Takeda halts development of GPR40 agonist, TAK-875 (fasiglifam) due to liver safety concerns – December 27, 2013

In disappointing news for Takeda, who has suffered setback after setback in diabetes on the regulatory front, the company announced today that it has voluntarily discontinued development of its first-in-class oral GPR40 agonist, TAK-875 (fasiglifam; for type 2 diabetes) due to liver safety concerns. The candidate was in phase 3 in the US, EU, and Japan, with a major (n=5,000) cardiovascular outcomes trial (CVOT) ongoing, along with 13 other shorter-term trials. Management had forecast a Japanese regulatory decision in FY2015 (calendar 2Q15-1Q16), with US and EU decisions following in FY2016-17 (calendar 2Q16-1Q18). It is unclear whether any other GPR40 agonists are in serious development (details below).

We are not surprised about Takeda’s decision, for several reasons we relay in this report, but primarily because any new drug that has even a small signal for hepatotoxicity (elevated liver function tests in some, as we understand it) will cause significant concern even in the absence of overt clinical evidence of liver damage. With a much more safety-sensitized FDA than years past – per the TZD-class journey, pancreatitis associated with incretins (even though unfounded) and more recent SGLT2 scrutiny –pharmaceutical companies have a lower threshold to halt development programs. Overall, we suspect that in an environment with a dozen fairly safe classes of drugs (if not most of them very patient friendly).

While mechanistically, GPR40 agonism has been billed as a “better SFU,” and was characterized as “revolutionary” by Takeda management as recently as the company’s F4Q12 update, we assume that the off-target toxicity that Takeda saw with a few thousand exposed was just too much risk for the company to take on. As background, TAK-875 was believed to improve glycemic control by stimulating insulin release in a glucose-dependent manner, which reduces the risk of hypoglycemia that accompanies sulfonylurea-stimulated insulin release (which occurs in a glucose-independent manner, so is often associated with hypoglycemia). Another of TAK-875’s benefits would have been that it would not have required dose adjustment in patients with renal impairment and did not seem to be associated with the low durability of SFUs. Takeda reported early phase 3 results for TAK-875 in May 2013 at the Japan Diabetes Society meeting showing that at 24 weeks, the 25 mg and 50 mg doses provided statistically significantly greater A1c reductions compared to placebo (placebo-adjusted decreases of 0.75% and 1.01%, respectively) from a low A1c baseline of ~7.8%. Detailed safety data were not released at the time, although TAK-875 was associated with a very low 2% incidence of hypoglycemia compared to 19% with SFU glimepiride. Phase 3 data on CV risk markers or weight have not been disclosed, although phase 2 data did not show any clear effect on weight in either direction.

Big picture, we wonder if Takeda would have halted development if this same safety signal had arisen five or ten years ago, in a very different regulatory and reimbursement environment. We are certainly sure they would have if the safety issues were serious, which we presume they are, although given that no data has been shared, this makes speculation difficult. Given that TAK-875’s CVOT was in the relatively early stages, we wonder if the cost of the CVOT, the unpredictable regulatory environment, the increasingly pressured reimbursement environment, and crowded diabetes competitive landscape collectively made further investment in a candidate with a potential safety signal financially untenable. The bar for new diabetes drugs is at a new high, and continues to get higher. Given that there are so many drug classes available (especially after the introduction of DPP-4 inhibitors and SGLT-2 inhibitors), new classes would need to meet a higher standard, demonstrating benefit beyond A1c lowering. Even if the safety signal Takeda saw was not itself a deal-breaker (which we believe it was, in this case), we imagine it might have seen the continued development of the candidate as not worth the risk and continued investment.

We still believe that industry and regulators may be able to do more to identify subpopulations in which drugs are safe and effective rather than ruling out approvals for everyone – as the ADA’s Chief Scientific and Medical Officer, Dr. Robert Ratner, often says, “The fact that a therapeutic agent is not appropriate for everyone does not mean that it is not appropriate for anyone.” Nonetheless, it is often difficult to identify the subgroups that would benefit most or suffer the least harm from a drug when the drug’s trials were not designed to do so, and it is also difficult to design a trial to identify such subgroups when one does not yet know what to look for.

Going forward, for new drugs to be successful, there must be some other signal in early studies besides lowering of A1C. We need to see improvements of some surrogate (even weight) to push the company to complete the registration studies and CVOTs.

This news certainly rounds out a year with more than its fair share of high-profile disappointments in diabetes drug development. As we recently reported, BMS is actually completely extricating itself from diabetes, citing significant recent increases in competition and price pressures, as well as relatively slow uptake of new medicines, as reasons why the diabetes market has become less attractive over the past year. In addition, Roche/Genentech is out of the picture on diabetes drugs, with the failure of its dual PPAR-alpha/gamma agonist aleglitazar. While Takeda has certainly been very committed to the global diabetes community about developing new therapies for patients, and while the press release seems to spell out that the company is in diabetes for the long run, the delays associated with alogliptin in 2008 and 2011 would certainly give any company pause on the regulatory front alone – especially since the FDA delays are hard to justify today. We hope that the discontinuation of TAK-875, taken together with the company’s recent discontinuation of two other diabetes-related compounds (the phase 1 glucokinase activator TAK-428 and the phase 2 diabetic neuropathy candidate TAK-428), is not an indication of the company’s disengagement with diabetes as a whole.

• Takeda had invested a great deal in TAK-875 and there was a lot of hope for this compound as monotherapy, combination therapy, and (given its clean profile) as part of a fixed dose combination. According to ClinicalTrials.gov, the phase 3 program consisted of at least 13 trials at the time the candidate was discontinued; these trials included a 5,000-person CVOT that had begun in June 2012 and was still recruiting patients. The CVOT was scheduled to complete in December 2018.
  
  • There was potential for Takeda to formulate a range of fixed dose combinations, were TAK-875 have been successful. These range from combinations with metformin or Takeda’s DPP-4 inhibitor Nesina (alogliptin) to combinations with SGLT-2 inhibitors, or other drugs in development. We had been excited about its potential use in a fixed dose combination and in the very strong Takeda sales force having another agent.
• Takeda did not provide any details or data on the exact safety signal that led to its decision, but from what we understand, there were aberrations in liver function tests. Cases of liver injury or failure are on the more severe end of the spectrum; these are possible, but less likely. We hope that Takeda eventually releases the data, although it may be a while before it is ever made public — as a recent example, Reata waited over a year between the termination of its phase 3 BEACON trial for bardoxolone methyl and the publication of the study’s results (see our report on those results and the NEJM publication).
  
  • Whatever liver safety concerns emerged in Takeda’s phase 3 program were not evident in the company’s phase 2 data. The 2012 Lancet publication of the phase 2 results stated that there were differences between the TAK-875 and placebo arms in terms of the percentage of patients with alanine aminotransferase, aspartate aminotransferase, total bilirubin, and creatinine kinase increases during treatment, although no specific data were provided (read our report on those results). That being said, safety signals involving relatively rare events often only emerge in phase 3.  
• Had TAK-875 had a clean safety profile and progressed to approval it would have competed most directly with DPP-4 inhibitors or SGLT-2 inhibitors as oral second/third-line therapies for monotherapy, and it may have had a slight advantage in A1c-lowering efficacy. We were, as noted, very excited as a new class, to see how it would have combined with either or both of these agents. It also would have competed against SFUs, which are challenging drugs due to hypoglycemia concerns and weight gain (along with potential cardiovascular and/or mortality risk) but are attractive from a cost standpoint. We await the CAROLINA and GRADE randomized control trials with great interest as they may be able to produce more conclusive data on SFUs’ potential safety risks — negative data on SFUs from either or both of these studies would be a blow against the class. While some would say a negative trial result would really open doors for “smarter” secretagogues, this all depends, of course, on development appetite, which certainly appears to be declining as far as Big Pharma goes (this may not be true with earlier stage compounds).
• It is unclear whether any other GPR40 agonists are in serious development. As of June 2013, BMS had listed a GPR40 agonist in “exploratory development” (somewhere between discovery and phase 2), although BMS has since transitioned to listing only candidates in clinical development on its online pipeline, and the GPR40 agonist is no longer listed. It is unclear whether AZ has picked up this candidate in its acquisition of BMS. Merck has also previously expressed interest in the mechanism, but to our knowledge, has not disclosed any active work on specific candidates. A major question, in our minds, is whether the likely hepatotoxicity signal is limited to TAK-875 or whether it is a GPR40 agonist class effect — we fear that the latter possibility may dissuade AZ, Merck, or any other companies that might be working on a GPR40 agonist to discontinue development.  
• The bar for new diabetes therapies is rising. Since 2005, when the first DPP-4 inhibitor was approved (Merck’s Januvia), branded therapies have been available that already have low rates of hypoglycemia, weight-neutrality or –loss, and relatively benign side effect profiles, combined with strong A1c drops. Arguably, few to no agents have a strong composite on all fronts, including a low hassle factor.
  
  • Given the higher (and ever-rising) expectations for diabetes drugs, some benefit beyond A1c lowering will likely be needed for companies to push companies to develop a candidate through registration studies and cardiovascular outcomes trials. As an example, incretin therapies were able to differentiate themselves through their weight neutrality or weight benefit (among other factors). We have not seen phase 3 data on TAK-875’s effect on weight, but phase 2 data presented at this year’s IDF showed no clear dose-dependent effect on weight (see page 9 of our IDF 2013 Day #2 Report). The hypoglycemia benefit associated with TAK-875 would have also provided an edge against SFUs but probably would not against DPP-4 inhibitors or SGLT-2 inhibitors. Given that there are so many effectively and relatively safe diabetes drug classes on the market, we could understand why drug manufacturers feel less pressure to push forward with therapies that may have significant side effects, although from a patient perspective, it really drives us crazy to hear this phrase (“twelve effective and safe classes”) since virtually all the classes to date have some shortcomings and we believe that more choices of agents would be beneficial for patients given the problems to date:
    • SFUs have hypoglycemia and weight gain and association with CV risk
    • Metformin is associated with nausea and can’t be taken by all patients (especially with those who have risk of lactic acidosis and/or renal problems)
    • TZDs are associated with weight gain, edema, and congestive heart failure;
    • DPP-4 inhibitors are associated with lower A1c drop (though this can be  addressed with combination therapy)
    • GLP-1 is associated with nausea for some patients
    • Insulin is associated with hypoglycemia and weight gain.
  • As the highly regarded researcher and current Chief Scientific Officer of the ADA, Dr. Robert Ratner, has said multiple times, “We still need better medications to assist all people with diabetes in achieving normal glucose concentrations without the risk of weight gain or hypoglycemia.”
  • The regulatory landscape alone is increasingly more knee-jerk on most potential safety issues, due in large part to the incretin-pancreatitis scare and concerns regarding TZDs and cardiovascular risk and cancers. Hepatotoxicity signals are particularly potent potential drug-killers, even if such signals stem from liver function tests and are not accompanied by overt clinical evidence of liver damage. Warner Lambert’s (later purchased by Pfizer) Rezulin (troglitazone) was associated with liver failure and had to be pulled from the market in 2000. A single case of what was then diagnosed as likely drug-induced liver injury was seen as a cause for concern during the initial FDA Advisory Committee for AZ’s SGLT-2 inhibitor Forxiga (dapagliflozin), along with a non-significant bladder cancer signal. Dapagliflozin’s sponsors were able to allay these concerns during the drug’s second FDA Advisory Committee meeting earlier this month, showing that the drug was unlikely to have caused most of the bladder cancer cases, and that the case of liver injury was due to other causes. However, the fact that the drug was almost completely derailed by these potential signals that did not result in major concerns is worrying for companies considering whether to move forward with the development of diabetes pharmacotherapies. This is particularly true in the era of major diabetes associations pushing for individualized therapy.
• Takeda has been plagued by setback after setback in diabetes, e.g., TZD Actos’ fall from glory due to the bladder cancer scare; nearly five years of delays for DPP-4 inhibitor alogliptin, along with two recent decisions to end development for two earlier stage drugs.
  
  • Looking back, the DPP-4 inhibitor delay seems unwarranted. As background, the company submitted its DPP-4 inhibitor Nesina (alogliptin) in the US in January 2008 before the FDA’s 2008 CV guidance had been implemented, but it did not ultimately receive an approval until 2013. It received a CRL in June 2009 calling for a CVOT (EXAMINE, whose results have now been released), another delay in 2011 when the FDA extended its PDUFA date by three months, another CRL in April 2012 asking for unspecified additional data, and an ultimate approval in January 2013.
• Takeda’s diabetes and obesity pipeline has two notable later-stage candidates with near-term launch potential: Contrave for obesity (in partnership with Orexigen) that has been submitted to the FDA and trelagliptin, a once-weekly DPP-4 inhibitor in phase 2. However, the news of TAK-875’s discontinuation disappointingly follows F2Q13’s termination of TAK-329 (a phase 1 glucokinase activator) and TAK-428 (a phase 2 candidate for diabetic neuropathy).

Table 1: Takeda’s remaining diabetes pipeline as of December 2013

Background Reading on TAK-875

• First phase 3 results shared at Japan Diabetes Society Meeting (May 16, 2013; press release): At 24 weeks, fasiglifam 25 mg and 50 mg provided statistically significantly greater A1c reductions compared to placebo (placebo-adjusted decreases of -0.75% and -1.01%, respectively). Takeda did not give detailed safety information, stating only that drug-related adverse events were mild to moderate and that the three treatment groups saw similar rates of adverse events, including hypoglycemia. The most frequently reported adverse events (≥5%) were nasopharyngitis and upper respiratory tract inflammation.
• TAK-875 enters phase 3 trials: It was a big turning point when Takeda announced that TAK-875 would enter phase 3 trials, since Takeda had been at work for some time on a drug to take the place of TZD Actos, which generated $4 billion in revenue at its peak. While alogliptin could have had an opportunity like this were it not for regulatory delays, being 4^th to market in DPP-4 inhibitor-land made this far less likely.
• Phase 2 results first presented at ADA 2011 (page three here) and subsequently at EASD 2011 (page 23 here). More phase 2 results were presented at this year’s IDF (page nine here).
• Publication of phase 2 results (February 27, 2012; Burant et al., Lancet). The TAK-875 50 mg dose provided the greatest A1c reduction, which was statistically significantly different from placebo (placebo-adjusted decreased of 0.99%) – this result is consistent with the A1c reduction observed in the phase 3 trial. Similar to the phase 3 results, the incidence of adverse events (including hypoglycemia) was similar between the TAK-875 groups and the placebo group. Results from this trial were previously reported at ADA 2011 and EASD 2011.

Close Concerns Questions

1. What were the nature of the liver safety issues that arose?
2. Given the drive toward individualized therapy, could a subpopulation be identified that was not affected by these liver issues?
3. Had the same safety signal arisen five or ten years ago prior to today’s climate of unpredictable regulators and increasingly pressured reimbursement, would the same decision to halt the drug have been less likely? More likely?
4. Given all of the challenges Takeda has encountered in diabetes, where might its management and/or board of directors estimate the ROI on diabetes drug development?
5. Since the “benefit” half of the risk/benefit evaluation is harder to prove, given the number of effective options on the market, will companies become more selective about the number and/or quality of candidates they select to progress into full-scale clinical development?

-- by Jessica Dong, Adam Brown, Manu Venkat, and Kelly Close

Editor’s Note: In our original report, published on December 27, 2013, we stated in error that Merck’s Januvia (sitagliptin) was the first incretin therapy on the market. We intended to state that Januvia was the first DPP-4 inhibitor on the market. We apologize for any confusion this may have caused.