–Sanjay Kaul plunges into the LEADER data pool
There will be a lot of discussion in the near future about the LEADER trial, the large new trial assessing cardiovascular outcomes with liraglutide (Victoza, Novo Nordisk). The trial is the latest in a series of large CVOTs with new diabetes drugs showing benefit— or at least no harm— with these drugs. It is no exaggeration to state that these trials are causing a major shakeup in the increasingly important area of type 2 diabetes, though it is still unclear exactly how these trials will be applied in clinical practice, since many of these drugs have already achieved enormous success in the marketplace despite the earlier absence of evidence regarding outcomes.
I asked clinical trial expert Sanjay Kaul, MD, of Cedars-Sinai Medical Center in Los Angeles, for his perspective on the trial, in which 9,340 people with type 2 diabetes and high risk for cardiovascular disease were randomized to liraglutide or placebo. With a median followup of 3.8 years there was a significant reduction in the first occurrence of death from cardiovascular causes, nonfatal MI, or nonfatal stroke (13% in the liraglutide group versus 14.9% in the placebo group, HR 0.87, CI 0.78-0.97, p<0.001 for noninferiority, p=0.01 for superiority).
Placing the trial in perspective, Kaul noted that “until last September, when the results of EMPA-REG OUTCOME trial were first announced, diabetes intervention trials had a checkered history with regards to cardiovascular outcomes for nearly 5 decades. Now within the space of 9 months we have 2 published trials with unequivocal evidence of cardiovascular outcome benefit, plus one that reported favorable top line results with Semaglutide, a long-acting GLP-1 agonist, in SUSTAIN 6.”
“Overall,” said Kaul, “the LEADER trial results along with the EMPA-REG OUTCOME trial results represent a clinical breakthrough in treating T2DM warranting a shift in the treatment paradigm to consider treatment with empagliflozin or liraglutide as a first-line therapy along with metformin in patients with T2DM and atherosclerotic CVD.”
Kaul says that it is now time to change the focus of clinical trials in this area. Until now the main goal of these trials has been to rule out excessive risk. Now we need to know how the drugs should be used in the much larger lower-risk patient population. “It is time to move beyond the restricted focus of ruling out unacceptable cardiovascular harm in high-risk patients,” said Kaul. CVOTs should be designed to assess outcomes “in lower-risk patients that are more representative of the patients we treat in clinical practice.”
Kaul said LEADER was “a well-conducted outcome trial where the quality and the quantity of the data yield a clear and convincing evidence of cardiovascular outcome benefit and an overall favorable benefit-risk profile for liraglutide.” Kaul compared LEADER with EMPA-REG and made a number of notable observations:
“1. The superior reduction in the primary composite endpoint (1.9% absolute risk difference) in LEADER was driven by a significant reduction in CV death (1.3% absolute risk difference) with numerical, but not statistically significant, favorable difference in nonfatal MI and nonfatal stroke. The favorable effect on nonfatal stroke in LEADER (HR 0.89) is in contrast to the numerical imbalance not favoring empagliflozin in EMPA-REG (HR 1.24). The absolute and relative risk difference in MACE-3, nonfatal MI, or hospitalization for unstable angina was similar to that observed in EMPA-REG trial; however, the p value for MACE-3 was more robust in the LEADER trial (p=0.01 vs p=0.04), thereby rendering it more stable to sensitivity analyses. The absolute and relative benefit in cardiovascular death, all-cause death, and hospitalization for heart failure was relatively greater in the EMPA-REG trial.
“2. The delayed separation of the KM curves in LEADER (>12 months for CV death and >18 months for all-cause deaths and hospitalization for heart failure) contrasts with the early separation of curves (<3 months) in the EMPA-REG trial. This might (at the risk of speculation) indicate different mechanistic pathways underlying outcome benefit. As with empagliflozin, the reduction in CV risk with liraglutide is unlikely to be attributable to the small but favorable differences in cardiometabolic factors such as HbA1c, BP or weight loss.
“3. A notable difference from the EMPA-REG trial is that silent MIs were counted in the primary composite outcome in LEADER and they were favorably impacted (HR 0.86). Among half of the patients in EMPA-REG in whom silent MIs were evaluated, treatment with empagliflozin was associated with a nonsignificant increase in risk (HR 1.28).
“4. Unlike the results of EMPA-REG trial, there is a treatment by renal function interaction in the LEADER trial, i.e., treatment benefit with regards to primary composite endpoint is only evident in those with moderate or severe renal impairment (eGFR <60ml/min) in LEADER. The interaction p value of 0.01 is unadjusted for multiple comparisons. It is unclear whether this interaction is also evident for CV and all-cause death as well.”
Kaul also pointed to several potentially important lingering questions about the trial, relating to a lack of effect on retinopathy, a possible increase in pancreatic cancer, and the effect of liraglutide on lower risk patients:
“1. Why was retinopathy endpoint not impacted favorably within the time frame that nephropathy endpoint was? What drove the reduction in nephropathy endpoint – ‘softer’ or ‘harder’ components? What accounts for nephroprotection with liraglutide?
- The increase in incident pancreatic cancer (13 vs 5, HR 2.60, 95% CI 0.93, 7.28 p = 0.06) is obviously a concern in light of the relative short follow up.
Why did patients without established CVD (19% of the overall cohort) not show reduction in primary composite outcomes? Was this lack of benefit also seen with respect to CV or all-cause death? If true, this has important implications for designing CVOT to show benefit in patients at lower CV risk. Because 99% of patients enrolled in EMPA-REG had pre-existent CVD, it remains to be established what the effects of empagliflozin might be in T2DM patients without CVD.
Double the number of planned events
Kaul also pointed out an important deviation from the investigators’ original plan. LEADER, he notes, “was an event-driven trial requiring 611 events to rule out a HR of 1.3 at a power of 90%. The observed annual primary endpoint event rate was 3.9% in the placebo arm (14.9% over 3.8 years), which was >2-fold higher than the expected annual event rate of 1.8%.” As a result, “the final number of observed events accrued (n=1306) exceed the planned number of events (n=611) by more than 2-fold.”
I asked the trial investigators, John Buse and Steven Marso, to explain this unusual circumstance. (In my experience it’s much more common for a trial to fail to reach the expected number of events than to exceed that number.) They explained that “the trial was prespecified to last at a minimum 3.5 years and up to 5 years for each participant to allow minimal exposure that was meaningful for assessment of not only the primary endpoint but also broader safety assessments. The annualized event rate was two fold higher than projected perhaps because there were more patients with prior CVD recruited than planned. We recruited faster than expected and enrolled somewhat more patients than expected,” they said.
Sanjay Kaul reports the following conflicts of interest: Boehringer-Ingelheim and Novo Nordisk.