Novartis has identified a genetic sequence that may be a risk factor for serious liver injury in patients receiving Prexige (lumiracoxib), a finding that may signal a development path forward for the COX-2 inhibitor.

Preliminary results from genomic testing indicate that an extended Major Histocompatibility Complex Class II haplotype is associated with Prexige hepatotoxicity at a p-value of 6.3 x 10, Novartis Global Head of Translational Sciences Timothy Wright told a Drug Information Association/FDA meeting on the Critical Path Initiative Sept. 15.

If the final results bear out these findings, Novartis will meet with FDA to discuss what information needs to be collected to verify that the biomarker can identify those with a predisposition to liver toxicity caused by the drug, Wright said.

FDA issued a second "not approvable" letter for Prexige last year (¹ Also see "No Big Surprise For Novartis: Prexige Is “Not Approvable” At FDA" - Pink Sheet, 27 September, 2007.). The previous letter, in 2003, requested more information to facilitate a risk/benefit analysis. The company had sought an indication for osteoarthritic pain and acute pain.

Safety concerns have plagued the drug elsewhere. Prexige was removed from the market in Australia after two patients receiving the drug died from serious liver injury, and in a number of European countries after the European Medicines Agency launched a safety review (² Also see "Sales of Novartis’ Prexige Suspended in U.K., Germany, Austria After EU Announces Safety Review" - Pink Sheet, 19 November, 2007.).

Even if a clear biomarker is identified, it is unclear what impact that would have on the product's safety. Genetic predisposition is only one factor in whether rare, drug-induced liver toxicity occurs, so "we may not be able to predict with 100 percent certainty who would be excluded against that toxicity from a given drug," Wright acknowledged. But "we would be able to predict whether they have a sensitivity."

Regulatory Pathway A Challenge

A big challenge, he said, is how to get approval for the drug based on a safety biomarker.

CDER Director Janet Woodcock agreed. The Novartis findings show the need "to develop the pathway, the regulatory policy and the standards for biomarkers," she said at the meeting.

To do so, FDA must resolve a number of issues, she pointed out. "What does it take to reduce the uncertainty of having a rare catastrophic event? ... What kind of demonstration would be needed to make people feel comfortable that that probability was reduced? ... That's really the issue. How much confidence do you have around the reduction of the risk, and we're going to have to have long conversations about this."

At this point, the pathway forward may be "insurmountable, or at least not attractive," Wright said, "but we're willing to work ... to bring this one in." In the current safety climate, a COX-2 inhibitor, regardless of its risk profile, would seem to have extreme difficulties getting approved by FDA.

Even if Novartis demonstrates that the biomarker can identify patients who should not receive Prexige, a genetic test suitable for clinical practice must still be developed. This is doable since the company's studies used a test based on polymerase chain reactions, Wright told reporters, but it would have to be verified for marketing as a companion diagnostic.

In 2007, FDA changed labeling for the blood thinner warfarin to advise about genetic tests that can identify patients with an increased bleeding risk at higher doses of the drug (³ Also see "Warfarin Label Change Falls Short of Recommending Genetic Testing" - Pink Sheet, 20 August, 2007., p. 8). However, the agency did not require use of the tests, which are not yet widely available.

Findings From DNA Analysis

In seeking a way forward for Prexige, Novartis conducted a retrospective examination of DNA samples from a subset of the 18,000-patient Therapeutic Arthritis Research and Gastrointestinal Event Trial (TARGET) using a case-control design.

Initial testing was a genome scan of 42 patients who had received Prexige and had alanine aminotransferase and aspartate aminotransferase at levels ?5x the upper limit of normal. The control group consisted of 180 patients who received ibuprofen or naproxen during the trial.

The result was a "highly significant association" between a single nucleotide polymorphism in the MHC Class II region and the Prexige recipients, Wright reported.

An analysis of DNA from three Prexige-treated patients who met Hy's law criteria for drug-induced liver injury found that all three carried at least one copy of the top associated allele in the MHC Class II region, he said.

A second set of analyses involved 37 Prexige-treated patients with ALT or AST of 3-5xULN and 185 patients from the non-Prexige arms of TARGET. A "strong association" was found between the same SNP and the Prexige patients.

The secondary analysis of 22 ibuprofen and 10 naproxen recipients with ALT or AST ?3xULN and a control group found no evidence of SNPs associated with the high ALT or AST readings.

The haplotype associated with Prexige hepatotoxicity was found during a human leukocyte antigen identification study. The preliminary results show a 99.8 percent negative predictive value for AST or ALT >5x ULN, Wright reported.

In addition to implications of the Novartis studies for Prexige, Wright suggested that the firm's identification of MHC association with hepatotoxicity provides potential insights into drug-induced liver injury pathogenesis. These are:

A drug or metabolite may act as "haptent" binding to a self peptide that is presented by MHC Class II molecules and recognized as foreign.

The specificity for binding the "modified" peptide may occur in a limited number of MHC alleles so only those with the risk allele are at risk for drug-induced liver injury.

Identification of the risk haplotype in the MHC Class II region may explain why animal toxicology studies lack predictability for human hepatotoxicity.

A genetic pathway for drug-induced liver injury is one of the first areas of investigation for the Serious Adverse Events Consortium, which is developing a genotyping infrastructure and creating a database of DNA variants associated with serious adverse events (⁴ Also see "Serious Adverse Events Consortium Expects Data On Genetic Variants Shortly" - Pink Sheet, 11 August, 2008., p. 9).

- Cathy Dombrowski (c.dombrowski@elsevier.com)

Novartis Eyes Liver Safety Biomarker As Potential Path To Prexige Approval

More from Archive

More from Pink Sheet