Memorandum

Sanofi Investor Call at American Diabetes Association 2014 – Toujeo (U300 insulin glargine) has been submitted; biosimilar Humalog (insulin lispro) advancing to phase 3 in 2H14 – June 19, 2014

Executive Highlights

  • Toujeo (U300 insulin glargine) NDA filed in the US and Europe; analysts question robustness of hypoglycemia data in phase 3 EDITION program.
  • Identity of one compound in the phase 1 biosimilar insulin program revealed: SAR342434 is insulin lispro (Lilly’s Humalog) and will advance to phase 3 in 2H14.
  • Management highlighted new partnership with Medtronic as an example of how Sanofi is working toward being a leader in integrating care for patients.

Sanofi held an investor call on Monday June 16 to discuss data presented on its diabetes portfolio at ADA 2014. The call focused on phase 3 registrational data for Sanofi’s new U300 insulin glargine formulation, Toujeo, which management announced has been submitted in the US and EU. Here we provide our top five highlights from the call, followed by the transcribed question and answer session and an Appendix detailing hypoglycemia findings from Toujeo’s phase 3 EDITION program (a subject of intense questioning during Q&A). Speakers on the call were Pascale Witz, Executive Vice President of Global Divisions & Strategic Development; Pierre Chancel, Senior Vice President of Diabetes; Dr. Ricardo Perfetti, Senior Medical Officer of Diabetes; and Dr. Matthew Riddle, Professor of Medicine at Oregon Health & Science University.

Top Five Highlights

1. Management announced that Toujeo has been submitted to the US FDA and reiterated that Toujeo had also recently been submitted to the EMA (as announced two days earlier in a press release about Toujeo data presented at ADA). With a standard 12-month review cycle, Toujeo could be the first “second-gen” insulin glargine formulation to the US or European market, with Lilly’s insulin glargine formulation delayed by Sanofi’s lawsuit, potentially until 2016. During Q&A, management noted that there have been no cardiovascular differences observed between Toujeo and Lantus and that Toujeo’s ultimate label could draw upon the safety claims made in Lantus’ label (e.g., presumably the CV safety labeling conferred by ORIGIN results. Management declined to make any pricing comments, but stated fairly emphatically that ensuring good access would be a high priority. They also stressed that converting Lantus patients converting from Lantus to Toujeo was a very high priority – presumably this is patients who are not at goal.  We aren’t sure how to interpret the pricing discussion exactly since there have been multiple price increases to insulin in 2014 in the US in particular; access to insulin analogs has become more challenging in some areas of the US like the Pacific Northwest in particular. We don’t have a good handle on reimbursement changes broadly speaking but know that in many parts of the EU, access to analogs is a major issue. At the same time, insulin is still less expensive for many type 2 patients than would be taking multiple orals or taking insulin and orals. We also believe in pricing premiums as this is what enables continued R&D investment, which itself enables medicines to be created and then become generic, which is very important for patients globally, especially in third-world countries. To be sure, pricing is hardly an issue with simple perspectives.

2. During the call, management presented results of a meta-analysis of EDITION 1, EDITION 2, and EDITION 3 (the three global type 2 diabetes studies for Toujeo, Sanofi’s U300 insulin glargine), which found that Toujeo and Sanofi’s gold-standard (for now) Lantus had equivalent A1c-lowering effects after six months (~1.1% reduction from 8.3% baseline in both arms). Meanwhile, Toujeo had lower rates of nocturnal and anytime hypoglycemia than Lantus (31% and 14% relative rate reduction, respectively) and less weight gain than Lantus (0.28 kg [0.62 lb] less). The percentage of individuals affected by one hypoglycemic event (confirmed ≤70 mg/dl or severe) was statistically significantly lower on Toujeo from baseline to six months (9% relative rate reduction for anytime hypoglycemia and 25% relative rate reduction for nocturnal hypoglycemia); the statistical significance also held true when looking online at baseline to week eight or week nine to six months. Results in of the Japanese type 2 diabetes study, EDITION JP2, were even more pronounced, with Toujeo providing a 55% and 36% relative rate reduction, respectively, in nocturnal and anytime hypoglycemia. We imagine Sanofi is very glad to have these results, given the notable success of Novo Nordisk’s degludec in Japan – we understand 20% of those on basal insulin are taking degludec. 

3. In Q&A, analysts questioned whether the hypoglycemia claims from the EDITION studies would be enough to convince regulators that Toujeo actually has a clinically meaningful hypoglycemia benefit. Much of the analysts’ skepticism is likely fueled by the hard stance the FDA took against Novo Nordisk’s hypoglycemia claims for its next-generation basal insulin Tresiba (for details see Tresiba’s Advisory Committee meeting). Regarding Toujeo specifically, many of the individual EDITION trials failed to demonstrate statistically significantly better hypoglycemia when the time period of the analysis varied (e.g., looking at the whole six-month duration of the trial or looking only from baseline to week eight or three months to six months; see table below in the Appendix). Analysts questioned whether Sanofi’s slicing and dicing of the timing in this way was legitimate, especially when the time period in the analysis was not part of the prespecified hypoglycemia endpoint. Sanofi’s original intention with separating out weeks one through eight from week nine through six months was to isolate the comparison to the period of time after titration had been optimized, the company said. Sanofi management maintained that even when statistical significance was not reached, the trend was always in the same direction no matter how hypoglycemia was defined and that this consistency was important to consider. We do think this will be perhaps a bigger deal when combining with GLP-1; we continue to believe it’s not about what advantages new insulins show alone but the advantages they are able to ultimately show in combination therapy. To be developed, of course, they need to be approved. 

4. Another notable happening during the call was that Sanofi confirmed that one of its biosimilar insulins in development is a version of Lilly’s rapid-acting analog, Humalog (insulin lispro) and that this compound (SAR342434) will advance to phase 3 in 2H14. SAR342434 has completed phase 1 in which it demonstrated similar activity and exposure as Humalog. Management disclosed that the phase 3 program will recruit ~1,000 patients for two studies: one in type 1 diabetes and one in type 2 diabetes, both using Humalog as the direct comparator. As a reminder, management disclosed during its 4Q12 financial update that it had two biosimilar insulins in phase 1 that were not based on any of Sanofi’s own molecules, suggesting that they were biosimilar versions of Lilly’s Humalog and Novo Nordisk’s Novolog (insulin aspart). As we understand it, the formulation patents for Humalog expired in 2013, and Novolog’s will expire in 2017.

5. Management commented on its announcement to partner with Medtronic (see item #5 under “Top Five Technology Highlights” in ADA 2014 Day #2), highlighting the partnership as an example of how Sanofi is leading the way toward integrating care for patients. Management remarked that it would take an “open innovation model” whose goal would be to think of a solution to help patients not have to worry about their disease 24/7.  That sounds great from our view and was reinforced in the Q&A discussion, where Sanofi management emphasized they will be considering a wide range of delivery technologies. That makes sense to us given the sizable percentage of patients not at their glycemic targets and the relatively smaller percentage of patients on insulin at all. We believe Sanofi is going to go a long way in helping patients with type 2 in particular when they use CGM and are able to get more sound therapeutic decisions (these decisions should be easier for doctors and healthcare teams to make when they can use CGM and easily see what is and is not going well with diabetes management), whether the data indicates the patients need MDI, basal insulin, physiologic insulin, GLP-1, combination GLP-1 and basal insulin, oral insulin, etc. As we previously wrote, we could imagine multiple products that could come out of the alliance, especially prefilled patch pen-like wearable devices (like Valeritas or CeQur) or simplified prefilled insulin pumps – we also think advice from HCPs to patients will be easier to develop and some will cover earlier stage medications such as GLP-1 (which will be a positive for Sanofi and Zealand). Additionally, we wonder if a concentrated insulin partnership is in the works, similar to the approach and Insulet and Lilly are taking. On oral insulin, management said it was both very very important but also not the first area of focus, given technical challenges. 

Questions and Answers

Q: What will the weight of the meta-analysis of Toujeo be with the FDA? Since it's mostly the high-dose patients that experience less hypoglycemia, is the meta-analysis something that could change the labeling?

A: Although I'm not an expert on decision-making within the regulatory agencies, I know that they will be interested in the entire submission package in relation to both type 1 and 2 diabetics. What we have presented so far, the pooled analysis of the first three of the EDITION studies, will certainly be part of what they'll consider. But I'm quite confident that they will want to look at the other studies including the Japanese type 2 diabetic study and in addition to that the Japnese type 1 diabetic study.

So, how convincing will it be to them, I think that's unknown by me. They, however, will be pleased with the consistency of the findings and the quality of the analysis as far as my opinion goes.

Q: In light of your recent deal with Medtronic, it is apparent that you are putting more emphasis on patches rather than oral insulin, based on your lack of oral GLP-1 or insulin in the upstream diabetes pipeline. So, what's your view on the orals and do you think that maybe patches or other drug delivery routes are a better path?

A: Well, we are going to consider a broad range of drug delivery technologies, and clearly, I mean, a patch would be an option that we are considering. Specifically, when you think about the heavy insulin users that's something that should be of interest to patients. And on the oral insulin, it certainly – oral delivery of GLP-1 may actually be more interesting, the most interesting. Oral insulin is something that is very challenging. We see a lot of both intra- and inter-variability, so it's probably not going to be our first area of focus.

Q: Even though glycemic control from the EDITION program was non-inferior to Lantus across all the trials, the dose needed to get equivalent glycemic control was 10% to 20% higher. What implications will this have on pricing?

A: There was indeed, throughout the studies, about a 10% or slightly greater difference in total units required to attain the same glucose control; however, it is unclear if this is statistically relevant. If systematic titration is used in clinical practice as in the studies, what people may expect is equivalent glucose control, less hypoglycemia, equal or less weight gain. Therefore, the 10% difference may not have very great clinical importance.

Additionally, Toujeo is a better tool, has a better profile, and each time that we have exposed physicians to this profile, it's quite clear: if access is not an issue, why shouldn't they go for Toujeo, either in an insulin naïve patient or in a switch patient? So it's very simple, we'll make sure that access is not an issue for Toujeo. Our directive is to maximize the penetration of Toujeo in the insulin naïve patients and make a very rapid conversion from Lantus to Toujeo. Now, pricing wise, it's too early to tell you about the pricing strategy, but of course, we want to make sure that access is not an issue.

Q: Solely based on the pre-specified secondary endpoint, across many of the trials, Toujeo was comparable to Lantus or just numerically lower but not statistically significant. So in terms of potential label discussions, how you're going to go and approach this with the FDA?

A: There's no one way to analyze hypoglycemic outcomes and between different populations and different ways of defining hypoglycemia risks that are probably different for a type 1 population or type 2 population and so on, it’s not very clear cut. So within the EDITION program, the selected main hypoglycemia endpoint was the after titration period from eight weeks to six months with the less than 70 mg/dL confirmed or severe definition for nocturnal. So a very specific category. Not all of the studies were this statistically significant difference, but the trend was always in the same direction. In some of the studies, the statistical power was low, but the nominal risk reduction was still of interest. So I believe that from the regulatory point of view, the most important characteristic of the finding so far will be to consistency, when all of the different categories of hypoglycemia, all the different definitions are concerned.

Q: Could you explain why week eight was picked as a cutoff point for the titration versus the maintenance days in the EDITION studies for the hypoglycemia measurement? A lot of the benefit seems to come in about the first 12 weeks, so if you picked week 12 as the cutoff, with 12 weeks to six months as maintenance, would you still see a maintenance period hypoglycemia benefit?

A: So, the time course question is an important and interesting one. First of all, the reason for selection of eight weeks was that in these studies, as in many other treat-to-target design studies, the most frequent visits in the majority of the titration of insulin dosing occurs in the first eight weeks. It's also true that the attainment of stable hemoglobin A1c values is usually not until 12 weeks but that's partly because of the delay between the titration and the attained glucose control by that measure. So, it was an arbitrary decision but the intention was to isolate the stable dose part of the study from the titration period. Now, why is that a plausible decision? One of the concerns when a new drug is studied, it's an artifact introduced by either continuing a previous treatment in this place, glargine U100, Lantus, and testing a new drug in comparison to it, usually there's a more cautious use of the new drug and so, you are unable to scientifically compare the first eight weeks or 12 weeks with the old drug with great confidence so that artifact was intended to be taken out the analysis by waiting until the titration was complete.

In my opinion, it's important to look at both the early period, the first eight week period and the eight-week period to six-month period and give them equivalent weight and scientific relevance, and the nice feature of these analyses is that the changes are pretty much the same in both of those periods of time and then the related question is to whether the rate of reduction of glucose with titration of Toujeo versus Lantus had an effect on hypoglycemia rates is an interesting one, we haven't done a full analyses to tease that out yet. My own opinion is that, that may account for some of the difference, but it doesn't account for the differences that are seen following the eight-week period, which in some of the studies, notably the Japan 2 study, where the differences in hypoglycemia rates continued to accelerate after the eight-week period. So obviously, we need more analyses,.I think there's a difference between the two insulins both early and late.

I think what we will need to do more and more is ask ourselves how this information is relevant from a patient point of view, rather than from a metric or arithmetic or data analysis point of view. Now experiencing hypoglycemia in the first week – the first few weeks of titration, we know is critically associated with the poor compliance and adherence. And we know that patients initiating insulin and the time they're titrating do experience frequent hypoglycemia. I mean the recent case of the Tresiba study showed that the hypoglycemia in the very first few weeks of titration, it offers a critical challenge for patients. So to us, a new insulin, like Toujeo, that is the initiation of treatment because it's not associated with the typical drawbacks of initiation, which are hypoglycemia or weight gain we think is a very relevant data.

Q: Going back to the Japanese trial and the consistency amongst the EDITION trials, the Japanese curve actually looks different than the EDITION 1/2/3 curves with a parallel curve after the titration in the 1/2/3 trials but a continuous divergence in the Japanese patients. Thoughts on what could account for that difference? Is it for instance that the Japanese patients are thinner, or are they actually just more aggressively managed, so that they get, relative to their body weight, higher insulin doses?

A: I think there maybe are two factors that, at least theoretically, may affect the difference in the shape of hypoglycemia accumulation curve. The first is that the Japanese type 2 study was exceptionally well conducted with a very high retention rate, and a very consistent titration rate. And the result of that, I think, is that this emphasizes the differences between the two insulins and allows the difference to continue to display itself as people continue to titrate and remain in the trial, the long end of the trial. But the other, and I think more important, reason is that the Japanese patients enrolled were smaller and leaner and, therefore, required much lower doses of insulin – even after the titration, about 25 units a day, and at lower doses like this, the differences in pharmacokinetics are probably more important. U100 insulin in those people, quite often, did not extend to 24 hour duration of action. And therefore there was a greater advantage in the leaner patients of the U300 preparation.

Q: We have been talking about disease management in type 2 diabetes since the early 1990s and, now, here is Sanofi in 2014 calling it Integrated Care and trying to solve the problem. So what has been wrong or what has been missing in all the previous disease management approaches, and what are you trying to do differently? And also, can you make it a little bit more tangible what we should be looking at, what you will be measuring to show that you're on the right track? And what would also be helpful if you give us some idea about timelines during which you want to come up with new solutions such as the patch pump?

A: Yes, you're right. There are a lot of talks about how to have a disease approach. I think however that the pharma companies have tried different type of approaches, and I do think that Sanofi has a very unique approach into this. And this is actually part of the reason why I was motivated in joining the company coming from GE Healthcare, where I have long-term medical device development experience as well as some pharma experience, as the last business I was running actually was pharma business within GE Healthcare.

So what is important when you're trying to do something like that in bringing two worlds together like the drug development and the devices. And actually I did have some experience of that in my last position, my previous company, to really operationalize it. It is to have enough expertise about the different areas you're trying to build in, but also have people who have the ability to understand the different worlds so that they can act as people who bridge the different worlds. So why I say that we are operationalizing is that first we have created this division, and I mean this Diabetes Division was created a couple of years, which increased the focus around the disease area itself. And then the next thing we did, and actually did recently, is that we are the first pharma company to have hired a Chief Patient Officer whose role is to really bring the patient at the center of how we develop these drug and these solutions, and in the couple of months that Anne Beal has been with us in the company working with Pierre's team, we already see a different approach in how we actually think about it.

The other important thing was actually the establishment of an integrated care center of excellence so that actually we can have or we can develop expertise and have people who understand the med-tech development, understand the software development, understand digital health, understand IT solutions with some experience of connecting that with the world of healthcare. So we brought experience actually from the outside, but this integrated care center of excellence led by Paul Sekhri who has a large experience both in pharma, but also with start-up and hi-tech companies in the world of devices, is working with the teams that we had in diabetes but also in our own device development.

So I do think that this is a change in the way you approach it. The development of devices is the core competency that was not necessarily core to Sanofi and the intent is not to reinvent everything, but it is to have enough people in-house who can actually enable to have this external innovation model, an external development model. So that's one big thing that makes it more tangible and more real and while why we are more comfortable with the progress that we can make here.

The other thing is that how do we measure it? I think here I have to say that the progress in technology has been extremely supportive of what we can hope for. Now, we all know that we're surrounded by quantified-self devices, you know with sensors, etc. I mean, why is that? It's just because the technology enables to track, measure and give feedback loop.

So that is true also for our integrated care approach. You can now think that you can have a lighter and more real-time feedback loop that can help you measure the impact. Now how to measure the impact on the adherence? I mean, we do need to start by demonstrating this in populations where we can have this overall measurement, either at a population level or at subset of patients' level so that you can actually measure the impact of using this solution.

Now one of the examples I mentioned earlier, for example diabeo in France, we do have a clinical trial going on, which was one of the first with the intent is to measure the clinical – the medical and economic value of leveraging mobile health in delivering healthcare.

Now to close, in terms of timeline, we are actually working on the project and we'll communicate those when we are ready to and so we'll do that in due time.

Q: It's clear that you are getting some additional benefits from this product, you're getting good A1c with lower hypo, but I think we've forgotten somewhat the original concept of the product was having the high unit dose concentration. We know that about a third of Levemir and Lantus patients need two shots, and this could replace it with one. You haven't really talked about much in terms of whether you're going to use that as part of the commercial positioning of this product at all, so could you talk to that?

A: Yes of course, there are some elements that are connected with these patients that are taking higher volumes. The benefit and the value of Toujeo is that we have this concentrated insulin, which means that you have one third of the usual volume that would be used by Lantus patients, for instance. So that makes the solution attractive. Now in terms of potential positioning this product or this solution is not only for people using large volumes of insulin. This – as I said, this solution is the new gold standard for people initiating insulin naïve patient or for people that have to switch and have to switch from Lantus to Toujeo. So, it’s not bad actually being able to avoid taking two shots instead of one shot instead of two shots. So, it's good. But clearly the strategy and the positioning of this product is not, at all, actually focused on this patient population. It's an additional benefit, but it's clearly the next gold standard.

From a scientific point of view, the data actually support that. If we look again, at the Japanese study, both the type 1 and type 2 study, the body weight of these people, the BMI, the insulin requirement are typically those of people requiring much less insulin addition compared to American or to European patients, and despite this very significant difference in insulin requirement, we see that the value of hypoglycemia reduction associated with Toujeo holds true. So, I think the science will support a much broader positioning.

Q: Returning to the issue of zero to eight weeks or nine weeks and beyond, you said there is no standard way to measure hypoglycemia. But, my question is, does that mean that regulators look equally at zero to eight weeks and nine to six months or do they – are they likely to weight zero to eight-week data more heavily than the pre-specified endpoint?

A: So, just let me add a comment on the hypoglycemia in the first eight-week question. I think the regulators, at least the FDA with which I'm more familiar, are showing increasing understanding of the complexities of diabetes management. And I believe that they will, of course, take very seriously the predefined hypoglycemia endpoint and insist that some favorable findings are present there, which I think there are. And they will also be interested in the first eight-week interval of time because they understand that in clinical practice, under different conditions either the early interval or the later interval maybe the more problematic or more important one. And so again, I do believe that both from the clinical point of view and the regulatory point of view the consistency of the findings throughout the studies over different time intervals and with different definitions is one of the important observations in the study so far.

In recent clinical development, we have seen that sometimes the pre-specified endpoint or the pre-specified window of observation did not match in terms of results with change in that window. So for instance, the time or definition of what is nocturnal, and with Toujeo we see that the expansion of the nocturnal definition does not change the significance of the data. The benefit of the first pace of insulin addition, the first eight weeks does not change in the second part of the study. The night benefit does not change when you analyze your time of the day analysis. The 70 mg/dL cut off for the defining hypoglycemia does not change when you look at 54 mg/dL. So what you see with Toujeo is that you don't have the partitioning of data, which some people will consider the demonstration of the fact that any observation is actually the result of chance. What you see is that the data partition, generally speaking, in the same direction and never partition in the opposite direction. The size of this effect is different in the different analysis but it's consistent from a direction point of view.

Q: Can you confirm that there have not been any signals, any CV safety signals of any sort and any of the EDITION trials with Toujeo?

A: Yes, I confirm there is not a different CV signal. And I'd like to also reemphasize that both in Europe and in the United States, the filing of Toujeo relies on the existence of the Lantus filing record. So, despite the fact that the technicality of filing is different in new drug application or a line extension both filing from a regulatory point of view are based on the existing record. The cardiovascular safety record both in Europe and in the United States has actually introduced the Lantus commercial label. And we're very pleased to see that the analysis of the EDITION program does not reveal any difference in terms of cardiovascular safety between the two products.

Appendix

Table 1: Hypoglycemia endpoints and reported six-month results from Toujeo’s phase 3 EDITION program.

Trial

Prespecified hypoglycemia outcomes (according to ClinicalTrials.gov)

Selected reported hypoglycemia outcomes (all vs. Lantus)

EDITION I

% of patients with ≥1 nocturnal hypoglycemia event from week 9 to endpoint, defined as severe and/or confirmed by plasma glucose ≤70 mg/dl between 0:00 and 5:59 hours

Statistically significantly fewer people on Toujeo had ≥1 severe or confirmed nocturnal hypoglycemia during month three to month six (21% relative rate reduction).

Occurrence of anytime hypoglycemic events was numerically lower in the Toujeo arm but did not reach statistical significance.

EDITION II

Statistically significantly fewer people on Toujeo had ≥1 severe or confirmed nocturnal hypoglycemia during month three to month six (23% relative rate reduction). Over the entire six-month treatment period, 27% fewer patients on Toujeo experienced a nocturnal hypoglycemia event, and 10% fewer experienced a daytime hypoglycemia event (both differences were statistically significant).

EDITION III

No significant reduction in % of patients with ≥1 severe or confirmed nocturnal hypoglycemic event (≤70 mg/dl) from nine weeks to six months. However, this difference became significant when the analysis was extended to include the full six-month study period, showing a 24% relative rate reduction. Percentage of patients with ≤1 severe or confirmed hypoglycemic event any time of day (not just nocturnal) was additionally significant the full six-month study period (25% relative rate reduction).

EDITION IV

Number of patients with various types of hypoglycemia events over 12 months.

No significant reduction in rates of anytime confirmed (≤70 mg/dl) or severe hypoglycemia during full six-month duration of trial, although Toujeo users had statistically significantly reduced nocturnal hypoglycemia in the first eight weeks of treatment (31% relative rate reduction).

EDITION JP I

Number of patients with various types of hypoglycemia events over 6 months.

% of patients reporting ≥1 confirmed (≤70 mg/dl) or severe hypoglycemia event was statistically significantly lower from baseline to week eight for both anytime and nocturnal hypoglycemia (16% and 29% relative rate reductions, respectively). For the time period of week nine to six months, statistical significance was not reached. For baseline to six months, statistically significance was reached for nocturnal hypoglycemia (15% relative rate reduction) but not anytime hypoglycemia.

 % of patients reporting ≥1 confirmed (≤54 mg/dl) or severe hypoglycemia event was statistically significantly lower from baseline to week eight, from week nine to six months, and from baseline to six months for both nocturnal and anytime hypoglycemia. Over the full six months, the relative rate reductions were 31% and 23% for nocturnal and anytime, respectively.

EDITION JP II

% of patients reporting ≥1 confirmed (≤70 mg/dl) or severe hypoglycemic event was statistically significantly reduced for nocturnal hypoglycemia from baseline to six months (38% relative rate reduction) and week nine to six months (42% relative rate reduction), but not from baseline to week eight. For anytime hypoglycemia, there was a statistically significant 31% relative rate reduction from baseline to week eight, but no statistical significance achieved from baseline to six months.  

 

-- by Jessica Dong, Sabrina Lee, and Kelly Close