As our understanding of the molecular basis of cancer has grown, so has the availability of tests to assess disease processes in dogs and cats. A cancer biomarker can be broadly defined as a molecule that indicates the presence of cancer or provides information about the likely future behavior of the cancer, which may include response to therapy.
To assess the potential value and application of currently available biomarkers in veterinary medicine, it is essential to understand process and pitfalls of biomarker discovery and validation. Analytic validity, clinical validity and clinical utility all need to be demonstrated. Despite decades of work and extensive resources, even in human medicine the number of biomarkers with clinical utility is limited.
Rapid expansion in genomic profiling is increasingly informing the care of human cancer patients. Biomarkers in veterinary oncology are still a developing field, with frequent additions to literature. Critical analysis is important as new information becomes available.
Biomarkers – Definition and Development
What is a Biomarker?
A cancer biomarker, also referred to as a marker or molecular diagnostic, is a molecule that indicates the presence of cancer or provides information about the likely future behavior of the cancer. Biochemically, biomarkers may include DNA, RNA, proteins or peptides, hormones, metabolites or markers that represent processes such as apoptosis, angiogenesis or proliferation. Biomarkers are detected in circulation, secretions, other body fluids or tissue.
The World Health Organization provides a broad definition, stating, “A biomarker is any substance, structure or process that can be measured in the body or its products and can influence or predict the incidence or outcome of disease.”
Biomarker Uses
In oncology, biomarkers may be used for a variety of purposes, including estimating the risk of developing cancer, screening for cancer, obtaining a definitive diagnosis, determining prognosis, predicting response to therapy, monitoring for disease recurrence and assessing response or progression in metastatic disease. Risk assessment involves evaluating epigenetic modifications as signatures of risk factor exposure.
If developing a biomarker for early cancer detection, many requirements must be met, and these can be challenging; few such markers exist even in human medicine.
The marker must detect cancer at an asymptomatic stage, a positive result must lead to decreased morbidity or mortality, the test should be inexpensive and safe, and it must have very high sensitivity and exceptional specificity.
Unlike early cancer detection, biomarkers used to obtain a diagnosis are used when the tumor is evident and the patient is symptomatic. In human medicine, no biomarkers are recommended as a sole diagnostic tool, but several are used to help with diagnosis or to subclassify diseases. A biomarker used for prognosis should predict the likely outcome both without and with therapy. Such markers should reduce the overtreatment of low-grade malignancies and prevent under-treatment of aggressive tumors.
Molecular biomarkers can play a key role in the selection and monitoring of therapy, individualized medicine, informing both drug and dose choices. Such markers should help reduce toxicity and save money by helping to modify doses and decreasing the use of drugs that will not be of benefit, as well as providing a way to assess if the treatment is working.
The FDA has noted that the judicious use of biomarkers will be important in decreasing the failure of clinical trials for new drugs by enriching the trial populations with specific molecular subtypes more likely to respond. A critical point is that “no prognostic marker can accurately predict outcome for an individual patient; it provides a probability estimate of outcome for a heterogenous population of patients.”
Biomarker Discovery and Validation
The process begins with preclinical exploratory studies to identify a marker. Two approaches can be taken: hypothesis-driven, whereby a marker is sought based on prior expectations, and discovery-based, which uses high-throughput genomics to screen normal and cases to identify a difference. Most biomarker discovery efforts fail at this early stage, often due to only modest differences between cases versus controls or large variability in marker levels in the healthy, control subjects.
Once a marker is identified, the next step is clinical assessment development. Analytical validity, clinical validity and clinical utility all need to be verified. The assessment should be robust, accurate and reliable. Clinically, it must divide the population into two groups with different outcomes, reflecting adequate specificity and sensitivity. With a low prevalence of disease, very high specificity or sensitivity is crucial for the assay to achieve strong positive or negative predictive values.
For the assay to have clinical utility, it must improve patient outcomes. Ideally, clinical utility is evaluated through randomized controlled trials. Most biomarkers fail at this stage due to low sensitivity or specificity, which limits their clinical usefulness.
Various study types can be used to assess clinical validity and utility. Retrospective longitudinal repository studies are one approach to evaluating a marker’s diagnostic ability for cancer screening. This testing paradigm uses stored samples from patients with known outcomes, comparing those in the preclinical phase of disease development with controls. Large sample sizes are required for this type of study, and samples must reflect the biological variability of the target population.
Prospective screening studies are another approach to determine whether the marker can detect cancer at an early stage (i.e., it performs as intended).
This type of study allows calculation of positive predictive value and false-positive rates. In this design, asymptomatic patients are screened and followed over time to determine if they have cancer and what stage it is in. This type of study involves active patients, and the screening tests will lead to diagnosis and treatment. Case-control studies for biomarkers used for early cancer detection are large-scale population studies designed to determine if screening reduces morbidity or mortality.
Bias in Evaluations of Biomarker Screening Validity
Several types of bias can occur when evaluating biomarkers for cancer screening and determining whether their use decreases morbidity or mortality. First, lead-time bias can make survival time with treatment appear improved, but this may be due to longer observed survival resulting from earlier detection, rather than an actual improvement in outcome.
Length bias is a type of prognostic selection bias in which slower-progressing cancers are more likely to be detected through screening programs and have better outcomes than more aggressive cancers identified after the onset of clinical signs.
Compliance bias relates to the extent to which patients (or owners in veterinary medicine) adhere to medical advice after screening. Many factors influence outcomes in this type of bias; for example, if screening requires significant commitment, those owners may also be more dedicated to treatment. This may create the appearance that pets do better because of biomarker use, when in fact the outcome is influenced by the preselection of compliant owners.
Selection bias may occur when healthier dogs are presented for screening or when more ill dogs are overrepresented. It can be difficult to determine the direction of this bias; for example, patients may respond better to treatment because they were healthier at baseline, or they may appear to do worse because they were more ill. However, these differences are not necessarily related to the screening assay or timing of diagnosis.
Overdiagnosis bias can make a biomarker appear useful for decreasing morbidity or mortality when patients are incorrectly diagnosed and therefore appear to do better because they do not have the disease. Care must be taken to avoid these biases in studies assessing biomarkers for cancer screening, and such studies should be critically reviewed with these biases in mind.
Some Diagnostics (Biomarkers) of Use in Veterinary Oncology
PARR: The PCR for antigen receptor rearrangement assay is used to determine whether lymphocytes in a lesion are polyclonal (and thus reactive) or monoclonal (consistent with neoplasia). The specificity of a monoclonal result for lymphoma is high, whereas the sensitivity is lower; therefore, the test is prone to false negatives when evaluating for lymphoma. This assay may help determine whether the tumor is of B- or T-cell origin, but flow cytometry and immunocytochemistry may be the preferred methods for immunophenotyping.
PARR can be performed on a variety of samples, including blood, tissue, fluids and even fixed, stained cytology slides. Reported specificity and sensitivity for lymphoma in dogs are 94% and 75%, respectively, and in cats are greater than 90% and approximately 65%.
Since the test was first developed, the release of genome assemblies for multiple species has enabled the development of clonality assays that cover all genes within a locus, which may decrease false-negative results. Several recent studies have evaluated improved assay techniques; however, clinical validity and utility will need to be further assessed.
Flow cytometry: Cell surface markers assessed via flow cytometry can be used to diagnose specific types of hematologic neoplasia. Identification of certain markers, such as CD45 and class II MHC, in dogs with lymphoma has been correlated with outcomes; thus, flow cytometry may be used to better characterize lymphoma in individual patients, helping to guide prognosis and therapeutic decisions.
Immunocytochemistry: Staining to determine the immunophenotype of lymphoma can be performed on cytologic samples, providing important prognostic information with minimal morbidity and cost.
Molecular profiling performed on 35 samples of canine lymphoma divided cases into three groups based on gene expression profiles: low-grade T cell, high-grade T cell and B-cell. Group classification correlated with event-free survival and may help predict prognosis in dogs with lymphoma.
Thymidine kinase 1 (TK1): This DNA precursor enzyme’s expression correlates with cell proliferation and turnover, and serum activity assays are available for dogs and cats. In dogs, TK1 levels may be of some benefit in diagnosis, prognostication and monitoring response to therapy for a variety of malignancies. In one study, TK1 testing combined with C-reactive protein levels was used to generate a Neoplasia Index as a means of screening for underlying cancer. As with many of these biomarker tests, this is no longer used.
The specificity of the test was relatively high; however, overlap remained in many cases in which neoplasia was not identified. In cats, serum TK1 was evaluated as a biomarker to differentiate gastrointestinal lymphoma from inflammatory bowel disease. While cats with lymphoma had significantly higher scores than normal cats, cats with inflammatory bowel disease or nonhematopoietic neoplasia showed overlapping scores, limiting clinical validity.
Canine lymphoma blood test
The cLBT involves measurement of serum levels of C-reactive protein and haptoglobin, which are then combined in an algorithm to provide a score. While the test may help detect early relapses, improved patient outcomes resulting from earlier detection will need to be demonstrated to validate clinical utility. This test is no longer offered.
MDR1: P-glycoprotein is a drug efflux carrier encoded by MDR1 (also known as ABCB1). Upregulation of MDR1 may lead to chemotherapy resistance, while dogs with a mutation may have increased sensitivity to many drugs, including several classes of chemotherapeutics. Testing for MDR1 mutation status may be considered for at-risk breeds.
Cardiac troponin I (cTnI): Identification of this myofibrillar protein in serum is a sensitive and specific marker for myocardial ischemia and necrosis. It has been shown to be elevated in dogs with pericardial effusion due to hemangiosarcoma compared with dogs with idiopathic effusion. In cases of pericardial effusion without a visible mass, this test may be useful in differentiating hemangiosarcoma from idiopathic effusion.
c-Kit: The tyrosine kinase receptor c-Kit may be used in cancer diagnostics in several ways. Its presence on gastrointestinal sarcoma cells can help confirm a diagnosis of a GIST (gastrointestinal stromal tumor). Mutations in the c-Kit receptor have been identified in several canine cancers, including mast cell tumors and lymphoma, and may serve as a prognostic indicator. Additionally, therapies that inhibit c-Kit signaling are available, and this biomarker may help predict response to therapy.
Serum Ki-67: Ki-67 is expressed only during the active cell cycle and is therefore a biomarker of cellular proliferation. When assessed in a small cohort of dogs (eight healthy, 13 with nonmalignant disease and 20 with malignancy), dogs with malignant tumors had significantly elevated serum concentrations. Sensitivity and specificity were both 100% when assessing tumors characterized as moderate to highly malignant based on pathologic features. Further work is needed to validate the assay and establish clinical validity and utility, including prospective trials for cancer screening in asymptomatic patients.
BRAF gene mutation: This mutation is found in approximately 80% of urothelial and prostatic carcinomas, and an assay has been developed to detect the mutation in urine. The assay is not affected by hematuria or bacteriuria. It has very high specificity but lower sensitivity. It may be helpful in noninvasively confirming a diagnosis of neoplasia. Further study is needed to determine whether it provides clinical utility in early detection by demonstrating improved outcomes.
About the Author
Laura Garrett | DVM, DACVIM (Oncology)
Dr. Laura Garrett is a Clinical Professor in Oncology, Coordinator of Communication Training, head of Specialty Medicine, and the Khan Family Chair of Veterinary Clinical Oncology at the University of Illinois College of Veterinary Medicine. Her training includes a DVM degree from University of Illinois in 1991, internship at University of Minnesota, 2 years in private practice, and residency at University of Wisconsin-Madison. After working in a New Zealand specialty clinic, she was on faculty at Kansas State University prior to returning to Illinois in 2006. Topics of publications include lymphoma, mast cell tumors, cytology, and communication skills. She has served as the President of the Veterinary Cancer Society and the Chair of the Board of Regents for the American College of Veterinary Internal Medicine. She has five times received the student-nominated Mell Faculty Award for empathy and honoring the bond between owners and pets.