proportional to tumor volume; short half-lives to allow early assessment of response to therapy; predictable increases and decreases in concentration responding to cancer progression and regression; and ability to be measured in a standardized and reproducible fashion in easily accessible specimens. In practice, tumor markers are most useful in evaluating the progression of disease status after the initial therapy and in monitoring the effectiveness of subsequent treatment. [1] [2] [3]

Despite the numerous tumor markers that have been identified, relatively few are in routine clinical use. Practice guidelines and recommendations for the use of tumor markers in a range of cancers have been developed by groups of scientists and clinicians from the National Academy of Clinical Biochemistry [4] [5] and the European Group on Tumor Markers,[6] although scientist and clinician viewpoints on the use of tumor markers do not always agree. The use of tumor markers is also included in Practice Guidelines in Oncology from the National Comprehensive Cancer Network,[7] and the American Society for Clinical Oncology has published recommendations for selected cancers. [8] [9] Practice guidelines from these and other national and international organizations are summarized by Fleisher and colleagues[4] and Sturgeon.[10]

Tumor marker use in the United States is limited compared with that in other countries as a result of required approval by regulatory agencies such as the U.S. Food and Drug Administration (FDA), which influences reimbursement. In addition, the FDA typically approves assays for specific clinical applications that may limit their use in other clinical settings. Currently approved markers and associated cancers and applications are listed in Table 20-1 .


Table 20-1   -- FDA-Approved Tumor Markers

Analyte

Associated Cancer

Designated Indication

Serum, Plasma

 

 

 CA15-3, CA27.29

Breast

Monitoring; recurrence

 HER-2

Breast

Monitoring

 Circulating tumor cells (CellSearch; whole blood)

Breast

Prognosis in metastatic disease

 CA125

Ovarian

Monitoring; second-look evaluation

 CA19-9

Pancreatic

Monitoring

 Total PSA, complexed PSA (cPSA)

Prostate

Detection in men aged ≥50 y in conjunction with DRE; monitoring; prognosis

 Free PSA

Prostate

Aid in distinguishing prostate cancer from benign prostate conditions in men ≥50 y with a total PSA of 4–10 ng/mL and nonsuggestive DRE in conjunction with total PSA (% free PSA)

 Prostatic acid phosphatase

Prostate

Monitoring

 Carcinoembryonic antigen

Colorectal, breast, lung

Monitoring; prognosis

 AFP

Nonseminomatous testicular

Monitoring

 AFP-L3

Hepatocellular

Risk assessment

 Des-γ-carboxy prothrombin

Hepatocellular

Risk assessment

 Thyroglobulin

Thyroid

Monitoring (in patients without thyroglobulin autoantibodies)

 Soluble mesothelin-related peptides

Mesothelioma

Monitoring

 β-human chorionic gonadotropin

None

Detection of pregnancy (not approved as a tumor marker)

Urine

 

 

 BTA stat, BTA TRAK

Bladder

Management in conjunction with cystoscopy

 NMP-22

Bladder

Diagnosis in symptomatic patients or those with risk factors; recurrence

 Tumor cell markers (ImmunoCyt)

Bladder

Management in conjunction with urine cytology and cystoscopy

Tissue

 

 

 Estrogen and progesterone receptors

Breast

Assessing the likelihood of response to therapy; prognosis and management

 HER-2

Breast

Identify patients eligible for trastuzumab (Herceptin) treatment

 EGFR

Colorectal

Identify patients eligible for cetuximab treatment

AFP, α-fetoprotein; BTA, bladder tumor antigen; DRE, digital rectal examination; EGFR, epidermal growth factor receptor; FDA, Food and Drug Administration; NMP, nuclear matrix protein; PSA, prostate-specific antigen.

 

 

 

METHODS

A new era in the tumor marker field began with the introduction of radioimmunoassay in the 1960s. In the 1970s, the development of enzyme-linked immunosorbent assays (ELISAs) and the discovery of monoclonal antibodies led to improvements in marker measurement and discovery of new markers. Cell-surface antigens identified with the use of monoclonal antibodies, such as the carbohydrate antigens CA125 and CA15-3, demonstrate improved clinical sensitivity and specificity compared with the oncofetal antigens. Recently, studies of oncogenes and tumor-suppressor genes, as well as development of molecular techniques such as recombinant DNA technology, polymerase chain reaction (PCR), and automated sequencing, have resulted in the understanding and use of tumor markers at the molecular level. [1] [2] New genomic and proteomic methods, such microarrays and matrix-assisted laser desorption/ionization-time-of-flight (MALDI-TOF) mass spectrometry [11] [12] will affect not only the discovery of new tumor markers, but also the way they are measured.

CLINICAL APPLICATIONS OF TUMOR MARKERS

Screening and Early Detection

Several factors must be taken into account when screening for a disease. The disease must be important, common, and the cause of substantial morbidity and mortality. An understanding of the natural history of the disease must ensure that early detection can affect the clinical course, and effective treatment must be available. The testing method also should be economical and noninvasive. When deciding to apply screening techniques, particularly if biochemical or immunochemical markers are used, an understanding of the analytic sensitivity (i.e., lowest detectable limit) and analytic specificity (i.e., extraneous interference) is essential. The precision of the assay (i.e., its ability to reproduce the results) also must be known and acceptable for use in large population studies. Clinical sensitivity and specificity, combined with the prevalence of the disease in the population, will affect the positive predictive value.[13]

Screening programs are most successful in regions or populations where specific cancers are highly prevalent. AFP concentrations have been used as a screening test for hepatocellular carcinoma in high -incidence areas including China, Japan, Taiwan, Africa, and Alaska. With a cutoff in the range of 10 to 20 mg/L, AFP has been shown to have a sensitivity of between 60% and 90% and a corresponding specificity of 70% to 80%, although hepatitis and cirrhosis also may cause elevated values. [1] [3] The American Cancer Society (ACS) has published specific recommendations for the early detection of breast, colorectal, prostate, and cervical cancers,[14] although among these cancers, the only serum tumor marker that is part of any screening algorithm is PSA.

Although controversy still exists on the value of early detection programs for prostate cancer, the ACS and the American Urological Association recommend PSA and digital rectal examination (DRE) annually beginning at age 50 for men with a life expectancy of at least 10 years.[14] It is recommended that prior to testing, men be offered information about the benefits and limitations of PSA testing so that they can make an informed decision with the assistance of their clinician. Other professional organizations, including the U.S. Preventive Services Task Force (USPSTF), American Academy of Family Physicians, and the American Medical Association advocate individualized testing decisions between patients and their clinicians, but do not recommend routine prostate cancer screening with either DRE or PSA. The USPSTF acknowledges that PSA testing is effective in identifying early prostate cancer, but did not find sufficient evidence that early detection improved health outcomes.[15]

Although PSA is for all intents and purposes organ specific, it is not cancer specific. PSA may be elevated (>4 ng/mL) in men with benign prostatic disease such as benign prostatic hyperplasia (BPH). PSA elevations also may occur with aging and with conditions such as prostatitis. Thus a number of methods have been proposed to increase the clinical utility of PSA for the early detection and diagnosis of prostate cancer, particularly in the diagnostic gray zone of 4 to 10 ng/mL, where there is significant overlap in PSA concentrations between prostate cancer and BPH. These approaches include age-specific reference ranges, PSA density, PSA velocity, and PSA molecular forms. [16] [17] The premise behind age-specific references (0–2.5, 3.5, 4.5, and 6.5 ng/mL for age ranges 40–49, 50–59, 60–69, and 70–79 years, respectively) is that lowering the upper end of the reference range in younger men would potentially increase sensitivity and aid in detecting organ-confined tumors earlier when surgery may be curative, whereas extending the range in older men would increase specificity, taking into account small increases in prostate volume, and PSA production and secretion with aging. However, because extending the range in older men may miss significant cancers, the use of age-adjusted ranges is controversial.

Assessment of PSA density, the ratio of PSA to prostate volume determined by transrectal ultrasound, is another method to increase PSA specificity; this assessment attempts to adjust for the increased volume often found in BPH. Limitations to PSA density include inaccuracies in measurement of prostate size and reduced sensitivity of prostate cancer detection. PSA velocity, similar to an approach used with CA125 for ovarian cancer, is defined as the change in PSA concentrations over time. A velocity of 0.75 ng/mL per year or more is suggestive of cancer. At least three PSA measurements 12 to 18 months apart are needed to calculate velocity. The effectiveness of this approach may be diminished by interassay and interlaboratory variation.

The most successful approach to increase the clinical utility of PSA in the 4- to10-ng/mL range is assessment of the free and complexed forms of the PSA molecule. In the early 1990s, it was discovered that the majority of PSA measured in serum is complexed to protease inhibitors (∼80% to 90%), with only a small portion in the free or unbound form. It also was discovered that men with prostate cancer, benign prostate disease, and no disease may differ in the proportions of free PSA and PSA bound to α1-antichymotrypsin. Although the mechanism currently is unknown, patients with prostate cancer have a lower percentage of free PSA (free PSA/total PSA) compared with men with benign disease. By using a cutoff of 25% in men with a negative DRE, 95% of cancers can be detected while sparing 20% of unnecessary biopsies.[16]

In other tumor systems, in most cases individual tumor markers lack sufficient sensitivity and specificity for screening. For this reason, an approach to improve their utility is to combine analysis of the marker with other procedures. In ovarian cancer, for example, strategies have included combining CA125 with ultrasound or using a two-stage strategy in which ultrasonography is performed only if CA125 concentrations are elevated. In a study of 4000 women, the specificity of CA125 plus ultrasound was 99.9% compared with 98.3% for CA125 alone. Other strategies that have been proposed to improve the specificity of CA125 include the use of multiple markers, including OVX1 and M-CSF. However, requiring elevation of all markers, while increasing specificity, sacrifices sensitivity; similarly, when markers are complementary, if elevation of only one marker is used as an indicator of disease, then specificity is lost. [18] [19]

Tumor Markers in Diagnosis

Lack of sensitivity and specificity limits the use of tumor markers for cancer diagnosis, although in contrast to screening of the general population, the prevalence of disease is likely to be higher in diagnostic situations. In most cases, histologic confirmation in tissue remains the gold standard for primary diagnosis, although tumor markers can still play a useful role, especially when combined with other methods, and may aid in differentiating benign and malignant disease or in identifying histologic tumor type.

For example, CA125, although still controversial as a screening marker, is more accepted as an adjunct in distinguishing benign from malignant disease in women, particularly in postmenopausal women with ovarian masses,[20] where elevated concentrations of CA125 greater than 95 U/mL in postmenopausal women can discriminate malignant from benign pelvic masses with a positive predictive value of 95%.[16] In premenopausal women, benign conditions resulting in elevated CA125 levels may be a confounding factor. The multiple markers approach has been applied to the preoperative discrimination of malignant and benign pelvic masses by using a number of analytic techniques. [21] [22]

AFP and hCG can play a useful role in the classification of germ cell tumors. AFP is the major serum protein of the early fetus synthesized by the fetal gut, liver cells, and yolk sac. hCG is a glycoprotein consisting of two distinct subunits and is synthesized and secreted by the placental syncytiotrophoblast. In seminomas, AFP is not elevated, but hCG is present in 10% to 30% of cases. Either hCG or AFP or both are produced by 60% to 90% of nonseminomatous germ cell testicular tumors at the time of diagnosis. Both markers are elevated in embryonal carcinoma (hCG > 65%; AFP > 70%), but these markers are not useful in teratomas. AFP is elevated in yolk sac tumors, whereas hCG is elevated in choriocarcinomas, and therefore useful in gestational trophoblastic disease as well. Typical reference values for AFP are 10 to 15 mg/L, whereas 5 IU/L is often used as a cutoff for hCG in testicular cancer. Clinically, assays for total β-hCG, measuring both intact hCG and the free β-subunit, may be preferable because of the production of free β-subunits in cancer.[1]

Other tumor markers play a role in testicular cancers. Placental alkaline phosphatase, another oncodevelopmental marker, can be elevated in seminomas and in a number of other cancers. Sensitivity is increased when it is combined with the nonspecific enzyme marker lactate dehydrogenase (LDH). LDH concentrations further correlate with tumor burden and, therefore, with prognosis. [1] [23]

Tumor markers also may play a role in the diagnosis of neuroendocrine tumors. [1] [3] [24] For example, the diagnosis of pheochromocytoma usually is established with an increase in the urinary excretion of catecholamines or catecholamine metabolites. Similarly, catecholamine metabolites vanillylmandelic acid (VMA) and homovanillic acid (HVA) are useful markers for neuroblastoma. Calcitonin is used in medullary thyroid cancer, with increased diagnostic sensitivity resulting from stimulation of calcitonin release with pentagastrin, whereas urinary 5-hydroxyindoleacetic acid (5-HIAA) is the primary test for overproduction of serotonin in carcinoid tumors. Specific circulating serum tumor markers may aid in the diagnosis of pancreatic endocrine tumors (e.g., insulin in insulinomas, gastrin in gastrinomas, glucagon in glucagonomas, and somatostatin in somatostatinomas). It also has been suggested that neuron-specific enolase (NSE), an isoenzyme of the glycolytic enzyme enolase found predominantly in neurons and neuroendocrine cells and elevated in neuroendocrine tumors, may be helpful in the diagnosis of small cell lung cancer in the small percentage of cases in which it is not possible to establish a diagnosis by biopsy.

Prognosis and Prediction of Therapeutic Response

The clinical staging of cancer can be aided by tumor markers, because the serum level of the marker typically reflects tumor burden. The marker value at the time of diagnosis may be used as a prognostic indicator for disease progression and patient survival. Tumor markers may be included as staging criteria for some cancers, as, for example, with CEA and colorectal cancer,[25] LDH in lymphoma,[26] or LDH, hCG, and AFP in testicular germ cell tumors.[27] They also may be combined with other clinical parameters in staging nomograms, such as PSA with preoperative clinical stage and biopsy Gleason score to predict pathologic stage in men with prostate cancer.[28]

In contrast to serum tumor markers, a number of tissue markers play a significant role in predicting response to therapy, particularly in breast cancer, where estrogen and progesterone receptors are used as predictive indicators for response to hormonal treatment and measurement of c-erbB-2 (HER-2) overexpression is used to identify patients for whom treatment with trastuzumab (Herceptin) may be beneficial.

Monitoring Disease

Most serum tumor markers are used to monitor treatment and progression of cancer. Markers may be used to determine the success of the initial treatment (e.g., surgery or radiation), detect recurrences of cancer, and monitor the effectiveness of treatment.[1] If surgery is successful, a marker level that was elevated before surgery should decrease afterward. The expected rate of decrease is projected from the half-life of the marker. If the half-life after treatment is longer than the marker's known half-life, it can be assumed that the treatment has not been successful in removing the tumor. The magnitude of marker reduction may, however, reflect the degree of success of the treatment or the extent of disease involvement. PSA is particularly useful for determining the success of initial surgical or radiation treatment of prostate cancer. If the tumor was organ confined and all prostatic tissue was removed, after radical prostatectomy PSA concentrations should decrease to undetectable levels. After allowing for sufficient clearance of pretreatment PSA (total PSA half-life of 2 to 3 days), finding detectable post-treatment PSA suggests remaining prostate tissue or the presence of metastases.[18]

In recurrent cancer after a successful initial treatment, marker values may not appear within the normal half-life. They may decrease to a steady level that is higher than normal, or remain within the reference interval for healthy individuals. A subsequent increase in the marker values suggests recurrence of the cancer.[1] Serum tumor markers may detect “biochemical recurrence of disease” before clinical evidence. Ultrasensitive PSA assays allow earlier detection of prostate cancer after radical prostatectomy. In patients with breast cancer receiving adjuvant chemotherapy, elevation of the breast cancer marker CA27.29 has been shown to indicate recurrent disease before any clinical evidence appears. Early detection of cancer recurrence may be helpful if it makes it possible to initiate early treatment or change therapy, but it is useful only if effective treatment is available.[13]

Levels of most tumor markers correlate with the effectiveness of treatment and response to therapy. In breast cancer, the concentration of markers such as CA27.29 changes with the treatment and the clinical outcome of the patient, as does CEA in colorectal cancer, CA 19-9 in pancreatic cancer, and CA125 in ovarian cancer. Marker values usually increase with progressive disease, decrease with remission, and do not change significantly with stable disease. Tumor-marker kinetics are, in general, more important than individual values, although interpretation of kinetic changes may be complicated. For example, marker levels in response to treatment may show an initial delay before demonstrating the expected pattern of change.[29]

The Working Group on Tumor Marker Criteria of the International Society for Oncodevelopmental Biology and Medicine[30] has published the following criteria for the interpretation of changes in tumor marker values:

  

   

“If no therapy is given, at least a linear increase in three consecutive samples (i.e., two time intervals) on a log scale should be registered to establish a recurrence. Usual intervals could be 3 months but are clinically determined. After a first increase, next samples should be taken after 2 to 4 weeks, irrespective of the absolute level.” If therapy is given, the changes in marker values should reflect the clinical progression of the disease.

  

   

“Progressive disease is defined by an increase in the marker level of at least 25%. Sampling should be repeated within 2 to 4 weeks for additional evidence. The sampling interval during therapy may depend on the type of tumor and should be related to clinical follow-up.” A decrease in marker value of at least 50% is indicative of partial remission “with the concept that tumor load is related to the changes in serum tumor marker levels.”

  

   

The Working Group also provided a general statement that “a complete remission cannot be determined by tumor marker levels, but if tumor marker levels are elevated, the clinical decision of complete remission based on conventional methods should be considered incorrect unless an explanation for the presence of an elevated level is given.”

ANALYTIC CONSIDERATIONS

A number of clinical and analytic limitations should be considered in the interpretation and use of tumor markers. [29] [31] [32] Physiological influences to be considered include biologic variability, effects of aging and menopause, and half-life and route of elimination. Tumor markers can be elevated as result of renal failure, liver failure, and cholestasis, depending on whether the marker is eliminated through glomerular filtration or metabolized by the liver. For example, serum CEA concentrations may be elevated in patients with liver disease. A false-positive PSA, for example, may be seen with prostatitis, while inflammation of serous membranes or biliary ducts may falsely elevate CA125 or CA19-9, respectively.[31] Smokers have higher concentrations of CEA in comparison to nonsmokers, and reference ranges for CEA in lung cancer typically are stratified by smoking status. This underscores the importance of selection of cutoff values with respect to selection of reference groups. Knowledge of whether the reference group is based on healthy individuals or patients with benign diseases will influence interpretation.[29]

Knowledge of the tumor marker level before surgery is important for subsequent use in monitoring, because markers are not 100% sensitive, even in advanced disease. Tumors can be nonsecreting, or host genetic factors can affect the ability to detect an antigen. For example, CA 19-9, a sialylated derivative of the Lewis(a) blood group antigen, is not expressed in the estimated 3% to 5% of the population with the Lewis(a-b-) genotype.[1] Treatment also may influence marker concentrations, with increased release with chemotherapy or surgery, such as the spike observed in CA125 concentrations after abdominal surgery.[3] Tumor manipulation also can influence concentrations, as evidenced by the increased PSA concentrations observed after prostate biopsy, transurethral resection, prostate massage, or other procedures. Although the influence of DRE is thought to be minimal, it is recommended that blood for PSA measurements be drawn before the procedure or a minimum of 1 week after the DRE.[33]

One of the most critical analytic considerations in tumor-marker result interpretation is the fact that different assays, specifically immunoassays, may give different values for the same marker. Differing results among assays can be attributed to calibration differences resulting from different materials, value assignments, antibody types and specificities, assay designs and kinetics, variations in reference ranges or cutoffs, and assay robustness. [31] [34] [35] Therefore, values cannot be used interchangeably, and it is recommended that patients be followed up using the same laboratory method. Clinicians also should be informed when laboratories change tumor marker assay methods and should be offered a crossover period and an explanation of the association between the old and new assays.

Factors affecting assay robustness include the measuring range of the assay, encompassing the detection limit and upper limit where sample dilution is not required; accuracy and precision of dilutional linearity; high-dose hook or prozone effect leading to falsely low results; and interferences from endogenous antibodies. Endogenous antibodies include either anti-analyte antibodies such as thyroglobulin autoantibodies or anti-reagent antibodies such as heterophile and human anti-mouse antibodies (HAMA). HAMA, which may be produced as a result of treatment with monoclonal antibodies or immunoscintigraphy with antibodies similar to assay antibodies, can bind to reagent antibodies with resulting falsely high or low values, depending on assay design. HAMA can be minimized by using chimeric antibodies in vivo and at the assay level by choice of assay antibodies and inclusion of nonimmune animal serum as a blocking technique. [1] [35]

FUTURE DIRECTIONS

With current advances in proteomic and genomic technology, the diagnosis of disease in the future will be based on a combination of methods, with classification based on molecular as opposed to morphologic features. Unique gene or protein profiles of multiple biomarkers, accounting for cancer heterogeneity, will be measured in tissue, cells, and body fluids. The analysis of panels of protein biomarkers may be performed by traditional immunoassay, antibody-based protein chips, or microarrays. Furthermore, many more diagnostic tests will be generated as the result of genomic and proteomic discoveries. For example, serum biomarkers for a number of cancers recently have been identified using mass spectrometry. [36] [37] Integrated diagnostic tools that combine these methods with molecular imaging techniques will be used, and bioinformatics will play a key role. The rapid translation of tests from the laboratory to the bedside will elevate the importance of laboratory testing in cancer diagnosis and care of the cancer patient.

REFERENCES

  1. Chan DW, Booth RA, Diamandis EP: Tumor markers.   In: Burtis CA, Ashwood ER, Bruns DE, ed. Tietz Texbook of Clinical Chemistry and Molecular Diagnostics,  3rd ed.. St. Louis: Elsevier Saunders; 2006:745-795.
  2. Diamandis EP: Tumor markers: past, present, and future.   In: Diamandis EP, Fritsche HA, Lilja H, et al ed. Tumor Markers. Physiology, Pathobiology, Technology, and Clinical Applications,  Washington, DC: AACC Press; 2002:3-8.
  3. Duffy MJ: Clinical uses of tumor markers: a critical review.  Crit Rev Clin Lab Sci2001; 38:225-262.
  4. Fleisher M, Dnistrian AM, Sturgeon CM, et al: Practice guidelines for the use of tumor markers in the clinic.   In: Diamandis EP, Fritsche HA, Lilja H, et al ed. Tumor Markers. Physiology, Pathobiology, Technology, and Clinical Applications,  Washington, DC: AACC Press; 2002:33-63.
  5. American Association for Clinical Chemistry : LMPG: Practice Guidelines and Recommendations for Use of Tumor Markers in the Clinic.  Available from http://www.aacc.org/AACC/members/nacb/LMPG/OnlineGuide/DraftGuidelines/TumorMarkers/TumorMarkersPDF.htm
  6. European Group for Tumor Markers (EGTM) : Consensus recommendations.  Anticancer Res1999; 19:2785-2820.
  7. National Comprehensive Cancer Network: Available at http://www.nccn.org/
  8. Bast Jr RC, Ravdin P, Hayes DF, et al: 2000 update of recommendations for the use of tumor markers in breast and colorectal cancer: clinical practice guidelines of the American Society of Clinical Oncology.  J Clin Oncol2001; 19:1865-1878.
  9. Locker GY, Hamilton S, Harris J, et al: ASCO 2006 update of recommendations for the use of tumor markers in gastrointestinal cancer.  J Clin Oncol2006; 24:5313-5327.
  10. Sturgeon C: Practice guidelines for tumor marker use in the clinic.  Clin Chem2002; 48:1151-1159.
  11. Zhu H, Bilgin M, Snyder M: Proteomics.  Annu Rev Biochem2003; 72:783-812.
  12. Colantonio DA, Chan DW: The clinical application of proteomics.  Clin Chim Acta2005; 357:151-158.
  13. Chan DW, Schwartz MK: Tumor markers: introduction and general principles.   In: Diamandis EP, Fritsche HA, Lilja H, et al ed. Tumor Markers: Physiology, Pathobiology, Technology, and Clinical Applications,  Washington, DC: AACC Press; 2002:9-17.
  14. Smith RA, Cokkinides V, Eyre HJ: American Cancer Society guidelines for the early detection of cancer, 2006.  CA Cancer J Clin2006; 56:11-25.
  15. Harris R, Lohr KN: Screening for prostate cancer: an update of the evidence for the U.S. Preventive Services Task Force.  Ann Int Med2002; 137:917-929.
  16. Polascik TJ, Oesterling JE, Partin AW: Prostate specific antigen: a decade of discovery: what we have learned and where we are going.  J Urol1999; 162:293-306.
  17. Sokoll LJ, Chan DW: Total, free, and complexed PSA: analysis and clinical utility.  J Clin Ligand Assay1998; 21:171-179.
  18. Bast Jr RC, Urban N, Shridhar V, et al: Early detection of ovarian cancer: promise and reality.  Cancer Treat Res2002; 107:61-97.
  19. Bast Jr RC, Badgwell D, Lu Z, et al: New tumor markers: CA125 and beyond.  Int J Gynecol Cancer2005; 15(Suppl 3):274-281.
  20. NIH Consensus Conference Development Panel of Ovarian Cancer: Ovarian cancer. Screening, treatment, and follow-up.  JAMA1995; 273:491-497.
  21. Woolas RP, Conway MR, Xu F, et al: Combinations of multiple serum markers are superior to individual assays for discriminating malignant from benign pelvic masses.  Gynecol Oncol1995; 59:111-116.
  22. Zhang Z, Barnhill SD, Zhang H, Xu F, et al: Combination of multiple serum markers using an artificial neural network to improve specificity in discriminating malignant from benign pelvic masses.  Gynecol Oncol1999; 73:56-61.
  23. Einhorn LH, Lowitz BB: Testicular cancer.   In: Casciato DA, Lowitz BB, ed. Manual of Clinical Oncology,  3rd ed.. Boston: Little Brown; 1995:228-236.
  24. Oberg K, Stridsberg M: Neuroendocrine tumors.   In: Diamandis EP, Fritsche HA, Lilja H, et al ed. Tumor Markers. Physiology, Pathobiology, Technology, and Clinical Applications,  Washington, DC: AACC Press; 2002:339-349.
  25. Compton CC, Fenoglio-Presier CM, Pettigrew N, Fielding LP: American Joint Committee on Cancer prognostic factors consensus conference: colorectal working group.  Cancer2000; 124:1739-1757.
  26. The International Non-Hodgkin's Lymphoma Prognostic Factors Project : A predictive model for aggressive non-Hodgkin's lymphoma.  N Engl J Med1993; 329:987-994.
  27. In: Sobin LH, Wittekind C, ed. Testis: TNM Classification of Malignant Tumors,  5th ed.. New York: Wiley-Liss; 1997:174-179.
  28. Partin AW, Mangold LA, Lamm DM, et al: Contemporary update of prostate cancer staging nomograms (Partin Tables) for the new millennium.  Urology2001; 58:843-848.
  29. Fateh-Moghadam A, Stieber P: Sensible Use of Tumour Markers,  2nd ed.. Marloffstein-Rathsberg, Hartmann Verlag AMBL; 1993:11-31.
  30. Bonfrer JMG: Working group on tumor marker criteria (WGTMC).  Tumor Biol1990; 11:287-288.
  31. Basuyau J-P, Leroy M, Brunelle P: Determination of tumor markers in serum: pitfalls and good practice.  Clin Chem Lab Med2001; 39:1227-1233.
  32. Schrohl AS, Holten-Andersen M, Sweep F, et al: Tumor markers: from laboratory to clinical utility.  Mol Cell Proteomics2003; 2:378-387.
  33. Price CP, Allard J, Davies G, et al: Pre- and post-analytical factors that may influence use of serum prostate specific antigen and its isoforms in a screening programme for prostate cancer.  Ann Clin Biochem2001; 38:188-216.
  34. Sturgeon CM, Seth J: Why do immunoassays for tumor markers giver differing results? A view from the UK national External Quality Assessment Schemes.  Eur J Clin Chem Clin Biochem1996; 34:755-759.
  35. Sturgeon CM: Limitations of assay techniques for tumor markers.   In: Diamandis EP, Fritsche HA, Lilja H, et al ed. Tumor Markers: Physiology, Pathobiology, Technology, and Clinical Applications,  Washington, DC: AACC Press; 2002:65-81.
  36. Zhang Z, Bast Jr RC, Yu Y, et al: Three biomarkers identified from serum proteomic analysis for the detection of early stage ovarian cancer.  Cancer Res2004; 64:5882-5890.
  37. Li J, Zhao J, Yu X, et al: Identification of biomarkers for breast cancer in nipple aspiration and ductal lavage fluid.  Clin Cancer Res2005; 11:8312-8320.