Lexicon announces positive topline proof-of-concept results for SGLT-1/2 dual inhibitor LX4211 in type 1 diabetes – April 14, 2014

Executive Highlights

  • LX4211 led to a ~26% placebo-adjusted reduction in bolus insulin, a 0.49% placebo-adjusted decrease in A1c, significantly improved time-in-zone, and a ~2 kg (~5 lb) weight benefit over four weeks in adults with type 1 diabetes.
  • Lexicon expects to meet with the FDA in coming months and aims to initiate a phase 3 trial in type 1 diabetes by the end of the year.
  • Management may discuss fast-track status with FDA.

This morning, Lexicon announced positive topline results from a proof-of-concept phase 2 study of its SGLT-1/SGLT-2 dual inhibitor LX4211 in adults with type 1 diabetes. Over the course of four weeks of treatment, patients on LX4211 saw statistically significant reductions in bolus insulin use (-32% with LX4211, -6.4% with placebo, p=0.007) while simultaneously improving A1c (-0.55% with LX4211, -0.06% with placebo, p=0.002, baseline A1c=7.9%). Patients also saw improvements in body weight (1.72 kg [~4 lbs] lost with LX4211, 0.50 kg [~1 lb] gained with placebo, p=0.005). Based on blinded CGM, the LX4211 group saw a statistically significant 12% increase in “time in zone” (time spent in the 70-180 mg/dl range) and an 11% reduction in time spent in hyperglycemia (see table below). In Q&A management noted that the treatment group did not see much of a change in basal insulin requirements, suggesting that LX4211 is primarily having an effect on prandial glucose. Management did not disclose much specific safety data, except to say that the drug was well tolerated and that there were no signals of significant concern. From the CGM data we saw, it appeared that, at the very least, LX4211 did not increase hypoglycemia from baseline levels. Lexicon expects to present full results at a meeting in 2H14.

There were a number of notable study design limitations that were understandable at the proof-of-concept stage, but still merit consideration. Patients were given a relatively broad glycemic target (70-180 mg/dl) and were not titrating to a very specific glucose level. Patients in both groups were told to take glucose measurements and adjust insulin as they had done in the past, which management characterized as a relatively “hands-off” protocol. This level of freedom and flexibility for patients in the study might increase the real-world relevance of the trial, but might also cloud the distinction between LX4211 being more effective and LX4211 being easier to take (both, of course, are valuable). Lexicon specifically selected a relatively low A1c window (7.0-9.0%) for the trial in order to capture patients that were more likely to be adherent to treatment.

During Q&A, management shared that the company hopes to meet with the FDA regarding LX4211’s continued development in type 1 diabetes in the next two months or so, and (contingent upon the outcome of the discussion) could initiate a phase 3 trial in type 1 diabetes by the end of 2014. Management remarked in Q&A that it would be “open” to discussing fast track status for type 1 diabetes, given the absence of oral drugs available (we imagine we might learn more about this following the company’s discussion with the FDA). Fast track status allows for more frequent communication between the FDA and sponsor as well as rolling review of an NDA submission (meaning the FDA would review application materials as they are ready vs. waiting for the complete application). Consistent with previous updates, the company appears to be willing and able to conduct phase 3 testing for type 1 diabetes independently, although it is still engaged in partnership discussion regarding LX4211 in type 2 diabetes (for which phase 3 development is much more costly due to CVOTs). Management also shared that (as opposed to the recently-completed proof-of-concept trial) the type 1 diabetes phase 3 trial could have A1c as its primary endpoint, and might be designed as a slightly larger program that could show a hypoglycemia benefit, although the glycemic target parameters would likely be the similar. A longer phase 3 trial would be expected to show more of an A1c reduction than was seen in phase 2 (management noted that the 0.55% A1c reduction seen after four weeks was similar to what was seen with Merck’s Januvia after 24 weeks). Lexicon is also interested in studying LX4211 in pediatric type 1 diabetes patients, who could see particular benefit in quality of life and ease of management with reduced bolus dosing.

  • While these results are good news for Lexicon, their impact on the ongoing partnership discussions for LX4211 may be complicated, especially given that Lexicon might pursue a phase 3 program for type 1 diabetes independently. However, we imagine that a joint partnership for type 1 as well as type 2 diabetes is not out of the picture as of yet, although if a partnership involves both indications, it is likely that Lexicon will have a greater role to play in the type 1 diabetes program. Partnership discussions for LX4211 have been ongoing for quite some time now; while during Q&A, management reaffirmed that the process is continuing as planned, at one point, management had implied the timing would be faster (we had been surprised at this at the time). Read our Lexicon 4Q13 Report for more on the partnership process for LX4211.
  • These results are part of a broader investment in the exploration of type 2 diabetes drug classes in type 1 diabetes. SGLT inhibitors show particular promise for this application, due to their insulin-independent mechanism of action. Last year, BMS completed a phase 2 study investigating the SGLT-2 inhibitor dapagliflozin (Farxiga/Forxiga; now owned solely by AZ) in type 1 diabetes patients (read our coverage of those poster presentations in our ADA 2013 SGLT-2 Inhibitor Report). J&J does not have any studies on its SGLT-2 inhibitor Invokana (canagliflozin) in type 1 diabetes registered on, although the much-respected Dr. Anne Peters discussed the off-label application at the recent Clinical Diabetes Technology Meeting. See our Novo Nordisk 4Q13 Report for details on the company’s LATIN T1D/ADJUNCT program investigating the GLP-1 agonist Victoza (liraglutide) in type 1 diabetes. Merck has tested its market-leading DPP-4 inhibitor Januvia (sitagliptin) in type 1 diabetes, and Lilly has also investigated its phase 2 glucagon receptor antagonist LY2409021 in type 1 diabetes.
  • Background on the design of the proof-of-concept phase 2 study of LX4211 in type 1 diabetes:
    • Pioneer phase: The trial began with a small, unblinded pioneer phase (n=3), with a more aggressive titration protocol (to determine safety). Results from this phase of the study were presented at the JP Morgan Healthcare Conference earlier this year.
    • Expansion phase study design: This phase of the study enrolled an additional 30 patients (for a total n of 33) who were either on insulin pumps or MDI. The 28-day, placebo-controlled, double-blind phase of the study was conducted in seven sites in the US, using the 400 mg daily dose used in type 2 diabetes. Patients in the expansion phase were given a fairly broad glucose target range (70-180 mg/dl) and were told to titrate their insulin as per their usual protocol – a relatively “hands-off” trial design (according to management). Blinded CGM was used to assess glycemic variability and time-in-range – Lexicon deserves major kudos for focusing on “time in zone” and parameters beyond A1c, and we hope to see that focus continue in phase 3, as well as in trials conducted by other companies. Interestingly, Medtronic’s Enlite was the CGM used for the trial. Some researchers do view CGM use in trials as a major hassle; while it surely is more work, we also think sometimes researchers might perceive the hassle associated with CGM as from another “generation” and may not realize it is easier to use now. It was terrific to hear about the “time in zone” data associated with LX4211; at the same time, we can imagine future “time in zone” data being even better with even more accurate CGM, especially in the hypo zone.
    • Inclusion criteria and baseline patient characteristics: Patients had to be age 18-55, on MDI or an insulin pump, and have an A1c between 7.0% and 9.0%. This A1c range (as opposed to the ~7.0-10.5% A1c range used in type 2 diabetes) was chosen to enrich the population with patients that were more likely to be adherent to treatment. Mean baseline A1c was 7.9% in both the LX4211 and the placebo groups. Notably, median age and median FPG were noticeably higher in the LX4211 group compared with placebo (by ten years and 10 mg/dl, respectively); despite the age discrepancy, diabetes duration was comparable between treatment groups. The effect of those slight imbalances is hard to evaluate, although we would expect to see fewer imbalances in phase 3 with a larger sample size.
    • Efficacy results: After four weeks of treatment, patients on LX4211 saw a 32% mean reduction in bolus insulin use, compared to a 6.4% mean reduction seen in the placebo arm (p=0.007). LX4211 led to a -0.55% mean reduction in A1c, compared to a reduction of 0.06% with placebo (p=0.002, baseline A1c=7.9%), and a 1.72 kg (~4lb) mean body weight reduction compared to 0.50 kg (~1 lb) weight gain on placebo. Blinded CGM showed a statistically significant improvement in time-in-range (p=0.003) and a reduction in time in hyperglycemia (p=0.002) with LX4211 (see the table below).

Table 1: CGM data on time in range, time in hypoglycemia, and time in hyperglycemia




Glycemic range





< 70 mg/dl





70 – 180 mg/dl





> 180 mg/dl





  • Safety: Lexicon’s presentation did not explicitly break out data on adverse events, and management only commented that the drug was safe and well tolerated. The CGM data did show the amount of time in hypoglycemia, although it did not indicate the severity of hypoglycemia or the number of individual episodes. For both LX4211 and placebo, time spent in hypoglycemia decreased from baseline, but the decrease seemed to be of a slightly smaller magnitude with LX4211. We do not see this difference as clinically significant, given the small study size. During the presentation, Lexicon CEO Dr. Arthur Sands suggested that the relatively low amount of hypoglycemia seen in the placebo arm was due in part to the diminished glycemic control seen in that group. It will be interesting to see full data on safety, including GI adverse events (which have been associated with some other SGLT-1 inhibiting agents, although they do not seem to be a major issue with LX4211 based on studies in type 2 diabetes).
  • Lexicon is considering ways to test LX4211 further “upstream” the progression of type 1 diabetes, in younger patients. Management commented during Q&A that LX4211’s ability to potentially simplify patients’ dosing regimen could be a major advantage for children in school and the broader young type 1 diabetes patient population (which faces a relatively high adherence barrier). If LX4211 were found to work well in children, the change in treatment plan so that children and teachers wouldn’t need to dose insulin during the day could be a major change.

Questions and Answers

Q: Could you comment on a timeline going forward for a phase 3 trial, keeping in mind there is no partnership in place yet, and also what this phase 3 trial might look like?

A: We’ve already submitted a meeting request to FDA for end of phase 2 trial for type 1 diabetes. We expect to get that in the next 2 months or so. Depending on how that goes we hope to initiate a phase 3 trial in type 1 diabetes by the end of this year. The phase 3 trial we’re considering for type 1 diabetes would have a primary endpoint of A1c reduction. So in that context, these results are very important. It’s difficult to know how these studies will work because patients automatically adjust their insulin, and so these results give us a lot of confidence and encouragement since despite patients adjusting their insulin in this study, we still saw A1c improvement. Thus, we do think that [A1c] will be a good primary endpoint for us in a phase 3 program for type 1 diabetes.

Q: How long do you think a phase 3 trial would last, and would you need a second trial?

A: We’re going to have to discuss that with the FDA. On one hand, we do feel that part of this profile has been very low rates of hypoglycemia, and we would benefit from a larger program by potentially being able to show significant reduction in symptomatic and severe hypoglycemia. These results are encouraging that they do merit discussion with the FDA over the pathway towards approval.

Q: Given that this directly addresses what the FDA is looking for in a type 1 drug, for which there are currently no oral therapies, do you think this would fit into a fast track for type 1?

A: Yes we do, we believe this is novel and medically important and we are open to discussing our options with the FDA in this matter.

Q: How does this data affect your partnership process and where you are in that process? Also, are you still looking to keep type 1 distinct from type 2 in future treatment?

A: We’re continuing our partnership discussions and feel that they’re going well. We have consistently said that we expect to play a bigger role in type 1, including partnership or any other arrangement, so that continues to be the expectation.

Q: Does this data change the pace of any conversations you may be having in terms of partnership? Or could you provide some color around how this data may or may not change the dialogue?

A: I think that the data demonstrates one of the elements of the profile that we wanted to see personally. So I think this is positive in our partnership discussions and increases our options. We’re continuing discussions that we’ve had ongoing here, and we’re hopeful that we come close to a conclusion soon.

Q: I wanted to clarify a few points on this data. On the hypoglycemia numbers, it looks like it may actually just be noise here, but the time in the hypoglycemic range was less in the placebo arm than in the treatment arm. If you looked at definitions of hypoglycemia, was there a numerical difference in favor of the placebo arm in this study?

A: I think the important thing is that the placebo arm is doing some insulin adjustment and if you recall the pie chart graphs, they are definitely going up in time spent in hyperglycemic range. It’s never surprising that if you drop insulin, you will see a drop in hypoglycemia, because you are losing some glycemic control. So what’s important with LX4211 without question, is that (looking at the pie chart) patients decrease time spent in hypoglycemia and increase time in euglycemic range and they’re achieving that with no increase in HbA1c. That’s the distinction between the placebo and treatment arms.

Q: The A1c numbers show that the placebo arm had a slight reduction in A1c. When you look at pie chart, it looks like they had significant time spent in the euglycemic range.

A: There is really not much change in A1c in the placebo arm right, 0.06? It’s hard to say that it’s anything more than noise and unchanged. The number of patients is small, so I don’t think you can expect an exact correlation between CGM readings and A1c; they are different aspects of glycemic control, but more or less they should go together, which I think they do.

Q: What was the titration paradigm, or the treatment target goal, provided to patients in this study? Were there specific target fasting plasma glucose levels that patients were asked to achieve with their insulin?

A: The patients were instructed to target glycemic control in the 70-180 mg/dl range. They were all given the same instructions and a reasonable range for their glycemic control readings, which they carry on per their normal routine by finger sticks. They make the judgment calls on adjusting their insulin based on their diet, meals, exercise, and other lifestyle issues. Essentially, both are given standard instructions and then we look at data at the end. Basically you want them to behave as they normally would and target their euglycemic range as normal, and those were the instructions given.

Q: So in the translation from this phase 2 trial into phase 3, would the FDA ask you to run a study under a very different set of conditions where patients are being asked not to keep insulin regimen and glucose levels roughly the same, but rather introduce a more aggressive treat-to-target paradigm for patients?

A: I believe the FDA would support current targets as delineated by ADA rather than to ask for some new set of targets for a phase 3 program. We believe the paradigm that we evaluated in phase 2 is applicable to phase 3.

Q: Just to be clear, do you feel confident that you have enough long-term data to go into phase 3 in either diabetes segment? Is there a plan to go into phase 3 on your own in type 1 or type 2 given the long partnership timelines we’ve seen?

A: In type 2 diabetes, we have all the data we need to advance into a phase 3 program. The phase 2 dose ranging study was very supportive and there is a popular study similar in scope to that of the SGLT-2 inhibitors that have moved forward. For type 1, we believe that these data are sufficient to move forward into phase 3. We don’t have broad dose ranging in type 1 diabetes. We feel that the type 2 data and the experience we’ve gained with our compound are very relevant and support our dose selection for type 1. Also the efficacy and safety we’ve seen support our dose selection for type 1. Given the nature of these results and what we’ve learned about this compound from both indications, we feel in a good position to move forward into phase 3 for both type 1 and type 2 diabetes.

Q: With regards to what happens if a partner doesn’t get signed, will you move forward alone?

A: Our expectation is that we will have a partner. But type 1 diabetes, in terms of a development program, is a different scope of program than type 2 and this is something we feel we could do if we chose to.

Q: Do the outcomes look any different for those who are on insulin pumps than those who are taking basal bolus injections?

A: We don’t have that subgroup analyzed yet. From the results we’ve looked at so far, we’ve seen general consistency and robust results in these data.

Q: Based on the data that you have from the pie chart as well as A1c numbers, what would you extrapolate the full-blown A1c effect to look like if you ran a longer 6-month trial or even just 12 weeks to get to the point of the typical period for measuring A1c?

A: Obviously the A1c effect depends on the half-life of red blood cells, so we know that it should be greater than what we see at four weeks. I would emphasize that we haven’t done longer studies in type 1 patients, but that we would anticipate a greater effect. I would emphasize that at four weeks, the 0.55% reduction we’re seeing is equivalent to the effect observed with sitagliptin at 24 weeks in phase 3. This is a nice effect already here at four weeks.

Q: Could you provide us a little more color around adverse events, specifically around any increases in volume or frequency of urination that patients may have experienced during the course of the study.

A: At this time since this is all very new, we are just disclosing that we are very pleased with the safety profile, and we won’t go into any details. There was nothing of significant concern, and the drug was very well tolerated

Q: Did you look at basal insulin reduction as well?

A: We did. There wasn’t much of a change in basal insulin. This contrasts with what we see in selective SGLT-2 inhibitors in similarly sized trials in type 1 diabetes. I think this will be an evolving point of differentiation for our agent where we’re seeing primarily this effect in bolus insulin use.

Q: In the pioneer cohort, there was an approximate 15-20% reduction in just the three patients in terms of basal, correct?

A: Yes, keep in mind for that group there was a more aggressive protocol for safety. There was a proactive reduction of basal insulin use to ensure they didn’t get hypoglycemia. So this study was more hands-off, asking patients to regulate themselves as they normally would.

Q: Is this the same type of protocol employed in selective SGLT2 studies?

A: Yes, they all implemented a proactive reduction pre-emptive of insulin of some percentage. Most studies were open label so they weren’t supposed to be controlled. They really saw no change in bolus insulin use and they were four to eight week studies.

Q: At the end of phase 2 meeting with FDA, do you anticipate pursuing type 1 separately from type 2?

A: We believe we’ll get some insight on that possibility as we discuss overall safety requirements for these two indications. We do think that LX4211 should be developed for type 2 but the regulatory interactions could give us insight into the timing of this, and if it’s possible that we emphasize type 1 then that’s something we want to understand.

Q: Will we see this data presented at ADA or later?

A: Well it’s too late to submit an abstract to ADA, but we will attempt to present at the next available major diabetes conference.

Q: Could you also give us any update on current succession plans?

A: We can only comment that the search process is ongoing.Q: Could you talk about any research that you might have done with patients, teachers, parents, families, or anyone that has a lot to think about in dosing insulin, especially fast-acting insulin, in children during the day? Could you talk about what getting something like this drug approved might enable?

A: This is speculative, but I think some of the next steps would be to move this drug in appropriate trials to younger age groups, and we’re studying this possibility right now stepwise, probably first with teens and young adults and then stepwise, younger. Of course those trials are intense and this is a patient population that (especially teens) can be notoriously poorly regulated and this compound could offer some potentially significant advantages in terms of simplifying insulin regimens depending on the insulin needs of patients. We’ve seen some fairly dramatic reductions in the need for bolus insulin injection or the quantity of it and that really could have a lifestyle improvement, in school time for example. We have to see how this plays out, but there is potentially significant benefit there, in terms of simplification of insulin regimen, while also providing glycemic control.

Q: It seems like getting rid of insulin dosing at school could be a really huge driver and that the FDA might be a little more open at looking at time in zone or time in range data. It seems to be that now that there are products that can measure that (hypoglycemia, time in zone, etc.), there may be more openness and interest in this area. Is this your impression as well?

A: I think this is definitely growing, now with the availability of CGM technology; it’s very powerful and quantitative in addition to regular safety reports on hypoglycemia. I do think it will be very important and again, we would hope that some of the benefits of daytime insulin use that you’ve alluded to could actually occur.

-- by Manu Venkat, Jenny Tan, Jessica Dong, and Kelly Close