Guidelines for Marker Development
DRAFT
Background
The PACCT Strategy Group developed the following document to help evaluate whether markers or assays are ready for use in clinical settings. It should be possible to determine what further steps need to be taken by critically examining available data. These guidelines are still in draft form. There are gaps that need to be filled and statistically valid study designs that need to be developed. We are putting the guidelines on this website to begin a dialog. Some of this material has been incorporated into an article in Seminars in Oncology (Hammond ME and Taube SE. Issues and Barriers to Development of Clinically Useful Tumor Markers: A Development Pathway Proposal. Seminars in Oncology 29(3):213-221, 2002). Comments on these guidelines can be sent to Dr. Sheila Taube (st29f@nih.gov).
Draft guidelines for marker development: Data required to proceed through the development steps
Step 1 - Considerations for preliminary identification of a potentially useful marker
- Is there a biological rationale for this marker? If so, the rationale can be used initially as part of the justification for the potential interest or value of the marker. If, however, a marker (or set of characteristics such as a cluster of expressed genes) is identified serendipitously or the biological rationale is not obvious, then preliminary data to support the suggestion that the marker may have some interest or be worth further investigation will be required. The data can include evidence that the marker predicts some outcome of interest; is associated with some clinical parameter; is prognostic or diagnostic; or appears to identify sub-populations of patients.
- Is there an assay system available that is working in at least one laboratory with reasonable reproducibility? Is there a reasonable scoring system?
- Has the marker been examined in normal as well as abnormal/diseased tissue? Has it been examined in different organ sites? This is not so much to establish exact distribution characteristics but to help determine the setting where the marker might have greatest value. The types of tissues that should be examined include benign diseases of the same organ and premalignant lesions. The source of so-called normal tissue should be described.
- Can a patient population be defined for which this marker may have utility? What is an expected range for the prevalence of this marker in populations of potential interest? Very rare (< 5%) or very prevalent (>95%) markers are likely to be useful in more circumscribed settings, e.g., they distinguish malignant from benign or definitively predict the response to therapy. Markers with intermediate prevalence may be useful to identify subsets of a patient population with different outcomes. The numbers of specimens that should be assessed at this stage will vary depending on the question being asked or the intended use of the marker. If prevalence is being assessed, then >20 specimens should be examined so that a marker present in 5% of cases would have a reasonable chance of being detected in the set of specimens. The numbers to be assessed for other questions will depend on the statistical design, the difference that would be meaningful to detect.
- Can the marker be measured in the types of specimens that will generally be available?
Step 2 - Development of an assay and assay refinement
- Define the specificity of the assay, i.e., what the assay is designed to measure.
- Establish initial optimum conditions of assay, e.g. types of specimens, fixation, antigen retrieval, reagents/components of the assay system, detection system, positive and negative controls, etc. Such assay characteristics have to be established for any proposed technique including molecular techniques.
- Define a scoring procedure and how the data will be reported.
- Assess the consistency of various characteristics of the assay, including expression in different types of specimens (including different organs, normal, diseased), staining patterns (membrane, nucleus, etc.) reproducibility of PCR or other molecular analyses, positive and negative controls, etc.
- Demonstrate intra-laboratory reproducibility under initial optimum assay conditions.
- Estimate prevalence of the marker on an expanded collection of targeted specimens. These specimens should be of the same type (fixed, frozen, etc.) for which the assay is being proposed, and the patient characteristics of this pilot set should be similar to the target patient group. Tissue reference sets would facilitate this step. In some cases, these reference sets can be prepared as tissue micro-arrays.
- If the marker is being evaluated as a prognostic indicator, correlation between the new marker and standard prognostic variables should be reported. This provides a suggestion as to whether the marker is likely to serve as an independent indicator. If it is highly correlated with a known marker, it may add information, but it may also not be of value. If the marker provides equivalent information about the disease but is easier to measure or has better assay characteristics, it may replace the older marker. If it does not correlate, it may provide new information about the disease.
- There should be a preliminary proposal of a clinical question for which the marker might be useful. Some pilot data should be presented supporting this (although statistical significance may not be attainable at this point).
Step 3 - Preliminary evaluation of clinical usefulness
- What clinical question is the marker intended to address? Diagnosis, prognosis, prediction of response, disease monitoring, etc.?
- Is the sensitivity or the accuracy of the assay adequate to address the clinical question? Does the marker correlate with other measures used in the same clinical setting? Does it provide additional information?
Step 4 - Assay standardization
- In addition to the assay characteristics addressed in Step 2 above, the robustness and reproducibility of the assay have to be addressed. What is the intra- and inter-laboratory reproducibility? What is the ability to scale up the assay to be able to assess large numbers of specimens reproducibly?
Step 5 - Clinical utility assessment
- This step is basically a reiteration of Step 3 with refinement of the clinical question. The intended use should be more clearly defined and careful statistical designs applied to studies that usually have to include larger numbers of cases.
- Proceed with a hypothesis-testing trial.
Steps 4 and 5 in practice are repeated, with refinements, probably 2-3 times prior to final validation steps.
Step 6 - Validation of both the assay and the clinical utility
- Is the assay adequately reproducible? The FDA has developed guidance documents for the kind of assay variability studies needed to document a product. The web site at: http://www.fda.gov/cdrh/devadvice/ provides the information required by FDA for different types of submissions.
- Do the results of the assay provide usable and important clinical information? Is the information reliable enough to support the intended patient management decision?
Thought should be given to how markers can be efficiently tested for more than one clinical use, e.g., prediction and prognosis, taking into account the example of the HER2 history.
Section Last Updated: 07/25/07
