Inside the FDA: Why Sanofi SA (ADR) (NYSE:SNY) IGlarLixi Was Delayed But Novo Nordisk A/S (ADR) (NYSE:NVO) IDegLira Wasn’t

Inside the FDA: Why Sanofi SA (ADR) (NYSE:SNY) IGlarLixi Was Delayed But Novo Nordisk A/S (ADR) (NYSE:NVO) IDegLira Wasn’t
novo nordisk logo

Update: Several hours after the initial publication of this report on September 2, Novo Nordisk announced that the FDA has extended the regulatory review for IDeglira, with a decision now expected in December. We will be presenting a separate analysis of this decision, what may have caused it, and its implications for both Novo Nordisk and Sanofi in the coming days. 

The race between Novo Nordisk A/S (ADR) (NYSE:NVO) and Sanofi SA (ADR) (NYSE:SNY) combination diabetes therapies is now down to the wire, with Novo pulling ahead after a delay was slapped on Sanofi. Novo’s combination drug IDeglira puts together liraglutide with insulin degludec. Sanofi’s IGlarLixi combines its own insulin glargine with lixisenatide.

This edition of Market Exclusive’s Inside The FDA series will delve into the primary transcripts to dissect exactly why Sanofi’s combo was delayed, whereas Novo Nordisk’s was not, as well as the implications for other companies in the diabetes space considering what happened in this case, and for other biotech companies pursuing combination drugs as well. We will begin with Novo Nordisk and IDegLira.

Most of this analysis will deal with IDegLira, and in the end we will come back to Sanofi in light of this and explain why, despite a positive panel recommendation for approval of IGlarLixi with only two no votes, the FDA still chose to delay its decision on the drug.

Novo Nordisk and IDegLira

As a brief introduction to the medications we are dealing with here, liraglutide is a subcutaneous GLP-1 drug whose activity is glucose-dependent. It generally leads to weight loss or at least weigh maintenance, and serves to activate the beta cells of the pancreas to produce insulin. Insulin degludec is a very long acting insulin analogue. IDegLira combines the two agents into a single injection.

The FDA panel discussion on this drug, which took place back in May, began with an FDA official commenting on the regulations involved in marketing combination drugs. As a reminder, FDA panelists are not bureaucrats. They are usually doctors or other professionals involved with patients in some way, invited by the FDA to vote. They do not generally have a bureaucratic mindset in terms of following regulatory guidelines robotically, if such a thing can be said.

Early on in the transcript of the discussion, before Novo Nordisk presented any data, an FDA official at the meeting noted that an improvement in glycemic control as measured by HbA1C over placebo is considered a valid surrogate of clinical benefit. HbA1C is basically a 3-month measure of blood glucose levels.

Clarifying this may seem prosaic, but that introduction is key for other diabetes-focused companies like Oramed Pharmaceuticals, Inc. (NASDAQ:ORMP), whose Phase II trial on an oral insulin pill used a placebo comparator. In other words, it is not necessary for a prospective oral insulin to go up against a subcutaneous therapy and display better results. For approval, it is enough to demonstrate efficacy over placebo.

FDA regulations regarding combination therapies like IDegLira are that the dosages have to target a significant part of the patient population for the disease, in this case diabetes. The FDA official raised the following question that the panel was instructed to consider before voting:Would it be reasonable, for example, to select an insulin-sensitive patient population who would only require low doses of the product since, at these doses, patients may be receiving a dose of 10 one of the component (sic) that is not contributing to the effect? Alternatively, in a population with severe insulin resistance, the dosage in the combination may be insufficient to provide long-term control because higher doses of insulin then can be delivered by the combination product will be required.

In other words, since the two drugs in this combination are fixed, it would not apply to the tail ends of the patient curve. That is, those early stage diabetics who only require low doses, and those late stage diabetics who require higher doses of one of the components than the combination can supply due to fixed ratios.

Novo understood this would be an issue beforehand, and so they tailored 5 different Phase III trials for different populations in order to test each separately and prove that the combination works for each population. The proposed populations are:

  1. Patients on neither GLP-1 nor insulin but who have failed other antidiabetic therapies like metformin
  2. Patients with inadequate glycemic control on GLP-1 alone
  3. Patients with inadequate glycemic control on insulin alone

The official further noted that individual titration (dosing) of each drug in the combination is not possible, and so it cannot apply in a blanket sense to the entire diabetes population. The official emphasized in outlining discussion points that combining a drug like insulin, with no theoretical maximum titration limit, with a GLP-1 drug like liraglutide with a very narrow dosing range, would present some disadvantages in terms of dosing flexibility. As we’ll see later on, Novo Nordisk was able to overcome these concerns with its data.

Novo Nordisk’s Presentation

The company began its presentation with the fact that the American Diabetes Association (ADA) recommends an HbA1C goal of below 7%. Between 60% and 80% of patients in two pivotal Phase III trials indeed achieved this goal. This is actually quite remarkable, because in the real world, Novo noted that only half of patients on GLP-1 drugs reach this goal, and only 30% of patients on insulin achieve it. The reduction lasted for up to 52 weeks, so this is long term benefit. The company then clarified that the combination drug is for treatment intensification only, and not for initial therapy.

Dr. Christopher Sorli, the principle trial investigator for these trials but not employed by Novo Nordisk, then made a key comment as follows, which also has broad implications for other companies in the diabetes space:

…long-term studies of intensive glucose control suggest that getting patients to goal early can have long-term benefits in reducing cardiovascular disease and mortality while other studies in patients with increased risk of cardiovascular disease suggest that intensive control may negatively impact mortality.

This comment is once again relevant for other companies in the space as it gives both Novo Nordisk and Oramed ammunition for the approval of the oral insulin pills they are both developing, just a factoid to keep in mind here. Meaning, even if any potential oral therapy is not as effective in controlling glucose as subcutaneous injections are, if it will encourage early stage patients to get on insulin earlier, there is more reason for these therapies to be approved.

Continuing on, the investigator notes that getting patients to take both therapies separately and concurrently would require more injections, which translates to less adherence and compliance. This may seem like it has nothing to do with either company because it is not their responsibility to get patients to comply, but as we will see in the reasoning for each vote, compliance has a lot to do with FDA approval.

Here we get to something seemingly inconsequential, but which was actually a pivotal factor in the delay of Sanofi’s IGlarLixi that was not suffered by IDegLira. That is, the titration algorithm, or how the patient calibrated his or her dose. The algorithm is very simple for IDegLira. The patient begins with a low dose, and increases by 2 units if blood sugar is above target, and decreases by 2 if below target. That’s it. (Keep this in mind.)

The results of one of the trials was a 1.44% decrease in A1C for insulin degludec alone, and a decrease of 1.28% for liraglutide alone. The combination saw a decrease of 1.91%, and was also associated with lower rates of hypoglycemia, or blood sugar levels that were too low. In general, it just gave better control. The mean A1C level attained after 26 weeks was 6.4%, which conforms to ADA targets mentioned above. More importantly, and what made IDegLira particularly impressive in the eyes of the panel, was that in order to achieve these levels, the IDegLira dose remained stable after week 12, but the insulin dose continued to increase from baseline. Meaning, the combination enabled patients to take lower insulin doses than otherwise with the same glycemic control.

Extrapolating from here, for any company pursuing any form of combination treatment for any disease that can show that the combination lowers the dose required, will probably impress any FDA panel.

But there was another factor that the panel was very impressed by. IDegLira patients exhibited no weight gain, while insulin-only patients did gain weight. In fact, over 52 weeks, patients on insulin alone gained weight, and patients on liraglutide alone, lost weight. It makes sense then that patients on the combination, in fact maintained weight. So not only did the dosages stabilize for the combination, but so did weight. IDegLira, however, was associated with weight gain compared to placebo and weight gain compared to GLP-1 alone.

By the end of Novo Nordisk’s presentation, the company was asked a question regarding what to do about patients who need more insulin than the maximum dose allows. Remember, the combination therapy does preclude simply upping the dose, because that also ups the dose of the other drug, which can be dangerous. The company answered that for those patients, individual components would be necessary.

Now, digressing for a moment to the sales question, would IDegLira cannibalize sales from Novo’s Tresiba and Victoza? To an extent, of course yes. However, as we have seen with the results, doses are stabilized with the combination, slowing down insulin resistance. That means patients can in theory live longer and spend more time on the combination before they exhibit insulin resistance, and if and when resistance kicks in and the maximum combination dose is not enough, they can then go to individual component therapy. Studies were not long enough to determine if and when insulin resistance kicks in on the combination therapy.

On net then, the sales implications are that the combination IDegLira will probably increase sales overall for the company by slowing the rate of insulin resistance, at which point patients will likely be forced to go on separate therapies anyway.

Here we get to an interesting part as it applies to any other company pursuing combination therapy for any drug for any disease. One panelist by the name of Dr. Meisel asked the following question:

And then if I may, a third question, are you considering these two agents to be additive or synergistic?

Additive means that the effect of taking the drugs together is the same as them being taken separately. Synergistic means that the effect of taking the drugs together is stronger than taking them separately. Novo’s Dr. Gough answered as follows:

We’re unable to say whether they have — how they’re working. The important point is that both components are contributing to the reduction in A1c across the dose range. But because our studies were treat-to-target studies, by the very nature of the design of the study, we cannot say how they’re working in terms of an additive effect. We can say there’s no evidence of a synergistic effect.

The significance of this answer is twofold. First, a company presenting to the FDA or an FDA panel should never speculate on something that the data cannot corroborate. Dr. Gough was being extra safe here rather than going out on a limb to state a hypothesis. That could have led to trouble because he could have been accused of making suppositions without evidence. Second, combination therapies do not necessarily have to have a synergistic effect. Additive is enough, again important for any company pursuing combination therapies.

Nevertheless, we can speculate as to what’s going on here, and that it may in fact be synergy. Since GLP-1 drugs are glucose-dependent and cause the pancreas to produce insulin in a glucose-dependent manner, what seems to be happening here is that the liraglutide component is obviating the need for more insulin, thereby slowing the insulin dosage escalation. The fact that everyone in the panel was sidestepping this elephant in the room just illustrates the overly cautious nature of these panels more than anything else.

Whether this interaction can technically be called synergy is a nomenclature question and not all that important. Keep in mind that FDA panelists are not FDA officials and usually not so bureaucratically-minded. Meaning, regarding their vote, many of them (not all) don’t particularly care if something meets a technical definition or not as long as they can see and understand an obvious effect, regardless of what it’s called. This, at least, is our opinion.

Dosing Questions

What seemed to be among the more pressing issues for the panelists was the dosing question. There are no units on the IDeglira pen that administers the drug. The panelists wanted units so patients can know how much of what they are putting into their bodies, and they wanted it clearly labeled as two drugs so patients don’t go adding insulin to IDegLira, which could result in an insulin overdose. In the end though, none of the panelists saw this as outweighing the benefits.

Reasoning Behind Each Vote And Implications

At the end of the meeting, each panelist was asked to provide a reason for their vote. The vote was unanimous in favor or approval. As we will see, the reasons provided did serve as a warning for the eventual delay of Sanofi’s IGlarLixi, which we’ll see even more of in the reasons provided for that combination drug. We won’t go into every vote, but key lines only.

One panelist, not a clinician but a patient representative, was particularly impressed with the idea of patient compliance due to one shot over two separate ones. Keep in mind that patient compliance was never actually tested. It is only an intuitive assumption, and yet it was key in inspiring a yes vote.

Another panelist, one Dr. Charlie Stanley, a pediatric diabetologist, also said something key for any future oral insulin product:

I voted yes. I mean obviously, as a pediatric diabetologist, I don’t have a 1 lot of experience with these agents for type 2 diabetes. I would love to see a 12-month plus weight data for this combination product because we really may be entering an era where that could have tremendous application with an insulin that perhaps even beyond 12 months, that the patients don’t gain weight the way people do on other insulin types of regimens.

The first thing to note about this is that children with diabetes, early stage of course, are not generally prescribed insulin or GLP-1. This is why he has little experience with the drugs. Second, he is particularly excited about the lack of weight gain from the combination versus insulin alone. If combining the therapies prevents weight gain, he sees this as a “new era”. This is significant because oral insulin would take care of both these issues and indeed usher in a new era.

First, it would allow insulin to be prescribed to early stage diabetic children due to the relative ease of taking a pill over injection. Second, oral insulin is theorized to prevent weight gain as well, some evidence of which we have seen in Oramed’s Phase II data for its own oral insulin formulation.

In other words, if it can be shown for either Novo Nordisk or Oramed or both that an oral insulin enables insulin therapy at a younger age and prevents weight gain, any FDA panel is likely to be very impressed, whatever the other drawbacks of an oral insulin may be.

Most of the panelists, despite their yes votes, expressed concerns over the lack of units and dosing questions. Those concerns were simply overridden by the good data. This, as we’ll see ahead, was a red flag for Sanofi’s IGlarLixi.

Why IGlarLixi Was Delayed

The FDA panel vote for IGlarLixi  was overwhelmingly in favor of approval by a vote of 12 to 2. This is because, just like IDegLira, the data for IGlarLixi was very impressive with similar benefits. Nevertheless, nearly every single panelist expressed serious concerns over the delivery device. First, in an effort to leapfrog Novo Nordisk and make their product better, Sanofi decided to use two different pens, one for a low dose combo, and another for a high dose. They may have thought that by having two pens it would open up more patient populations, as those with higher insulin resistance would be able to use the higher dose pen. This would not be an option with IDegLira of course.

Sanofi was perhaps too ambitious here.

What ended up happening was that the panelists, even the ones that voted for approval, were universally concerned about the pen due to the potential for dosage confusion. Panelists raised these same concerns with IDegLira, but since there was only one delivery pen, the concerns were not as potent as with IGlarLixi. The reasons for each vote on IGlarLixi are given on page 380 on this document where you will find that nearly every panelist expressed the same concern over the delivery pen confusion potential.

Particularly, the concerns were based on a Human Factors Study that showed that 1 out of 15 pharmacists erred in regard to the dosage and gave a patient the wrong pen. Further, 1 out of 45 nurses had trouble identifying which pen corresponded to which dose. Had Sanofi been a little less ambitious and stuck with one pen only, the delay would probably never have happened.

As it is now, the question of whether IGlarLixi gets approved depends on whether there is a quick fix to this problem. The best case scenario is that Sanofi changes the color on the pens or something to make it more obvious which is which and the FDA OK’s it. This is problematic, because another Human Factors study may be required to corroborate the effect of any change, and that will take time. Sanofi doesn’t have that data yet.

The worst case scenario is that Sanofi simply nixes one of the pens for now and sticks with the lower dose pen, dealing with the higher dose pen later when more Human Factor studies can be conducted. In such a case it is very likely that IGlarLixi with only one pen will be approved.


From these documents, the implications for other diabetes companies are as follows. First, dosing confusion is crucial. Companies need to make sure that their delivery products are clear, and units are understood, and also that one unit does not apply to two different drugs. These are actually mutually contradictory, so best to stick with no units it seems.

There is a Catch 22 here because IDegLira has no units and looks like it will be approved without additional delay, but IGlarLixi has units for the insulin component that can be confused with the GLP-1 component, and this was considered a serious problem by panelists as well. Best to keep the units unlabeled, and stick to a simple device without trying to get too clever.

Second, FDA panelists for diabetes drugs are most impressed by lack of weight gain, patient compliance, and ease of administration. The implications for oral insulin being developed by both Novo Nordisk and Oramed Pharmaceuticals are positive, because these therapies may cause less weight gain because the insulin is delivered to the liver through the portal vein, and will probably increase compliance. Whether or not they are as effective as subcutaneous insulin appears at this point to be less of an issue, as long as they are somewhat effective.

Third, counting the votes by itself is not sufficient to determine what the FDA will actually do in response. As we can see here, the vote for IGlarLixi was 12-2, and still the FDA delayed its review at great cost to Sanofi. The reasoning behind each vote is critical, and the writing was on the wall.

Fourth, for any company pursuing a combination drug in general, synergy does not seem to be required. An additive effect appears to be sufficient.