This story originally ran on May 10 and has been updated to include additional information from the FDA.
In a paper published May 10 in the online edition of Nature Biotechnology, researchers from the Predictive Safety Testing Consortium’s Nephrotoxicity Working Group detailed the process behind the group's 2007 submission of urinary renal safety biomarkers for Food and Drug Administration and European Medicines Agency review.
The submission process, as described in the paper, provides useful clarification on current FDA and EMEA biomarker review procedures as well as a roadmap for how future biomarker studies may need to proceed in order to gain regulatory approval.
The "progressive" qualification process described in the paper, which takes the proposed context for the biomarker's use into account during review, is one that the FDA has been developing with regard to biomarker research over the last four years, Federico Goodsaid, associate director for operations in genomics at FDA, told ProteoMonitor.
The agency, which has 10 different biomarker projects in some stage of the qualification pipeline, has formalized the progressive qualification process over the last year and will be releasing a draft guidance detailing the formal steps of the process this July, Goodsaid said. Also this summer, the International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use will be releasing guidelines covering the structure, content, and format of biomarker qualification submissions — a release that Goodsaid said will "dovetail very nicely" with FDA's draft guidance.
The PSTC's data resulted in FDA and EMEA approval in 2008 for the limited use of five protein biomarkers — kidney injury molecule-1, albumin, total protein, β2-microglobulin, and cystatin C — in nonclinical and clinical drug development, with clinical approval pending for two other protein biomarkers — clusterin and trefoil factor 3. As such, the submission represented a milestone for the use of biomarkers in drug development, the authors said in the Nature Biotechnology article.
The lack of reliable tests to predict nephrotoxicity in humans is a significant impediment to the development and application of new drugs. Traditionally, the FDA and EMEA have required drug companies to submit the results of two blood tests — blood urea nitrogen and serum creatinine — to evaluate nephrotoxicity. These tests, however, can only detect kidney damage a week after it has started to occur and cannot indicate precisely in what part of the kidney damage is located.
The biomarker tests submitted by the PSTC detect kidney damage within hours of its occurrence and can identify which parts of the kidney are affected. Such detection could allow for the development of drugs previously considered non-viable because of the inability of current tests to detect and monitor early renal damage.
Specifically, the researchers made three claims regarding the seven biomarkers submitted for review: Urinary KIM-1, CLU, and albumin can each outperform and add information to BUN and SCr assays as early diagnostic biomarkers of drug-induced acute kidney tubular alterations in rat toxicology studies. Urinary TFF3 can add information to BUN and SCr assays as an early diagnostic biomarker of drug-induced acute kidney tubular alterations in rat toxicology studies. And total urinary protein, cystatin C, and β2-microglobulin can each outperform SCr assays and add information to BUN and SCr assays as early diagnostic biomarkers in rat toxicology studies of acute drug-induced glomerular alterations or damage resulting in impairment of kidney tubular reabsorption.
Fundamental to the success of the PSTC’s submission was the proposal of a “rolling biomarker qualification concept,” a framework that allowed researchers to submit the biomarkers for review with the understanding that regulators would evaluate their validity solely for the uses supported by the submitted data.
“What we realized was that whatever you do, you cannot validate a biomarker for everything,” Frank Dieterle, project manager at Novartis Pharma AG and lead author on the Nature Biotechnology paper, told ProteoMonitor in an interview.
“You would need to cover any potential drug, any potential clinical situation to have universal qualification of a biomarker — and that’s not possible. So we started with the term 'qualification,' which means that you qualify the use of a biomarker for a certain limited purpose.”
[ pagebreak ]
Under this framework, researchers were able to gain regulatory approval for the biomarkers’ use within a narrow context, which could then be expanded and modified as additional data became available. Presently, the PSTC biomarkers are qualified for voluntary use as additional evidence of renal damage in rats in conjunction with traditional tests and may also be used in clinical trials on a case-by-case basis to gather additional data regarding their usefulness for monitoring drug-induced kidney damage in humans.
The PSTC submission also established a series of best practices for qualifying renal biomarkers — specifically, the use of creatinine concentrations to normalize samples of different urine dilutions, the use of data from toxicity studies to support biomarker qualification, the appropriate use of multiplicity corrections, and the reporting of data using receiver operating characteristic curves.
Most crucial, Dieterle said, was creating a standardized histopathology lexicon that researchers could use across labs and institutions. Because histopathology was the “gold standard” against which biomarker data was compared, it was essential that researchers and regulators agreed on standards of interpretation.
Standards of interpretation aside, histopathology evaluation practice was one of the primary points of debate between regulators and the PSTC. Specifically, regulators were concerned that, because the PSTC’s histopathology readings were not done as blind evaluations, bias could have been introduced at this point in the study. However, Dieterle said, blind pathology readings are not a standard in the drug industry, and adopting them would require significant additional resources.
“We want to use the samples and studies that we are using anyway for drug development,” he said. “Otherwise we would waste a lot of resources. It could even be unethical to run additional animal studies if you could use the current ones.”
One potential way to resolve the issue, Dieterle said, would be to perform limited blind studies that could then be compared to traditional histopathology readings to test them for signs of bias.
Goodsaid declined to provide specifics on FDA's position regarding standards for histopathology readings in biomarker qualification studies, saying that FDA would release a draft guidance of histopathology standards this summer around the same time as the broader draft guidance on biomarker qualification. He did note, however, that histopathology standards should be as "objective" as possible.
"The problem is that in biomarker qualification you have to make sure that you have an objective measurement that serves as your point of reference," he said. "In the case of non-clinical testing you are often using histopathology as that standard, and you want to make sure that that standard is objective as possible. That's what this discussion is about, and the guidance will go into that discussion."
Regulators also raised questions about the PSTC’s submission of an “exclusion analysis” dataset that left out results from nephrotoxicant-treated animals without positive composite kidney histopathology scores — thereby discarding possible false-positive results. According to Dieterle, excluding these animals avoided potential misinterpretation in the case that the biomarkers were operating in a prodromal manner. However, regulators preferred drawing conclusions from the full dataset.
“One way forward is that we do a series of experiments where we really prove that our biomarkers on a molecular level are present earlier than pathology,” Dieterle said.
Such experiments would include time-course studies to establish whether biomarker signals precede histopathological evidence of renal damage and expanded histopathology studies to determine if biomarker changes might be related to histopathological changes not detected by standard light microscopy.
The PSTC is also consulting with regulators on the development of studies to provide data on biomarker baseline values across various subject populations, the goal being to gain approval for the biomarkers in a general clinical application context. Also under discussion is the design of studies using drugs with known nephrotoxicity — including intravenous contrast medium, gentamicin, and cisplatin (Platinol) — to be run concurrently with clinical assay development and validation.
Looking beyond the case of these particular biomarkers, the PSTC study represents a significant advance for models of biomarker qualification more generally, according to the authors of the Nature Biotechnology article
“A qualification framework existed only on paper for the FDA, and for the EMEA didn’t exist at all when we started to do this research,” Dieterle said.