Lilly/BI announced that DPP-4 inhibitor Tradjenta met its primary endpoint in the CAROLINA trial, demonstrating non-inferiority to sulfonylurea glimepiride on three-point MACE. No numbers were provided and, while not explicitly stated, it seems Tradjenta did not show superiority to glimepiride. Full results will be presented on June 10 at ADA 2019 in San Francisco.
These results represent a key development in the long-standing debate over sulfonylureas’ CV safety. All SUs currently have a black box label for increased CV mortality; while recent literature reviews have found little evidence of elevated CV risk with the class, we suspect “real world” is different (not all patients will have HCPs who help them choose the best SFU, dosinng etc). CAROLINA fills an important void in providing sound, head-to-head evidence that speaks to this question. On this front, University of Toronto’s Dr. Dan Drucker said, “The demonstration of non-inferiority is extremely valuable for the diabetes community and will likely go a long way towards settling the long-standing debate surrounding the cardiovascular safety of the SU class.” On the other hand, concern remains over the hypoglycemia, weight gain, and potential beta cell burnout also associated with SUs. Unfortunately, broad affordability and accessibility has made these agents fixtures in treatment algorithms, with ~half of diabetes prescriptions going to SUs. Still, ahead of seeing the full readout, it is tentatively reassuring to the millions of patients currently on these agents that they do not seem to come with increased CV risk.
Lilly/BI announced this morning topline results from the CAROLINA CVOT of DPP-4 inhibitor Tradjenta (linagliptin) vs. sulfonylurea glimepiride: Tradjenta met its primary endpoint of non-inferiority to glimepiride on three-point MACE (non-fatal MI, non-fatal stroke, and CV death). No numbers were provided in the topline release, and while not explicitly stated, we can reasonably assume that CV superiority was not achieved. Full results from the study will be presented on June 10 at ADA 2019 in San Francisco.
CAROLINA results were and remain highly-anticipated due to the trial’s unique use of glimepiride as an active comparator. Whereas some negative effects of sulfonylureas – hypoglycemia risk and weight gain – are well-established, their impact on long-term CV outcomes has been long debated (more on this below). CAROLINA offers important granularity and robust data on each of these issues, and a neutral topline result can be tentatively viewed as reassuring to the millions of patients currently taking these agents.
On this front, Dr. Dan Drucker (University of Toronto) expressed optimism over these results: “The demonstration of non-inferiority is extremely valuable for the diabetes community and will likely go a long way towards settling the long-standing debate surrounding the cardiovascular safety of the SU class. We await the full trial results, particularly the rates of hypoglycemia, with great interest.” To be sure, we hesitate to draw any strong conclusions before seeing the full readout – presumably it will also be valuable to see weight comparisons.
We’re curious to see how CAROLINA will be interpreted by the broader diabetes community and what effect these results may have on prescribing practices, treatment algorithms, and even insurance coverage. Sulfonylureas have been much-maligned in recent years amongst highly-respected KOLs (see criticisms from Drs. Ralph DeFronzo and Jay Skyler among others), with some advocating for their complete elimination from treatment algorithms. Nevertheless, the broad affordability and accessibility has made these agents fixtures in treatment algorithms, with ~half of diabetes prescriptions going to SUs. Here, we also note that there seems to be very meaningful heterogeneity within the SU class, with glimepiride and gliclazide (the latter only available OUS), to our understanding, the better agents among them.
CAROLINA enrolled 6,033 patients with type 2 diabetes; notably, 35% of participants had established CVD, a relatively low fraction balanced by a median >6 years of follow-up.
Sulfonylureas and Potential CV Harm
The link between sulfonylureas and potential CV harm has been long and hotly debated, especially in light of their status as one of the most widely used drugs in the treatment of type 2 diabetes – a high-risk population at baseline. The worry that SUs may have a deleterious effect on CV safety first emerged in the 1960s, when the University Group Diabetes Program found an increased risk of all-cause and CV mortality with sulfonylurea treatment vs. placebo. Although this study was not designed/powered to test for CV safety, it has led all sulfonylureas in the US to harbor a label indicating an association with increased CV mortality. The study has also since been criticized for not standardizing its population for higher CV risk in the sulfonylurea group, and its applicability to modern diabetes treatment paradigms has also been questioned. Mechanistically, researchers have also pointed to sulfonylureas having a negative CV effect due to binding on the potassium ATP channel in cardiac myocytes and other cells.
In 2013, Dr. Julio Rosenstock and colleagues reviewed published literature to evaluate the impact of sulfonylureas on CV outcomes in modern, long-term RCTs in which they were compared in a head-to-head fashion versus an active comparator or were used as part of a treatment strategy. This review of 15 trials found no indication of an increase in the incidence of CV events with the use of sulfonylureas – nonetheless, the authors stressed that the data analyzed was limited, and no adequately powered, formal head-to-head CVOT existed at the time. CAROLINA results importantly fill this void, while also seeming to reaffirm the review’s findings by demonstrating an absence of increased CV risk with SU treatment.
Commercial and Class Implications
Lilly management has noted in the past that positive CAROLINA results could have accelerated attrition away from sulfonylureas and toward more efficacious diabetes therapy classes. On Lilly’s 4Q17 call, President of Lilly Diabetes Mr. Enrique Conterno expressed his view that “most classes will benefit [from positive CAROLINA results], and it will not be limited to DPP-4s or to Tradjenta in particular – we’ll also see an acceleration for SGLT-2s (in particular, Jardiance) and GLP-1s.” In light of neutral results, though, we would still expect these trends – already taking place – to continue. From where we stand, it’s difficult to model either situation: We don’t know an HCP who “likes” SUs, but most begrudgingly use them because they’re incredibly cheap – and it’s unclear how CV superiority of Tradjenta would change that dynamic. At what point would payers and HCPs change their coverage decisions and prescribing habits? Nonetheless, we do hope to see the use of sulfonylureas steadily decline, especially with the emergence of more efficacious diabetes therapies that not only provide superior glucose-lowering and hypoglycemia avoidance, but also cardio- and reno-protective benefits.
CAROLINA is now the fifth DPP-4 inhibitor CVOT to complete, with all establishing the CV safety of the class. Of note, CAROLINA is the only of these against an active comparator, and by far the longest DPP-4 CVOT with a median follow-up of >6 years (in fact, this outpaces almost all CVOTs, in all classes). See our DPP-4 CVOT landscape here. The TECOS (Merck’s Januvia), EXAMINE (Takeda’s Nesina), SAVOR-TIMI (AZ’s Onglyza), and CARMELINA (Lilly/BI’s Tradjenta) CVOTs have all been published, and each found non-inferiority for a DPP-4 vs. placebo on three-point MACE. That said, SAVOR-TIMI did show increased risk for heart failure hospitalization with Onglyza.
Phase 3 of Linagliptin vs. Glimepiride: The Basis for CAROLINA
Notably, a previous two-year long phase 3 trial comparing linagliptin to glimepiride (n=1,519) hinted at a reduced relative CV risk with linagliptin (HR=0.46, 95% CI: 0.23-0.91, p=0.0213) on a four-point MACE endpoint (CV death, non-fatal MI, non-fatal stroke, or hospitalization for unstable angina). Importantly, the study was designed to assess glucose control instead of CV outcomes, and was not powered for the latter. Only 38 total CV events occurred in the shorter study, making these results suggestive and hypothesis-generating at best, rather than a definitive finding of CV superiority for linagliptin vs. glimepiride. Indeed, as Dr. Benjamin Scirica has pointed out, the phase 3 programs for DPP-4 inhibitors were quite misleading in that they indicated substantial cardioprotection that did not pan out in longer, larger CVOTs.
Although CAROLINA may not have demonstrated CV superiority of Tradjenta over SUs, it’s important to remember the key improvements DPP-4 inhibitors have brought to the diabetes treatment arsenal. Sulfonylureas are well-known to increase hypoglycemia risk, and weight gain; the increased risk of hypoglycemia with sulfonylureas is especially worrisome. Many also worry about beta cell burnout. Notably, phase 3 data of Tradjenta compared to glimepiride showed significant reductions in both hypoglycemia (7% vs. 36% of participants; p<0.0001) and severe hypoglycemia (1% vs. 12%; p<0.0001) with Tradjenta treatment, and we fully expect similar trends to emerge in CAROLINA.
--by Martin Kurian, Ann Carracher, and Kelly Close