AHA (American Heart Association) 2018

November 10-12, 2018; Chicago, IL; Day #1 Highlights – Draft

Executive Highlights

  • This first installment of AHA 2018 highlights is rich with new data and analyses, headlined by results for the DECLARE CVOT of AZ’s SGLT-2 inhibitor Farxiga. In DECLARE, Farxiga gave a 17% relative risk reduction (HR=0.83, 95% CI: 0.73-0.95) on the co-primary composite of hospitalization for heart failure + CV death, driven by a significant 27% reduction on HHF. Very notably, that co-primary endpoint was well-balanced across the primary (n=~10,000) and secondary (n=~7,000) prevention cohorts. That said – or perhaps, because of these data – Drs. Javed Butler and Subodh Verma both argued that the “primary” and “secondary” distinction so commonly applied to CVOTs and atherosclerosis holds little water when it comes to HHF and CKD risk. As such, they said, SGLT-2s should be used more broadly for incident heart failure risk reduction, irrespective of effects on MACE outcomes.

  • Full data from Lilly/BI’s real-world EMPRISE study of Jardiance and SGLT-2s in HHF reinforces a class-wide and molecule-specific benefit on this endpoint, in patients with and without established CVD. Full data, following the release of topline results ~one week ago, demonstrate a 44% risk reduction on broadly-defined HHF with Jardiance vs. DPP-4 inhibitors (HR=0.56, 95% CI: 0.43-0.73) over ~five months (n=~35,000) – that number jumps to 53% when considering only specific HHF. The broader SGLT-2 inhibitor class (n=~225,000) was similarly associated with a 58% reduction in HHF-specific (HR=0.42, 95% CI: 0.35-0.50) and a 32% risk reduction on HHF-broad (HR=0.68, 95% CI: 0.63-0.74).

  • Significant buzz surrounded results from Amarin’s REDUCE-IT CVOT for Vascepa (icosapent ethyl), which gave a 25% relative risk reduction compared to placebo on a primary endpoint of CV death, MI, coronary revascularization, and unstable angina (HR=0.75, 95% CI: 0.68-0.83, p<0.00001). The secondary-prevention cohort (~70% of enrollment), saw a clear 27% relative risk reduction (HR=0.73, 95% CI: 0.65-0.81), vs. a non-significant 12% relative risk reduction in the primary prevention cohort (HR=0.88, 95% CI: 0.70-1.10). Among 8,175 participants, ~58% had type 2 diabetes, and benefit was very consistent between those with and without diabetes.

  • Saturday of AHA also saw the formal launch of ADA/AHA’s new multi-year initiative, “Know Diabetes by Heart,” which aims to raise awareness about and ultimately reduce heart failure, CV death, heart attack, and stroke in patients with diabetes. Lilly/BI, Novo Nordisk, and Sanofi are supporting the program, and the website currently offers discussion guides for patients, a sign-up link for ADA’s free Living with Type 2 Diabetes Program, and registration links for the partnership’s Ask the Expert Q&A series – we expect significant expansion in the coming weeks and months. See below for more on this important work, from ADA’s Chief Scientific, Medical, and Mission Officer Dr. Will Cefalu and AHA’s Chief Medical Officer for Prevention Dr. Eduardo Sanchez.

  • Finally, a late-breaker from Sanofi/Regeneron examined cost-effectiveness of PCSK9 Praluent utilizing ODYSSEY Outcomes data. The study, presented by Dr. Deepak Bhatt, found that the maximum annual price of alirocumab to achieve cost-effectiveness at $100,000 per QALY would be $13,357 when considering those with baseline LDL-C ≥100 mg/dl and only $2,083 when considering those with baseline LDL-C <100 mg/dl. These are particularly robust cost effectiveness data for a class that remains highly underutilized due to price and should offer clinicians some guidance on which patients stand to benefit most efficiently from Praluent.

Hello from Chicago, where we’re bringing you our first set of data-rich AHA 2018 highlights, including results from AZ’s DECLARE CVOT for Farxiga, Amarin’s REDUCE-IT CVOT for Vascepa, Lilly/BI’s real-world EMPRISE study for Jardiance, and Sanofi/Regeneron’s ODYSSEY Outcomes Economics Study on Praluent cost effectiveness.

Stay tuned for a second installment of highlights coming your way soon, including CARMELINA and CAMELLIA late breakers, a look at the future of SGLT-2s and heart failure, exhibit hall coverage, and more.

Top Six Highlights

1. Full DECLARE Results: 27% Risk Reduction on Hospitalization for Heart Failure with Farxiga Drives Superiority on Co-Primary Composite, Balanced Across Primary and Secondary Prevention; No Superiority on MACE or Components; Impressive Renal Effects and Safety Profile

Dr. Stephen Wiviott presented full results from the DECLARE CVOT to a standing-room-only crowd: AZ’s SGLT-2 inhibitor Farxiga (dapagliflozin) reduced risk for the co-primary endpoint of hospitalization for heart failure/CV death by 17% vs. placebo (HR=0.83, 95% CI: 0.73-0.95, p=0.005 for superiority), making it the first SGLT-2 to show significant benefit on heart failure. Dapagliflozin was non-inferior to placebo on three-point MACE (the other co-primary endpoint), though results trended in favor of the SGLT-2 inhibitor (HR=0.93, 95% CI: 0.84-1.03, p<0.001 for non-inferiority, p=0.17 for superiority). DECLARE is unlike any other SGLT-2 CVOT to date and carries significant implications for the class and for patients. Most pertinently, DECLARE solidifies the heart failure benefit of dapagliflozin specifically and SGLT-2 inhibitors broadly – if there was any question before – by including HHF in a primary endpoint (heart failure was only assessed as a secondary endpoint in Lilly/BI’s EMPA-REG and J&J’s CANVAS). In an interview with our team, AZ management underscored heart failure as the most prevalent and important CV endpoint for patients with type 2 diabetes. Indeed, patients with type 2 are at a 2.5-times greater risk of developing heart failure compared to a background population, and heart failure is just as costly to the healthcare system as cancer, according to AZ CEO Mr. Pascal Soriot.

DECLARE’s large (~60%) primary prevention cohort also makes the CVOT unique, and piques our curiosity about the potential for a CV indication that applies to the entire type 2 diabetes population rather than only to those patients with established CVD. Importantly, HHF/CV death was significantly reduced in both primary (HR=0.83, 95% CI: 0.71-0.98) and secondary prevention participants (HR=0.84, 95% CI=0.67-1.04). This consistency across primary and secondary prevention cohorts should support a new indication for reducing risk of heart failure (and possibly CV death) in a much broader population than ever before. To this end, AZ management reiterated to us that they plan to submit the data to regulatory agencies as soon as possible, likely in 1H19. Our biggest question, perhaps, is whether FDA will grant a dual HHF/CV death indication because this was the primary endpoint, or an indication for HHF only, the outcome which drove superiority. An indication solely for heart failure would parallel Jardiance’s indication solely for CV death, which drove overall MACE results in EMPA-REG. For context, AZ management explained that the decision to pair CV death with HHF in the study design was based (very rationally in our opinion) on the compelling results from the EMPA-REG trial indicating Jardiance’s reduction of CV death. AZ hypothesized that heart failure may be the driver of CV death improvements (as opposed to atherosclerotic events like stroke or MI), making HHF and CV death a natural primary endpoint pairing.

On a similar note, AZ emphasized to us that because the results of the trial are much more generalizable, it should give PCPs and HCPs greater confidence in prescribing Farxiga: They won’t have to spend as much time thinking about how to determine which patients may benefit from Farxiga, given the broad population studied. AZ management acknowledged the immense educational opportunity ahead of them with prescribers, both with respect to raising awareness of heart failure as the most prevalent CV complication of diabetes, as well as the general benefit of Farxiga for a wide variety of patients. Looking forward, the Dapa-HF (n=4,500 with established heart failure with reduced ejection fraction; results expected 2020) and DELIVER (n=4,700 with established heart failure with preserved ejection fraction; results expected 2020+) trials have the potential to extend Farxiga’s now-known benefit on heart failure to even more patients, namely those with established heart failure and without type 2 diabetes.

Simultaneous to Dr. Wiviott’s presentation, results from DECLARE were published in NEJM: “Dapagliflozin and Cardiovascular Outcomes in Type 2 Diabetes.” Key findings are summarized as follows: 

  • Hospitalization for Heart Failure & CV death: On the co-primary composite endpoint of CV death and hospitalization for heart failure (HHF), dapagliflozin demonstrated a significant 17% reduction compared to placebo (HR=0.83, 95% CI: 0.73-0.95, p=0.005 for superiority, event rate: 4.9% vs. 5.8%).

  • Three-Point MACE: On DECLARE’s second co-primary endpoint, dapagliflozin was non-inferior to placebo on three-point MACE, though results trended in favor of the SGLT-2 inhibitor (HR=0.93, 95% CI: 0.84-1.03, p<0.001 for non-inferiority, p=0.17 for superiority, event rate: 8.8% vs. 9.4%). We wonder if, over more time, the effect may have reached significance.

    • Moreover, dapagliflozin was not superior to placebo on any component of three-point MACE. It came closest on MI, with a hazard ratio of 0.89 (95% CI: 0.77-1.01). Both CV death (HR=0.98, 95% CI: 0.82-1.17) and ischemic stroke (HR=1.01, 95% CI: 0.84-1.21) showed resoundingly neutral results. See the table below for how these compare to results from other completed SGLT-2 CVOTs.

  • HHF: Farxiga gave a 27% reduction on HHF (HR=0.73, 95% CI: 0.61-0.88), making HHF the dominant driver of the primary endpoint result. We find this result unsurprising, given consistent risk reduction on HHF in both EMPA-REG OUTCOME and CANVAS, as well as in real-world data studies.

  • CV death: The hazard ratio for CV death was 0.98 (95% CI: 0.82-1.17), slightly favoring Farxiga over placebo. In a conversation with our team, Dr. Naeem Khan (VP of CVMD, AZ) suggested that this result in likely a function of population; indeed, data from other SGLT-2 CVOTs (below) support this notion.

  • All-cause death: Interestingly, a trend in favor of dapagliflozin on all-cause mortality (HR=0.93, 95% CI: 0.82-1.04) was driven by non-CV death (HR=0.88, 95% CI: 0.73-1.06). In an interview with our team, AZ management explained this is likely an artifact of the collective benefit of dapagliflozin on many factors (renal endpoints, hypoglycemia, severe adverse events, etc.) rather than a specific benefit on any one non-CV outcome.

As a reminder, DECLARE is the longest (median 4.2 years follow-up) and largest (n=17,160) SGLT-2 CVOT to date. The primary prevention cohort (multiple CV risk factors but no established CV disease at baseline) alone (n=10,189) is bigger than the total enrolled populations of both EMPA-REG OUTCOME (n=7,020, <1% primary prevention) and CANVAS (n=10,142, 34% primary prevention). We hope that annual follow-up on these patients, even at a basic level, will be possible. A few other points for consideration include:

  • Benefit on HHF/CV death was virtually identical in the primary and secondary prevention cohorts – a huge win for AZ. The hazard ratios for the composite were nearly the same in the primary (HR=0.83, 95% CI: 0.71-0.98) and secondary (HR=0.84, 95% CI=0.67-1.04) prevention groups (the latter narrowly crossed unity, though of course these analyses were not powered for superiority). The fact that benefit on HHF/CV death persists regardless of baseline risk is of paramount importance for prescribing practice and for AZ’s ability to garner a primary prevention indication. For three-point MACE, data was neutral in the primary prevention cohort (HR=1.01, 95%: 0.86-1.20) but trended in favor of dapagliflozin in the secondary prevention cohort (HR=0.90, 95% CI: 0.79-1.02). This raises the question of whether dapagliflozin’s benefit on atherosclerosis might rest primarily in patients with type 2 diabetes and established CV disease. To be sure, a post-hoc analysis of CANVAS presented at last year’s AHA found significant three-point MACE benefit to Invokana (canagliflozin) among secondary prevention participants but not among primary prevention participants; meanwhile, Invokana reduced HHF risk in both cohorts. It seems likely that SGLT-2s as a class offer MACE benefit only in those with comorbid diabetes/CV disease, but perhaps their benefit on heart failure extends across the spectrum of baseline CV riskDr. Itamar Raz has predicted as such, and DECLARE certainly supports this notion. AZ management emphasized this finding to us in an interview, noting that no trial has found atherosclerotic benefit with an SGLT-2 inhibitor in a primary prevention population (though none have tried).

  • On a composite renal endpoint comprising 40% decrease in eGFR to <60 ml/min/m2, end-stage renal disease (ESRD), or renal death, dapagliflozin demonstrated an extremely impressive 47% relative risk reduction (HR=0.53, 95% CI: 0.43-0.66). This furthers our understanding of nephro-protection as a class effect of SGLT-2 inhibitors, following positive CREDENCE news, and it bodes well for Dapa-CKD (dapagliflozin in chronic kidney disease patients, with or without type 2 diabetes). However, when the endpoint also included CV death (as was pre-specified), the benefit was lower, at a 24% relative risk reduction (HR=0.76, 95% CI: 0.67-0.87). Moreover, given hierarchical testing, our understanding is that this finding is technically exploratory.

    • With respect to DECLARE’s design, it should be noted that only patients with an eGFR >60 ml/min/m2 were enrolled in the trial (CANVAS and EMPA-REG both enrolled patients down to eGFR of 30 ml/min/m2), as FDA does not currently recommend dapagliflozin for use in patients with an eGFR below this threshold. As reflected in the table below, the average baseline eGFR in DECLARE (85 ml/min/m2) was 8-11 units higher than that of CANVAS and EMPA-REG OUTCOME. For context, Dr. David Fitchett explained at ESC 2017 that this original contraindication/recommended dose reduction for all SGLT-2 inhibitor products as eGFR falls below 60 ml/min/m2 is not for safety concerns, but rather for a presumed attenuation of metabolic efficacy. Notably, on systolic and diastolic blood pressure outcomes, SGLT-2 inhibitors have been shown to offer greater reductions with decreasing renal function. Farxiga just received a positive CHMP opinion for an expanded indication down to an eGFR of 45 ml/min/m2, based on results from the DERIVE trial, and FDA is currently reviewing a similar update in the US.

  • Dapagliflozin’s safety profile was strong. Most notably, there was no significant difference in amputations with dapagliflozin vs. placebo (1.4% vs. 1.3%, respectively), nor was there any imbalance in bone fractures (5.3% vs. 5.1%). Significant benefit with dapagliflozin was found in treatment-emergent severe adverse events (34% vs. 36%, p<0.001), major hypoglycemia (0.7% vs. 1.0%, p=0.02), and bladder cancer (0.3% vs. 0.5%, p=0.02). On the other hand, safety signals with dapagliflozin were seen in treatment-emergent adverse events leading to drug discontinuation (8.1% vs. 6.9%, p=0.01), DKA (0.3% vs. 0.1%, p=0.02), and genital infections (0.9% vs. 0.1%, p<0.001). Farxiga’s advantage on major hypoglycemia is particularly important to note, in our view. We have always thought this was a significant benefit of the entire SGLT-2 class, and this point isn’t raised as often as it could be with doctors. The same is true for weight loss benefit (see below).

  • Dapagliflozin showed a significant benefit on body weight compared to placebo (LS mean difference=1.8 kg, 95% CI: 1.7-2.0, p<0.001), stimulating weight loss from a baseline BMI of 32 kg/m2 on average. As we see it, HCPs need to be made more aware of the weight loss benefit associated with SGLT-2 inhibitors, because this is an outcome that matters to patients and that affects health and quality of life.

  • On metabolic outcomes, dapagliflozin demonstrated a significant reduction in A1c vs. placebo over 48 months (LS mean difference=0.42%, 95% CI: 0.40-0.45, p<0.001). This result is comparable to what was seen with empagliflozin in EMPA-REG OUTCOME (0.36% relative A1c reduction) and with canagliflozin in CANVAS (0.58% relative A1c reduction).

Comparison of Relevant Results from CANVAS (for Invokana) vs. EMPA-REG OUTCOME (for Jardiance)

This chart is for illustrative purposes only, as differences in enrolled population and trial design make it inherently impossible to draw comparisons between molecules. We continue to believe that a large outcomes trial examining SGLT-2 inhibitors and GLP-1 agonists in comparison to each other would be immensely valuable to patients, clinicians, and payers.


Farxiga (DECLARE)

Invokana (CANVAS)


Enrollment, (% w/ Baseline CVD)

17,160 (41%)

10,142 (66%)

7,020 (>99%)

Baseline eGFR required (mean)

>60 ml/min/m2 (85 ml/min/m2)

>30 ml/min/m2 (77 ml/min/m2)

>30 ml/min/m2 (74 ml/min/m2)

Median Follow-up

4.2 Years

2.4 years

3.1 years

Three-Point MACE


(HR=0.93, 95% CI: 0.84-1.03, p<0.001 for noninferiority, p=0.17 for superiority)

14% RRR

(HR=0.86, 95% CI: 0.75-0.97, p=0.0158 for superiority, p<0.0001 for non-inferiority)

14% RRR

(HR=0.86, 95% CI: 0.74-0.99, p=0.038 for superiority, p<0.001 for non-inferiority)

CV Death


(HR=0.98, 95% CI: 0.82-1.17)


(HR=0.87, 95% CI: 0.72-1.06)

38% RRR

(HR=0.62, 95% CI: 0.49-0.77, p<0.0001 for superiority)

Non-Fatal MI


(HR=0.89, 95% CI: 0.77-1.01)


(HR=0.85, 95% CI: 0.69-1.05)


(HR=0.87, 95% CI: 0.70-1.09)

Non-Fatal Stroke


(HR=1.01, 95% CI: 0.84-1.21)


(HR=0.90, 95% CI: 0.71-1.15)


(HR=1.24, 95% CI: 0.92-1.67) 

Hospitalization for Heart Failure

27% RRR

(HR=0.73, 95% CI: 0.61-0.88)

33% RRR

(HR=0.67, 95% CI: 0.52-0.87)

35% RRR

(HR=0.65, 95% CI: 0.50-0.85)

Renal Endpoints

24% RRR

(HR=0.76; 95% CI: 0.67-0.87)

40% RRR

(HR=0.60; 95% CI: 0.47-0.77)

46% RRR

(HR = 0.54)

Glycemic Control (A1c reduction)

0.42% (95% CI: 0.40-0.45)

0.58% (95% CI: 0.56-0.61)


Weight Loss

1.8 kg (95% CI: 1.7-2.0)

1.8 kg (95% CI: 1.51-1.70)

~2 kg

2. Drs. Javed Butler & Subodh Verma Promote Broad SGLT-2 Inhibitor Use for HF Reduction + Renal Function Irrespective of MACE Effect; Challenge Primary/Secondary Prevention Distinction in HF Paradigm; Call for Expansion of Diabetes Care Considerations

In his discussant following the DECLARE presentation, Dr. Javed Butler asserted that SGLT-2 inhibitors should be used in type 2 diabetes patients to reduce heart failure risk, irrespective of their effect on the classic three-point MACE. Supporting this claim was compelling data from a recent Swedish cohort study: Controlling A1c, smoking, LDL, blood pressure, and albuminuria attenuated risk for MI (HR=0.84, 95% CI: 0.75-0.93) and, non-significantly, stroke (HR=0.95, 95% CI: 0.84-1.07) relative to matched controls without diabetes. However, the hazard ratio for heart failure hospitalizations remained elevated (HR=1.45, 95% CI: 1.34-1.57) even after managing these other risk factors, suggesting that type 2 diabetes confers heart failure risk independent of A1c, smoking, LDL, blood pressure, and albuminuria. Moreover, once patients with type 2 diabetes develop heart failure, their mortality risk skyrockets and fewer than 10% survive five years, compared to ~80% of those with type 2 diabetes without heart failure. With this in mind, Dr. Butler stated that any therapy indicated for improving heart failure outcomes (as dapagliflozin likely will be, based on DECLARE results) should be a welcome addition to the diabetes care armamentarium. We couldn’t agree more, and we hope that as medications prove their efficacy on heart failure, they will be better utilized in diabetes care alongside lifestyle modification.

  • During a talk on heart failure and diabetes, Dr. Subodh Verma (University of Toronto, Canada) highlighted a recent comment piece he co-authored in The Lancet, which contextualized the use of SGLT-2 inhibitors in primary and secondary prevention populations. Entitled “Pump, pipes, and filter: Do SGLT-2 inhibitors cover it all?,” the editorial discusses a simultaneously published meta-analysis by Dr. Marc Sabatine et al. examining SGLT-2 CVOT data published to date, including the recently-presented DECLARE full results. In this comment, Dr. Verma emphasizes that these data demonstrate that cardio-renal benefits appear to vary across the diabetes continuum (see reproduced graphic below), juxtaposing those at risk vs. those with established CVD. From his perspective, risk reduction for MACE events with a drug seems to be limited to secondary prevention populations with established CVD, while the effect of SGLT-2 inhibitors in preventing heart failure hospitalizations and protecting renal function appears to apply to both secondary and primary prevention populations. On this note, Dr. Verma challenged the conceptual separation of primary and secondary prevention in relation to the risk of hospitalization for heart failure and renal disease. He pointed out that the meta-analysis of SGLT-2 CVOT data indicates a strong association between declining renal function and HHF, irrespective of atherosclerotic vascular disease. And so, while the distinction between primary and second prevention populations may be appropriate for atherosclerotic outcomes, Dr. Verma thinks that this framework is not useful when considering renal disease and hospitalization for heart failure. We’re curious to see what framework Dr. Verma would instead use to better differentiate between risk levels in these patients – or, does he not believe risk stratification is necessary at all and all type 2s should be treated with a therapy that lowers heart failure and renal risk? Overall, we’re thrilled to see Dr. Verma strongly support the broadened use of SGLT-2 inhibitors as first-line therapy after metformin regardless of whether the diabetes patient has established atherosclerotic vascular disease, chronic kidney disease, or heart failure.

  • Dr. Butler called for an expansion of the diabetes care paradigm beyond control of A1c, CV risk factors, and MACE endpoints, to include risk of heart failure and CKD. As he put it, “regardless of what primary endpoint was chosen in these clinical trials based on the guidance from regulators, all CV outcomes are important to the patients and doctors, including CKD, HF, and MI and stroke – and, even worse, CV mortality.” Drawing on Dr. Verma’s recent Comment, Dr. Butler pointed out that the terms “primary” and “secondary” prevention come from the perspective of atherosclerotic CV disease, which is closely related to the traditional primary outcome of three-point MACE and comes from regulatory guidance – however, it doesn’t necessarily reflect what’s most important to patients or the healthcare system. To this point, Dr. Butler cited a 2010 article by the venerable Dr. Rich Bergenstal, which demonstrated that congestive heart failure (CHF) is the single-most prevalent complication in patients with diabetes relative to those without diabetes. Bringing all of this together, Dr. Butler concluded his argument with the assertion that any quality improvement effort that does not consider HF and CKD will likely fall short of the potential benefit patients can receive. We certainly agree that heart failure and CKD cannot be ignored in diabetes care, and we agree with Dr. Butler that diabetes management must account for all aspects and comorbidities of the disease, not just those recommended to be used as primary outcomes.

3. Lilly/BI’s Real-World EMPRISE Study Confirms Benefit on Hospitalization for Heart Failure with Jardiance, SGLT-2 Class in US Cohort; Balanced Across Both Primary and Secondary Prevention

Following the release of topline results ~one week ago, Lilly/BI presented a jam-packed poster on the EMPRISE study, examining the real-world impact of Jardiance on heart failure hospitalizations. Importantly, two definitions of hospitalization for heart failure were used: HHF-specific refers to a primary HF discharge diagnosis, while HHF-broad refers to a HF discharge diagnosis in any position. Among people with type 2 diabetes starting Jardiance (n=17,539) or a DPP-4 inhibitor (n=17,539), those receiving Jardiance saw a 53% relative risk reduction (HR=0.47, 95% CI: 0.26-0.85) for HHF-specific over a brief mean follow-up of 5.4 months. HHF-broad was reduced by 44% (HR=0.56, 95% CI: 0.43-0.73) with Jardiance vs. DPP-4s. The 44% risk reduction was reported by Lilly/BI is the recent press release on EMPRISE. Very notably, these results were consistent between patients with (n=4,217) and without (n=13,243) established CVD at baseline, inasmuch as the confidence intervals overlapped; the HHF-broad risk reduction was 47% (HR=0.53, 95% CI: 0.39-0.72) among those with a history of CVD, compared to 65% among those without prior CVD (HR=0.35, 95% CI: 0.20-0.61). We would expect risk reduction to be greater in secondary prevention patients vs. their primary prevention counterparts, though in this case, we attribute the discrepancy to short follow-up time and the fact that this was an observational study. There was very little difference based on 10 mg vs. 25 mg dose of Jardiance, which gave 47% vs. 54% risk reductions, respectively. The specific event rate (IR/1,000 person-years) was 5.4 in the DPP-4 group vs. 2.1 in the Jardiance group (broad rates were 19.9 vs. 10.5, respectively). See the full table of results below.

  • Encouragingly, the broader SGLT-2 inhibitor class showed a similar 58% relative risk reduction for HHF-specific (HR=0.42, 95% CI: 0.35-0.50) and a 32% relative risk reduction for HHF-broad (HR=0.68, 95% CI: 0.63-0.74). This part of the EMPRISE analysis compared 112,017 SGLT-2 initiators to an equal number of DPP-4 initiators over a mean follow-up of 6.8 months. There was similar consistency between primary and secondary prevention cohorts at this level (see table below).

  • For comparison, EMPA-REG OUTCOME identified a 35% relative risk reduction on hospitalization for heart failure (HR=0.65, 95% CI: 0.05-0.85), a 38% risk reduction on CV death (HR=0.62, 95% CI: 0.49-0.77), and a 32% risk reduction on all-cause death (HR=0.68, 95% CI: 0.57-0.82) with Jardiance in people with type 2 diabetes and established CVD. The event rate (IR/1,000 person-years) was 14.5 in the placebo group and 9.4 in the empagliflozin group, roughly double that seen in the real world, which is a reflection of EMPA-REG OUTCOME’s enriched population.

    • Heart failure benefit with SGLT-2s has also been found with J&J’s Invokana in CANVAS, which found a 33% risk reduction on HHF (HR=0.67, 95% CI: 0.52-0.87), and AZ’s Farxiga in DECLARE, which found a 27% risk reduction on HHF (HR=0.73, 95% CI: 0.61-0.88). Real-world evidence has strongly corroborated this effect, most notably through the AZ-sponsored CVD-REAL program. CVD-REAL (n=309,046), in the US and western Europe, found a 39% reduced risk for HHF with all SGLT-2s vs. other diabetes therapies (p<0.001), while CVD-REAL 2 (n=470,128), a more multi-national cohort, found a 36% reduced risk for HHF (p=0.001).

  • It’s very encouraging to see Jardiance’s heart failure benefit corroborated in a real-world cohort, demonstrating that the HHF risk reduction translates to routine clinical practice. That said, when it comes to real-world evidence, the issue of confounding remains a significant point of discussion. To be sure, RWE has an important role to play, but in the case of EMPRISE, for example, there’s the inherent risk that people who received SGLT-2s/Jardiance were different from those who received DPP-4s in some way that cannot be controlled for. We remain very interested in learning more about how the EMPRISE design (comparing those on SGLT-2s vs. DPP-4s) compares to the CVD-REAL design (SGLT-2s vs. all other diabetes classes) and whether one design is more robust than the other. It is somewhat limiting that EMPRISE examines only US patients.

  • Study design: EMPRISE utilized data from two commercial databases (Optum Clinformatics and Truven Health MarketScan) and Medicare fee-for-service data, only from the US. The study cohort was matched 1:1 via propensity scoring; over 140 baseline characteristics were controlled. This specific analysis is based on data from August 2014 to September 2016 but, as Lilly/BI have previously stated, EMPRISE will eventually include data through 2019. No SGLT-2 inhibitor or DPP-4 inhibitor use in the prior year was allowed. Future analyses will include a greater number of patients and examine clinical, safety, and economic outcomes. Lilly has stated a goal to examine >200,000 patients from the same databases.

  • Baseline characteristics: The DPP-4/Jardiance and DPP-4/SGLT-2 cohorts were both very well matched across baseline characteristics, drug use, comorbidities, and healthcare utilization. Among those in the DPP-4/Jardiance analysis, mean age was 59 years old, 46% were female, and mean A1c was 8.5-8.6%; 8% were previously hospitalized (the time frame is unclear), while 25% had a history of CVD, 5% heart failure, 7% CKD, and 76% hypertension. Among the DPP-4/SGLT-2 cohort, mean age was 61 years old, 46-47% were female, and mean A1c was 8.6%; 8% were previously hospitalized, while 26-27% had a history of CVD, 6% heart failure, 7-8% CKD, and 76-77% hypertension.

4. REDUCE-IT Proves It with Impressive Full Results: Highly Significant Benefits on Primary (25% RRR) and Secondary Endpoints, Driven by Secondary Prevention Group; Benefit in Large Diabetes Cohort (~58% of Participants)

After weeks of buildup following a splashy release of positive topline results, the full presentation of REDUCE-IT CVOT (n=8,175) results did not disappoint, demonstrating highly significant results for Vascepa (4 g/day icosapent ethyl) vs. placebo on both primary and key secondary endpoints. Following the presentation, full results were also published in NEJM: “Cardiovascular Risk Reduction with Icosapent Ethyl for Hypertriglyceridemia.” Dr. Deepak Bhatt (Brigham and Women’s Hospital) took the stage to present these full results in front of a packed-beyond-capacity room. As a reminder, the study’s primary prevention cohort was comprised entirely people with type 2 diabetes over age 50 with ≥1 CV risk factor, so these results are highly relevant to people with diabetes, and somewhat lower CV risk at that; overall, ~58% of participants had type 2 diabetes. The secondary prevention cohort included those over age 45 with established CVD and comprised ~70% of the study population. Read on below for key takeaways on Vascepa’s impressive benefits on primary and secondary endpoints, effects on primary and secondary prevention cohorts, safety data, and a brewing controversy surrounding the use of mineral oil in REDUCE-IT’s placebo arm, which may have contributed to a higher rate of CV events:

  • Primary Endpoint Results: Vascepa delivered a 25% relative risk reduction compared to placebo. On the prespecified composite primary endpoint of CV death, MI, coronary revascularization, and unstable angina, treatment with icosapent ethyl gave a highly significant 25% relative risk reduction vs. placebo (HR=0.75, 95% CI: 0.68-0.83, p<0.00001). Dr. Bhatt individually read out each of the “zeros” – there were actually seven – in this miniscule p-value (accompanied by hearty laughter from the audience), underscoring just how solidly Vascepa conveyed this risk reduction. The absolute risk reduction was 4.8%, with a number needed to treat of 21 (95% CI: 15-33) to prevent one primary endpoint event over the median follow-up of 4.9 years. The total number of adjudicated primary endpoint events was 1,606, with a rate of 17% in the icosapent ethyl group vs. 22% in placebo.

    • Key Secondary Endpoint Results: Vascepa gave a 26% relative risk reduction on three-point MACE compared to placebo. On this prespecified key secondary endpoint (CV death, MI, and stroke), treatment with icosapent ethyl led to a highly statistically significant 26% relative risk reduction (HR=0.74, 95% CI: 0.65-0.83, p=0.0000006). This corresponds to an absolute risk reduction of 3.6%, with a number needed to treat of 28 (95% CI: 20-47) to prevent once three-point MACE event over 4.9 years.

  • Vascepa’s benefit was most strongly seen in secondary-prevention patients, although Vascepa still trended toward benefit in the primary prevention cohort. In REDUCE-IT’s secondary-prevention cohort (which represents ~70% of the study population), treatment gave a clear 27% relative risk reduction (HR=0.73, 95% CI: 0.65-0.81) on the primary composite endpoint. Meanwhile, a non-significant 12% relative risk reduction was seen in the primary prevention cohort, with the confidence interval crossing unity (HR=0.88, 95% CI: 0.70-1.10). To be sure, REDUCE-IT was not powered to demonstrate significant benefit in primary prevention alone, but given the somewhat narrow overlap between these two confidence intervals, our assessment it that there’s a clear trend toward greater benefit in higher-risk patients. Certainly, this doesn’t mean that a longer follow-up would not have shown benefit in primary prevention, but we can only speculate on that point. On this note, we’re curious to see how this data will play into discussions with regulatory agencies concerning Vascepa’s potential label expansion – will an indication still be pursued for a primary prevention population? This situation brings to mind the LEADER CVOT for GLP-1 agonist Victoza, which enrolled a similarly-sized primary prevention cohort and achieved significance on MACE strongly driven by the secondary prevention patients; FDA limited Victoza’s CV indication to those with established CV disease.

  • Notably, consistent benefit was seen regardless of diabetes status; REDUCE-IT enrolled ~58% of its population with patients with diabetes, including the ~30% primary prevention cohort. Patients with diabetes at baseline experienced a 23% relative risk reduction (HR=0.77, 95% CI: 0.68-0.87), while patients without diabetes saw a 27% relative risk reduction (HR=0.73, 95% CI: 0.62-0.85). We find this data to be highly compelling in terms of Vascepa’s potential in reducing CV risk in patients with diabetes. Importantly, the primary prevention cohort was entirely composed of patients with diabetes “and at least one additional risk factor.” A total of 4,787 patients with diabetes were enrolled in the study, with 2,394 from the primary-prevention cohort and the remaining 2,393 having established CVD and included in the secondary prevention group. Given fairly different effects between the primary and secondary prevention cohorts, however, we assume that the majority of risk reduction in the diabetes population was seen in those patients with established CVD.

  • Vascepa either demonstrated or trended toward benefit on all individual and composite ischemic endpoints in prespecified hierarchical testing. This includes separate analyses on CV death or non-fatal MI, fatal or non-fatal MI, fatal or non-fatal stroke, and other endpoints. See the figure below for the complete rundown. The benefit across the board on these prespecified endpoints is surely impressive, and stands in contrast to somewhat heterogeneous results from CVOTs we’ve seen with diabetes drugs (e.g., positive topline results are sometimes driven specifically by one individual components of three-point MACE).

  • Adverse events signals for Vascepa treatment included significantly higher rates of atrial flutter, diarrhea, peripheral edema, and constipation. Most notable was the higher rate of hospitalization for atrial fibrilization or flutter (3.1% vs. 2.1% for treatment vs. placebo, p=0.004); still, absolute rates were still low for these events. More broadly, overall rates of serious adverse events between the two groups were well balanced: Both rates of serious adverse events (30.6% vs. 30.7%) and serious adverse events leading to withdrawal of study drug (2.2% vs. 2.2%) were essentially identical. As a whole, we view Vascepa treatment as having a relatively clean side effect profile, strengthening its position as a potential CV risk lowering agent (we note that many patients remain very resistance to statin therapy).

  • Future studies will examine the mechanism of action for Vascepa’s CV benefit as well as its cost-effectiveness as a treatment option. Dr. Bhatt noted during his presentation that no comment can be made at this time on a potential mechanism of benefit for Vascepa. However, we do note that the simultaneous NEJM publication of the study posits that the timing of divergence for the Kaplan-Meier curves suggests a delayed onset of benefit –reflecting time needed for either triglyceride-related effects or other putative mechanisms to take hold. Antithrombotic, antiplatelet/anticoagulation, membrane-stabilizing, and coronary plaque stabilizing effects were also hypothesized. Dr. Bhatt remarked that detailed biomarker and genetic analyses are planned to investigate these mechanisms; no timetable was given for when results from these studies may be disclosed. Dr. Bhatt also mentioned that cost-effectiveness analyses are also planned – currently, Vascepa costs $3 per month for most patients using a co-pay card, and total cost of the medication for the system at large stands at ~$2,400/year. A low number needed to treat, in the 20s range, bodes very well for cost effectiveness.  

  • Amarin and Vascepa’s commercial success will partly hinge on how willing and able HCPs are to prescribe this relatively novel therapy. On this front, the reception of REDUCE-IT at AHA was certainly encouraging, as ~80-90% of the audience indicated in a post-session poll that they would be willing to prescribe Vascepa in high CV risk patients with moderately elevated triglycerides. While this is certainly a start, those on the front-lines of patient care – primary care practices – will be critical in getting Vascepa to as many patients who may benefit as possible. It appears as if Amarin realizes this as well: On the company’s 3Q18 earnings call, management explained that, in their expansion of targeted physicians for Vascepa educational marketing, around 85% are GPs, compared to 7% cardiologists and 5% endocrinologists.

  • Casting a shadow over these positive results is the emerging concern that mineral oil used in the placebo arm interfered with baseline statin therapy, causing an uptick in CV events. To better mimic the color and consistency of Vascepa, Amarin chose to use placebo capsules containing mineral oil. However, some cardiologists have suggested that the mineral oil may be one factor behind certain biomarkers in the placebo group trending the wrong way (specifically, LDL-C and C-reactive protein), possibly by interfering with statin absorption and artificially elevating rates of CV events in the placebo group. We’re still learning more on this front and, but our current understanding is that, although the use of mineral oil may have contributed to the impressive magnitude of the risk reduction seen in the Vascepa group, it’s unlikely that this was the driving factor behind the result. The decision to use mineral oil in the placebo arm was approved by FDA in discussions during the design of REDUCE-IT – we’ve heard more about the issue from Amarin and will be back with more soon.

ADA’s Chief Scientific, Medical, and Mission Officer Dr. Will Cefalu and AHA’s Chief Medical Officer for Prevention Dr. Eduardo Sanchez officially launched AHA/ADA’s multi-year partnership, “Know Diabetes by Heart,” offering details on the what, why, and how of the initiative during two events on Saturday. The partnership’s dedicated website already has a plethora of resources and educational materials devoted to the critical link between diabetes and CV disease. The initiative is funded by Novo Nordisk, Lilly/BI, and Sanofi.

  • What: Know Diabetes by Heart is an initiative to raise awareness about and ultimately reduce heart failure, CV death, heart attack, and stroke in patients with diabetes. The initiative targets patients, providers (first and foremost PCPs), and healthcare systems.

  • Why: We thought Dr. Cefalu summed it up particularly well in saying: “because the problem [of diabetes and cardiovascular disease] is too large to tackle alone.” He emphasized the extreme cost of diabetes on the healthcare system: “One out of every two Americans has a glucose problem. We can’t afford that. It cost $327 billion to care for diabetes in 2017.” That said, he continued, diabetes care is at an incredibly exciting point, with >40 new products on the market since he trained as an endocrinologist 30 years ago. Unfortunately, CV risk among those with diabetes is still a tremendous problem, and the increasing number of people with diabetes makes the epidemic all the more difficult to fight. Indeed, in the initiative launch video, ADA CEO Ms. Tracey Brown explained that 228 people with diabetes die every day, primarily from CV disease and complications. Dr. Sanchez offered color on how AHA is approaching the initiative, articulating, “The AHA has a new mission statement: To be a relentless force for longer, healthier lives. I believe this gives us license to do all sorts of things we haven’t done before. In my mind, this says that we must be working with ADA.” He explained that, for people over age 60 with type 2, CVD attenuates life expectancy by about 12 years – but CV risk is modifiable and life expectancy can be extended if we do certain things better, which is what this partnership aims to promote. In his assessment, patients need to be made more aware of their personal risk and providers need to discuss it more proactively.
  • How: Four large-scale aims are outlined below. In truth, details on the specifics of each of these goals were sparse during the launch (we heard quality improvement initiatives and streamlining navigation within healthcare systems as two ideas), though both Dr. Cefalu and Dr. Sanchez underscored the importance of accountability in sticking to these endeavors. Currently on the partnership’s website are discussion guides for patients to spark conversation with doctors, a sign-up link for ADA’s free Living with Type 2 Diabetes Program (printable guides, monthly newsletters, access to ADA’s Diabetes Forecast, and an online community with local events), and registration links for the partnership’s Ask the Expert Q&A series, which will answer patient questions asked by phone, online, or live.

    • (i) Raising awareness and understanding of the link between diabetes and cardiovascular disease;

    • (ii) positively empowering people to better manage their risk for cardiovascular disease, heart attacks, and strokes;

    • (iii) supporting health care providers in educating their patients living with type 2 diabetes on cardiovascular risk and increasing their patients’ engagement in prevention of cardiovascular deaths, heart attacks and strokes;

    • (iv) and following improved guidelines and treatment algorithms that take a patient-centered, individualized, and holistic approach to diabetes/CV disease care, such as the recently published ADA/EASD consensus statement.

  • As part of the partnership’s holistic approach, we heard three fantastic “mini-presentations” demonstrating how the intimate connection between CVD and diabetes extends beyond CVOTs:

    • UCSF’s Dr. Jonathan Butler demonstrated how cumulative psychosocial stress is associated not only with CVD but also type 2 diabetes in women over the age of 60, using data from the Women’s Health study. The study demonstrated that increasing quartiles of cumulative stress (adjusted for race, ethnicity, socioeconomic status, and CV disease risk factors) were associated with increasing risk of incident diabetes, with the highest hazard ratio for a quartile reaching ~2.5. Indeed, noted in the press release for the partnership’s launch were recent focus groups conducted by the ADA and AHA in September, which reinforced that many people living with type 2 diabetes experience distress related to the day-to-day management of this chronic disease. They also reported feelings of hopelessness, which can decrease the likelihood of taking action to reduce the long-term complications of diabetes. We find the multi-dimensional impact of diabetes on physical, mental, and emotional well-being too often understated and are encouraged by the holistic approach being taken by the partnership.

    • Yale’s Dr. Murrium Sadaf delineated the relationship between mental illness and diabetes in victims of sudden, unexpected death. Over the course of two years, ~30% of sudden unexpected death victims (aged 18-64) had diabetes. Mental illness was shockingly more prevalent in those with diabetes (57%) than those without diabetes (35%), with prevalence of mental illness higher in those with “severe diabetes” (64%) compared to “mild diabetes” (53%).

    • Columbia’s Dr. Nour Makarem presented an intriguing study suggesting that, when it comes to eating, it matters not only how much one eats but also when they eat it affects – temporal eating pattern impact diabetes and CV risk factors. The study population consisted of 12,708 cancer- and diabetes-free patients from the Hispanic Community Health Study/Study of Latinos. It was found that those who consumed ≥30% of their energy (in calories) after 6 pm had higher fasting glucose (93.7 vs. 93.0 mg/dl; p=0.001), insulin (12.4 vs. 11.6 mU/L; p=0.003), HOMA-IR (2.9 vs. 2.7; p=0.001), systolic BP (118.7 vs. 117.5 mmHg; p=0.004), and diastolic BP (72.2 vs. 71.0 mmHg; p<0.0001). While these differences don’t seem all that clinically meaningful in and of themselves (particularly for such a large sample size), the consistency between them paints an interesting pattern.

6. ODYSSEY Outcomes Economic Study Shows PCSK9 Praluent ~Three Times More Cost Effective in Those with Baseline LDL-C ≥100 mg/dl; Max Price of $6,319 Per Year at $100,000 Per QALY

In his second Saturday late-breaker, Dr. Deepak Bhatt presented results from the ODYSSEY Outcomes Economics Study of Sanofi/Regeneron’s PCSK9 inhibitor Praluent (alirocumab), demonstrating greater cost-effectiveness for the agent when used in those with baseline LDL-C >100 mg/dl. In the overall intention-to-treat population from Odyssey Outcomes, the maximum annual price of alirocumab to achieve cost-effectiveness at $100,000 per QALY (quality-adjusted life year) would be $6,319. However, that number climbed to $13,357 when considering those with baseline LDL-C ≥100 mg/dl and plummeted to $2,083 when considering those with baseline LDL-C <100 mg/dl – indicating that Praluent use is most cost effective in patients with higher baseline LDL-C. In other words, “The higher the baseline LDL-C, the higher the value of alirocumab…” These data led Dr. Bhatt to conclude that, considering the absolute clinical benefit and cost-effectiveness analyses, Praluent might offer good value in patients with recent ACS and baseline LDL-C of at least 100 mg/dl. This is the most robust cost-effectiveness analysis we’ve seen on PCSK9s to date, and it offers significant insight on which patients stand to benefit most efficiently from the class. Our sense has very much been that, despite strong thought leader enthusiasm for the LDL-lowering benefits of the two-member class, clinical enthusiasm and uptake have been hampered by cost barriers and utilization management procedures. Moreover, we’ve observed that the field simply isn’t sure which patients should receive PCSK9s, and physicians are wary of prescribing such costly agents to the “wrong” patients. To this end, we’re interested to see what more can be done to characterize patients who garnered the most benefit from Praluent in ODYSSEY Outcomes and also in FOURIER for Amgen’s Repatha.

  • Cost effectiveness is a fairly subjective measure. The $100,000 willingness-to-pay threshold is, as discussant Dr. Andrew Moran explained, the midpoint of commonly-accepted thresholds ranging from $50,000-$150,000 per QALY. The number equates with a payer’s willingness to invest in new treatments for gains in patient health; on the flip side, the lower the number is, the better it controls healthcare costs. Dr. Moran noted that PCSK9s have been a fairly unique situation in this regard: Cost-effectiveness analyses rarely impact drug pricing, but we’ve seen both manufacturers slash prices partly in response to these emerging data.

    • Dr. Moran corroborated the findings presented by Dr. Bhatt: He cited (i) another recent analysis setting value-based pricing at $100,000/QALY from $4,000 to $9,000 annually and (ii) ICER’s recommended Praluent price, at the same QALY threshold, of $5,300 if baseline LDL-C >100 mg/dl and $2,300 if >70 mg/dl (annually). The second of these is particularly well aligned with this analysis, given both used ODYSSEY Outcomes data.

  • PCSK9 inhibitors remain remarkably expensive and difficult to access. For reference, Amgen recently dropped the cost of PCSK9 inhibitor Repatha from $14,000 to $5,850 per year; Sanofi/Regeneron have agreed to lower Praluent’s list price in exchange for simplified access, but we’re not certain where the average list price currently stands. That said, we’re not exactly sure how list price relates to the QALY analysis, given complex rebating and discount schemes, though the point stands that PCSK9 inhibitors remain very, very expensive.

  • As a reminder, ODYSSEY Outcomes (n=18,924) enrolled patients with recent ACS (1-12 months) on a high-intensity statin dose; after a median follow up of 2.8 years, Praluent conferred a 15% risk reduction vs. placebo on the primary composite CV endpoint (HR=0.85, 95% CI: 0.78-0.93) and a 15% reduction on all-cause death (HR=0.85, 95% CI: 0.73-0.98). However, the magnitude of benefit was greater in those who started with LDL-C ≥100 mg/dl: Those participants saw a 24% risk reduction on MACE (HR=0.76, 95% CI: 0.65-0.87) and a 29% reduction on all-cause death (HR=0.71, 95% CI: 0.56-0.90). (Dr. Moran noted that analyzing effects by LDL-C subgroup was not pre-specified.) For comparison, the HR on all-cause mortality was 0.95 for those with LDL <100 mg/dl; of course, the trial wasn’t powered to show benefit, but the difference is quite striking. Our understanding is that the greater magnitude of benefit in the ≥100 mg/dl subgroup drives the superior cost effectiveness.

  • Dr. Bhatt identified four primary limitations of this analyses: First, modeling was done with the HR for all-cause mortality, which was nominally significant due to the study’s hierarchal testing plan. Second, US costs were applied to the study’s global population. Third, cost only includes event-based costs, not follow-up costs (though, he notes, this makes the results more conservative). Last, he noted that factors outside high LDL-C may identify additional patients in whom Praluent offers favorable cost effectiveness – LDL-C shouldn’t be the only factor considered. One key strength, he added, is that this analyses leverages very robust data from a long-term CV outcomes trial, allowing for fewer assumptions than other cost analyses.

  • In terms of methods, a specific cost was applied to CV death and non-fatal events, based on Medicare reimbursement rates adjusted to 2018 dollars. Costs were adjusted to commercial rates for those <65 years old. For CV death that cost was $20,225, compared to $18,862 for non-fatal MI without revascularization, $12,617 for non-fatal ischemic stroke, and a staggering ~$40,000 for ischemia-driven coronary revascularization or unstable angina. Importantly, cost of concomitant medications and follow-up were not included, which actually makes these costs an underestimate. This analysis estimated long-term survival probability by extrapolating based on age-specific mortality rates in the ODYSSEY Outcomes placebo arm, and the observed HRs from the trial were applied to this curve.


-- by Ann Carracher, Martin Kurian, Peter Rentzepis, Payal Marathe, and Kelly Close