New Drug Development Tools (DDT): Connecting the Dots Between Biomarker Discovery and Improved Patient Outcomes

For at least two decades, biological markers, or biomarkers, have been promoted as key advances in scientific discovery that provide a measure of biological function. Researchers and, indeed, the public expect biomarkers to be useful in delivering science-based improvements in health care, particularly in clinical diagnostics. After all, a biomarker is defined as, “A characteristic that is objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes or pharmacologic responses to therapeutic intervention.”1 Thus, biomarkers are expected to be diagnostically significant in aiding the clinician in sorting through the various potential causes underlying macro symptoms such as fever, pain, nausea, and hives.

Biomarkers have intuitive appeal to scientists and clinicians. Many have been reported, but elevating these discoveries to practice has been slow. After discovery, many dots must be connected between validation, development of assays, adoption of assays, etc. Results, in terms of improved patient outcomes, are disappointing by any standard.

As organizations charged with improving health care, the U.S. FDA and European Medical Agency have responded with regulatory qualification of biomarkers, where the agencies seek to provide a framework to advance discovered biomarkers to assays and other uses. Qualification includes validation of the testing platform and recognition of “fit for purpose.” “Purpose” includes documentation of intended use, including sample availability and utility in practice.

In May 2012, the FDA announced three Drug Development Tools Qualification Programs to facilitate development of protocols for: 1) biomarkers, 2) clinical outcomes assessments, and 3) animal models. “Qualification” means that, once a tool is qualified, the FDA recognizes that the drug development tool (DDT) can be relied upon in development and regulatory review. The DTT is context specific. Qualification does not mean that the DDT can be used in clinical diagnostic or therapeutic purposes without further regulatory review. It does, however, recognize that the tool qualifies as suitable for intended use in a specific context.

The expected benefits of the DDT program are: 1) expediting drug and diagnostic development and regulatory review, 2) stimulating scientific collaboration in development of DDTs, 3) encouraging adoption of qualified DDTs in regulatory processes, 4) promoting development of DDTs for unmet needs, and 5) supporting innovation in drug development.

Biomarkers

Many scientists and regulators were frustrated with the arduous path that proponents of a particular biomarker needed to traverse to gain acceptance. It seems to be an informal trail of building consensus. For example, an academic research team could discover and claim a “possible biomarker.”

This merits a publication, but then what? How can a research group remove the “possible”? The discoverer is probably a graduate student or postdoc. They move on and are replaced with others who want to make similar discoveries to advance their careers. From society’s viewpoint, discovery is only the first step. The candidate biomarker needs to be validated and the assay reduced to practice. This requires development, regulatory approval, manufacturing, marketing, and technical support.

The DDT qualification process outlined by the FDA consists of more than 20 steps, starting with the sponsor contacting the FDA.2 A biomarker cannot be qualified without a reliable means with which to measure it. The assay results should not depend on the measurement technology. Only three biomarkers are listed on the FDA’s web site as of April 2013. These were all started prior to the formal program announcement in May of 2012. Obviously, this is not much activity. The lengthy process proposed by the FDA will certainly test the patience of any submitter.

One concern is how to add newly discovered co-markers. Systems biology is showing that many maladies can have multiple markers. One just does not give a complete picture.

Clinical outcomes assessments (COAs)

In contrast to biomarkers, I anticipate that COAs will advance due to strong economic incentives. One of the goals for the DDT program is encouraging development of qualified DDTs to solve unmet needs. For example, reducing readmission to hospitals is a major thrust of the Affordable Care Act and certainly an unmet need. Two thousand hospitals in the U.S.A. face the prospect of losing $170 billion in funding in the next 10 years from reductions in Medicare, Medicaid, and the fiscal cliff, if their rate for readmissions within 30 days is excessive.

An article by Bill Malone discusses the readmission problem for patients suffering from heart failure or acute myocardial infarction (AMI).3 Cardiac biomarker assays (creatinine kinase-MB [CK-MB], troponins, and myoglobin), combined with electrocardiograms (ECGs), are useful in the diagnosis of a cardiac event, but have not shown prognostic utility.

There is a need for assays that clinicians can use to decide when a patient is ready for discharge, and provide useful prognostic guidance to help the patient avoid relapse and readmission. Several research groups are working on this problem. ST2 is a protein marker that shows promise as a cardiac biomarker. The level of ST2 is useful in assessing the severity of cardiac remodeling, including tissue fibrosis. Increasing levels indicate worsening risk of heart failure. ST2 has been approved by the FDA. Critical Diagnostics (San Diego, CA) markets an ST2 assay. Using a computer model, they claim the assay can reduce readmits for heart failure by 17%. Malone also notes that more prospective studies are still needed.

Malone also reports a retrospective study of changes in NT-proBNP (N-terminal prohormone of brain natriuretic peptide) concentration during treatment and predischarge also correlated with readmission and death within six months. Again, a prognostic evaluation is needed. Galectin-3 is another promising biomarker. Christopher deFilippi proposed a scoring system that identified patients with a 3× greater risk of 60-day readmission or death.4,5

Malone reports that hospitals are looking for other metrics that correlate with readmissions. For example, Intermountain Healthcare (Salt Lake City, UT) developed a multivariant prognostic signature for probability of readmission. Included are data from assays for hematocrit, red cell distribution width, mean platelet volume, mean corpuscular hemoglobin, total white blood cell count, sodium, potassium bicarbonate, creatinine, glucose, and calcium. The algorithm gave useful prognosis of patient mortality in 30 days, six months, and five years. However, acceptance by nonaffiliated hospitals seems to be slow. This is frustrating, since the information is often available in modern electronic records but is unused.

Retrospective studies with other metrics of patients originally admitted with heart problems show that other factors contribute to the prognostic signature of readmission. Frailty is one. Lack of support can overwhelm a frail patient, leading to relapse. A frail patient may not be able to follow instructions. Demographics—including ethnicity, family size, and cultural history—also appear to be relevant. The trick will be to develop treatment plans that improve patient outcomes.

Admittedly, counting readmissions within 30 days or another period is a gross metric encompassing a host of unrelated causes. P. Nelson Le, M.D. (InterSystems Corp., Cambridge, MA), reports that approximately 20% of Medicare patients are readmitted for the same diagnosis within 30 days, leading to an associated cost that is near $17 billion. Broadening the focus a bit to “directly related” to the original admission increases the number to only about 33%. These rough statistics show that “other factors” account for the majority of readmissions. For example, hospitalization due to “hit-by-truck” within 30 days counts as a readmit even if the original admission was a cardiac problem.

I think the FDA should be congratulated for recognizing and constructing a flexible framework that encourages, supports, and potentially qualifies novel solutions to current unmet problems with multivariant maladies.

Animal models

Developing useful prognostic signatures of therapeutic and toxic effects across different species is the third major program for DTT qualification. All too often, predictions of efficacy and safety based on screening compounds in rodents, zebra fish, pigs, dogs, etc., do not correlate with results observed in humans. Perhaps this is due to our limited knowledge of their systems biology. The FDA and drug industry in general would like to find biomarkers that correlate between species, assuming these markers exist.

For example, Dr. Donna Mendrick of the FDA’s Center for Translational Research recently discussed the problem of increasing adverse drug reactions (ADRs) associated with marketed drugs.1 In 2005, almost 90,000 serious ADRs were filed, including more than 10,000 deaths. Worse yet, the accuracy of predictions of toxicity is much less than 100%. Toxicity testing of drug candidates in animals misses about 30% of the ADRs in humans. Plus, rare, but serious, ADRs have been missed in the small number of humans tested in clinical trials.

Mendrick went on to focus on hepatotoxicity. The state-of-the-art needs improvement: Around half of the drugs that cause hepatotoxity in humans were not detected in preclinical animal testing. About 1% of hospitalized patients develop drug-induced liver injury (DILI). About 1000 drugs have been linked to liver injury. Existing markers (serum alanine aminotransferase [ALT] and bilirubin) are not very useful, especially as prognostic signatures.

Acetaminophen is a leading case in point. It is a common component of many OTC drugs. It alone accounts for about half of all acute liver failures in the U.S. Treatment options include N-acetylcystine, or liver transplant. About two-thirds of the patients have spontaneous remission, but it cannot be predicted which ones will recover without transplant.

One study of liver toxicity of acetaminophen involved oral gavage of male Sprague-Dawley rats. The urine was examined for miRNA-122, which appears to be liver-specific. Overall, 10 miRNAs were altered by treatments with high doses of acetaminophen. The miRNA levels in urine of human overdose patients is also being examined. The study is ongoing.

The sbv IMPROVER Challenge for 20136 will also focus on developing tools for trans-species drug safety and efficacy. This may be basic science today, but small bits of knowledge have a way of being used by creative people to solve real problems.

In summary, the DDT program is a good first step to connect the many dots between discovery of a possible biomarker to improved health care. Hopefully, scientists will make use of the programs.

References

  1. Mendrick, D.L. Biomarkers Working Group. Clin. Pharmacol. Ther.  2001, 69, 89–95; NCTR, FDA 10-2-2012.
  2. http://www.fda.gov/Drugs/DevelopmentApprovalProcess/DrugDevelopmentToolsQualificationProgram/ucm284076.htm.
  3. Malone, B. The race to reduce readmissions. Clin. Lab. News Apr  2013, 1A.
  4. Bettencourt, P.; Azvedo, A. et al. N-terminal-pro-brain natriuretic peptide predicts outcome after hospital discharge in heart failure patients. Circulation  2004, 110, 2168–74.
  5. Kansal, P.; de Boer, R.A. et al. Use of galectin-3 to create a simplified heart failure rehospitalization risk model. Circulation  2012, 126, A17458.
  6. Peitsch, M.C. Editor’s page. sbv IMPROVER: Species Translation Challenge open to the scientific community for submissions. Am. Lab. June/July  2013, 45(6), 6–8.

Robert L. Stevenson, Ph.D., is a Consultant and Editor of Separation Science for American Laboratory/Labcompare; e-mail: rlsteven@yahoo.com.

Comments