Toward Building Better Biomarkers Statistical Methodology
Event Details
Clinical Considerations
There is concern regarding the clinical utility of biomarkers in kidney disease (Acute Kidney Injury [AKI], Chronic Kidney Disease [CKD], various forms of glomerulonephritis [GN], and polycystic kidney disease [PKD]) for prediction of diverse clinical outcomes (such as loss of renal function, development of cardiovascular disease, diminution of quality of life, death), as well as in drug development (for refinement of risk stratification or drug response). There is dissatisfaction with the results of published biomarker studies and the lack of uptake of biomarkers in clinical practice in patients with kidney disease. Finally, we don’t know the answers to several key questions. Some of these are listed below.
- Of what use is this biomarker to me or my patient in the clinic?
- Has “context of use” or “fit purpose” been identified, or do the present studies evaluate biomarkers on convenience samples?
- How can we deal with the paradox of well-established but crude measures which biomarkers must enhance?
- Can we develop markers for kidney disease that are superior to S[Cr] and proteinuria?
- Does the Framingham study provide a better framework for the evaluation of biomarkers—or do biomarker studies have to be designed de novo for each context?
- How does biomarker science allow us to move from a population focus to individual considerations: the patient and the physician in the consultation room (or at the bedside)—considering a distinct outcome?
- Can we as biomarker scientists develop physician / patient perspectives—assessing risk for the individual, or outcome for the individual with drug therapy?
Statistical Considerations
Have we developed a true set of biomarker statistical analyses? Or are we fitting old techniques to new contexts / issues? What does a biomarker showing independent association with a distant outcome in a Cox regression mean? What are the key elements of design for a meaningful biomarker study? What metrics are appropriate for assessing incremental value of biomarkers?
- What does “prediction” mean?
- Is “prediction” different for present measures (eg eGFR) compared to future outcomes (risk of future decrement in renal function, need for ESRD treatment, or death)?
- How can we make predictive models in kidney disease patients more precise?
- Can causal links between biomarker levels or changes identified by present statistical methods be strengthened?
This workshop will grapple with these thorny questions. Breakout groups, composed of biostatisticians and clinical nephrologists, will consider the issues and put forth approaches for moving forward.
Event Logistics
Location
Natcher Conference Center
NIH Building 45
Room F1/F2
Bethesda,
MD
20894
Contacts
Program Contact
Paul L Kimmel, M.D., M.A.C.P., F.A.S.N.
NIDDK
Meeting Logistics
David Henson
The Scientific Consulting Group, Inc.