ESC 2018 (European Society of Cardiology)

August 25-29, 2018; Munich, Germany; Days #1-2 Highlights – Draft

Executive Highlights

  • We’ve learned enormously from ESC 2018 in Munich, and we’ve got so much to share in terms of new data and stellar insight on diabetes. What’s been most provocative? Karolinska’s Dr. Peter Ueda presented a cohort study of more than 34,000 new starts on SGLT-2s vs. GLP-1s, identifying an increased risk for amputations with the former (HR=2.23, 95% CI: 1.37-3.91). Notably, only 1% of the sample was taking Invokana, adding another layer of questions to the SGLT-2 safety discussion. In line with CANVAS, and very unsurprisingly, high-risk patients saw the most lower-extremity amputations (those with previous amputations, PAD) – how risk is defined in each trial, of course, isn’t standard.

  • CVOT news at ESC has not disappointed: Full CAMELLIA results pointed to resounding neutrality on CV outcomes and a very mild effect on CV risk factors with Belviq, but the obesity therapy did give a 19% reduction in new-onset diabetes (HR=0.81, 95% CI: 0.66-0.99). Now we’re looking forward to responder analyses for the 39% who achieved ≥5% weight loss.

  • An intriguing FOURIER post-hoc suggested that PCSK9 inhibitor Repatha may have greater effects in patients with metabolic syndrome; the study also found that these patients had substantially higher baseline CV risk to begin with (we do not view the latter as surprising).

  • We learned that AZ has added a new CVOT to the Farxiga lineup: DELIVER will investigate the SGLT-2 in patients with HFpEF, complementing Dapa-HF trial (now fully enrolled) in HFrEF – a smart move, given the greater unmet need in and increasing prevalence of HFpEF. As a reminder, HFpEF refers to heart failure with preserved ejection fraction; we are seeing more and more companies with SGLT-2s study this population, which seems to overlap more substantially with diabetes.

  • Other highlights included two compelling talks from Drs. Subodh Verma and Darren McGuire. In asserting that the CV benefits of SGLT-2s should extend to patients without diabetes, Dr. Verma explained the class’ effects on interstitial volume, myocardial function, and cardiac energetics (check out his recent paper with Dr. John McMurray for the full take). Dr. Darren McGuire boldly called for GLP-1/SGLT-2 dual therapy in patients with diabetes and ASCVD (atherosclerotic cardiovascular disease) whenever possible. He also advocated for an end to first-line metformin in diabetes care.

  • We’ve also enclosed coverage of five companies in the exhibit hall, where we were thrilled to see Novo Nordisk promoting the 26% reduction on three-point MACE with Ozempic, as per SUSTAIN 6 – a benefit recognized on the product’s EU label, though without a formal CV indication. In line with sky-high enthusiasm for SGLT-2s at this meeting, Jardiance was prominently featured in Lilly/BI’s large booth and the exhibit hall generally. 

Hallo from Munich, where we have been busy at ESC 2018! On the ground, we’ve been more-than-busy keeping pace with the amount of diabetes- and obesity-specific learning to be had in the capital of Bavaria. Below, you’ll find our top six highlights from the first two days of the meeting, plus exhibit hall coverage. Stay tuned for our second installment of ESC highlights, and see what we’re most looking forward to by reading our preview here!

Top Six Highlights

1. Observational Study Identifies Amputation Signal with SGLT-2 Inhibitors, Driven by High-Risk Subgroup (Prior Amputation or PAD); Only 1% of Sample Taking Invokana

Karolinska’s Dr. Peter Ueda presented an observational cohort study (n=34,426) identifying a ~doubled risk of lower limb amputations (HR=2.32, 95% CI: 1.37-3.91) among patients starting on an SGLT-2 inhibitor vs. a GLP-1 agonist. The SGLT-2 cohort experienced 40 total events (for comparison, there were 187 total amputation events in the CANVAS program), compared to 22 in the GLP-1 cohort (2.7 vs. 1.1 events per 1,000 patient years; the rate in CANVAS was 6.3 with canagliflozin vs. 3.4 with placebo). The audience audibly reacted to the amputation finding, and the presentation was referenced multiples times later in the day. Dr. Ueda emphasized that, as in CANVAS, this effect was driven by the subgroup of patients with a history of peripheral arterial disease (PAD) and lower limb amputations (HR=2.13, 95% CI: 1.05-4.32; 30.2 vs. 13.2 events/1,000 years). Furthermore, the absolute risk difference per 1,000 patient years was 1.5 in the total cohort (95% CI: 0.4-3.3), compared to 14.9 in the high-risk cohort (95% CI: 0.7-43.8) so a major difference there. The numerical breakdown of events was not immediately clear from the presentation. Of particular note, this study was mainly made up of patients on Jardiance and Farxiga, with very few on Invokana – specifically, 38% of those studied took Lilly/BI’s Jardiance (empagliflozin, n=6,506) and 61% took AZ’s Farxiga (61%, n=10,454), while 1% of SGLT-2 participants in this study were initiated on J&J’s Invokana (canagliflozin, n=254). In the aftermath of the CANVAS readout at ADA 2017, Lilly/BI were quick to distinguish Jardiance from Invokana and to emphasize that empagliflozin did not increase amputation rate in EMPA-REG OUTCOME, though as a reminder, amputations were recorded and adjudicated differently in CANVAS vs. EMPA-REG. Thought leader opinion on this issue is mixed – we’ve heard from Dr. Jay Skyler that he prefers to prescribe another SGLT-2 inhibitor rather than Invokana due to the amputation signal seen in CANVAS, while others, including Prof. Philip Home, have pointed to a possible class effect. Dr. Ueda made sure to underscore during Q&A that these results in no way confirm that amputations are a class effect of SGLT-2s, although they do add a new layer of information to the ongoing debate. He pointed to AZ’s DECLARE, the largest-ever diabetes CVOT at ~17,000 patients including a well-defined subgroup with PAD, as the next major potential data source on this question. All in all, both RCT and real-world data on SGLT-2s and amputations remains mixed: The real-world EASEL study found an increased risk of amputations with the SGLT-2 class (more patients were on Invokana rather than Jardiance or Farxiga), but OBSERVE-4D did not; both investigations were sponsored by J&J. At ACC 2018, Dr. Mikhail Kosiborod indicated that AZ’s CVD-REAL dataset may be used to look at amputations, and we’re more eager for that analysis than ever, given this dataset contains many hundreds of thousands of patients with diabetes. One thing is abundantly clear: the conversation and controversy over amputation risk with SGLT-2s hasn’t subsided though most would agree that it’s far better to reduce risk in patients where possible in the earlier stages of diabetes so that they don’t find themselves in the “very high risk” population. Dr. Naveed Sattar referenced the presentation as “very provocative” later in the day and we’ve heard attendees ask about this presentation during the Q&A of at least two entirely separate sessions.

  • Interestingly, the hazard ratio in this observational study (HR=2.23, 95% CI: 1.37-3.91) was numerically higher than the hazard ratio for amputations seen in CANVAS (HR=1.97, 95% CI: 1.41-2.75), though the confidence intervals clearly overlap and the comparators (GLP-1s vs. placebo) were different. In the end, we continue to agree with the majority of thought leaders that the CV, renal, and glucose-lowering benefits of SGLT-2s far outweigh this safety question for those who are not at advanced stages of their diabetes, provided that proper risk mitigation strategies are in place. What are these strategies? So far we’ve heard patient selection against PAD and prior amputations; diligent foot monitoring; and strong HCP- and patient-facing education. The latter is still a bit vague and we’re not sure what form this would take though we can imagine required education being quite appealing from a regulatory perspective.

  • This analysis also identified a signal for increased risk of DKA with SGLT-2 inhibitors vs. GLP-1 agonists (HR=2.14, 95% CI: 1.01-4.52). However, Dr. Ueda noted that there were few events overall (19 in the SGLT-2 group vs. 11 in the GLP-1 group; 1.3 vs. 0.6 events per 1,000 patient years). While insurance claims data has corroborated case reports of increased DKA risk with SGLT-2s in type 2 diabetes, our sense is that most thought leaders consider this risk extremely minimal in the vast majority of patients. If DKA does result from SGLT-2 use in type 2, it’s the result of a perfect storm of precipitating factors.
  • This study utilized national registries from Sweden and Denmark, considering all new users (≥35 years-old) of either SGLT-2s or GLP-1s from July 2013 to July 2016. Patients were excluded based on previous use of either class, severe renal disease, end-stage illness, drug misuse, and hospital admission in the past 30 days. Then, 17,213 patients from each group were propensity-score matched 1:1 based on 66 health and demographic covariates. GLP-1 was chosen as an active comparator based on comparable profile in terms of CV benefit and utilization in treatment algorithms, as well as the fact that GLP-1s haven’t been associated with any similar safety signals as SGLT-2s. We thought this was a very interesting choice, because it’s true that many HCPs are now asking how to determine ideal patient candidates for a GLP-1 agonist vs. an SGLT-2 inhibitor (as well as both agents in combination). Median follow-up was 270 days (~nine months), though 25% of participants had 508 days of available data (almost 17 months), which is longer than similar observational studies. A 90-day grace period between prescription refills (i.e. when the refill should have happened and when it actually happened) was allowed. Only 19% of patients in either cohort had a history of CVD at baseline, a key difference from previous SGLT-2 and GLP-1 CVOTs, though surprisingly, this doesn’t seem to have a significant bearing on these particular results. Sensitivity analyses in the Swedish cohort, involving adjustment for A1c, BMI, smoking status, albuminuria, and eGFR, gave results similar to the primary analysis but were not reported.

  • There were no signals for bone fracture (HR=1.11, 95% CI: 0.93-1.33), acute kidney injury (HR=0.69, 95% CI: 0.45-1.05), serious UTI (HR=0.89, 95% CI: 0.67-1.19), venous thromboembolism (HR=0.99, 95% CI: 0.71-1.38), or acute pancreatitis (HR=1.16, 95% CI: 0.64-2.12). Previously, an increased risk for bone fracture had been identified with Invokana in CANVAS (HR=1.55, 95% CI: 1.21-1.97) but not in CANVAS-R nor any other SGLT-2 inhibitor trials.

2. AZ Initiates DELIVER, a Second Heart Failure Outcomes Trial in HFpEF; Joins Dapa-HF Study in Patients with HFrEF

As announced on Thursday, AZ has expanded its heart failure program for SGLT-2 inhibitor Farxiga (dapagliflozin): The company will conduct a second, 4,700-participant outcomes trial – DELIVER – in patients with heart failure with preserved ejection fraction (HFpEF). The study is currently expected to complete in June 2021, according to Also per AZ, recruitment is now complete for Dapa-HF, the first Farxiga heart failure outcomes trial enrolling patients (n=4,500) with HF with reduced ejection fraction (expected completion December 2019). Importantly, both trials are enrolling patients with and without diabetes. They will investigate the same primary endpoint (time to first occurrence of CV death, hospitalization for heart failure, or urgent care heart failure visit), as well as a similar set of secondary endpoints (e.g. total number of CV death/hospitalization for heart failure events, all-cause mortality, etc.). While we didn’t necessarily expect this trial, it doesn’t come as an enormous surprise. As we’ve learned more about heart failure, specifically in the context of diabetes and obesity, we’ve wondered why AZ initially chose to conduct Dapa-HF in patients with HFrEF. Dr. Mikhail Kosiborod has explained that HFpEF is more so associated with adiposity and insulin resistance (the cardiac manifestation of obesity), while HFrEF typically results from a direct myocardial insult. Globally, as the obesity epidemic has climbed, HFpEF has become the more common version of heart failure, according to Dr. Kosiborod. What’s more, traditional myocardium-targeted heart failure therapies have not been as effective in HFpEF, due to the disease’s different etiology – this is one reason by Dr. Kosiborod and others are so excited about the potential of the SGLT-2 inhibitor class to fill an unmet need in heart failure prevention and treatment. To this end, Lilly/BI have initiated both the EMPEROR HF-Preserved and EMPEROR HF-Reduced trials to evaluate Jardiance’s (empagliflozin) potential in treating both types of heart failure. It makes sense that AZ would want to follow suit. Sanofi/Lexicon’s SOLOIST-WHF for SGLT-1/2 dual inhibitor sotagliflozin does not have ejection fraction restrictions. While we’re not certain what limitations would have been placed on a heart failure indication to come out of Dapa-HF alone (i.e. would it only indicate Farxiga for HFrEF?), we’re glad that AZ will now, pending positive results, have data to support the efficacy of Farxiga across the ejection fraction spectrum.

3. FOURIER Analysis Indicates PCSK9 Inhibitor Repatha May Have Greater CV Benefit in Patients with Metabolic Syndrome, Likely a Function of Higher Baseline CV Risk; Does Not Increase Risk of Progression to Diabetes

Dr. Prakash Deedwania presented a FOURIER post-hoc comparing patients with metabolic syndrome (n=16,361) to those without (n=10,981), and he showed how the former group may benefit more from Amgen’s PCSK9 inhibitor Repatha (evolocumab) in terms of cardioprotection. Notably, pre-existing metabolic syndrome confers higher CV risk in the first place. On the primary composite endpoint of CV death, MI, stroke, hospitalization for unstable angina, and coronary revascularization, participants with metabolic syndrome saw a 17% risk reduction with Repatha vs. placebo (HR=0.83, 95% CI: 0.76-0.91, p<0.001), resulting in a number needed to treat of 43 to prevent one MACE. Meanwhile, those without metabolic syndrome at baseline did not see a significant CV benefit with Repatha vs. placebo (HR=0.89, 95% CI: 0.79-1.01, p=0.069, NNT=59). On the key secondary outcome of three-point MACE (non-fatal MI, non-fatal stroke, or CV death), participants with baseline metabolic syndrome saw a striking 24% relative risk reduction on Repatha (HR=0.76, 95% CI: 0.68-0.86, p<0.001), while those without did not experience a significant benefit (HR=0.86, 95% CI: 0.74-1.01, p=0.07). Dr. Deedwania further detailed that patients with metabolic syndrome and no diabetes at baseline actually saw a 35% risk reduction on this secondary endpoint. LDL reduction with evolocumab, however, was comparable across subgroups: The cohort with pre-existing metabolic syndrome saw a mean 58% reduction in LDL with evolocumab vs. placebo, compared to a 61% mean reduction in the cohort without metabolic syndrome (p<0.00001 for both). Of course, FOURIER was not powered to identify benefit in either of these subgroups; as such, it’s encouraging to see a strong benefit in the metabolic syndrome subgroup, but we certainly cannot conclude that Repatha holds no CV benefit for patients without metabolic syndrome (and we should absolutely keep in mind that all FOURIER participants were in need of additional LDL-lowering on top of statins to get to target). It’s also worth noting that patients with metabolic syndrome to start had a ~20% higher baseline CV risk from which to improve: Cumulative risk for the primary endpoint was 15.8% in this “more sick” group (down to 13.5% with Repatha) vs. 12.9% in the “less sick” group (down to 11.2% with Repatha). Risk for the secondary endpoint was 10.9% in the metabolic syndrome group (down to 8.6% with Repatha), vs. 8.5% in the non-metabolic syndrome group (down to 6.8% with Repatha). It’s easier to demonstrate a benefit in a higher-risk population, which is why diabetes CVOTs have traditionally enrolled a majority of patients with baseline CVD or multiple CV risk factors: It takes less time to show superiority with a higher event rate, so it’s not surprising that the metabolic syndrome group would see “better” results, particularly given that median follow-up in FOURIER was only 26 months.

  • These results hold important lessons for the clinical use of PCSK9 inhibitors: Because both Repatha and Sanofi/Regeneron’s Praluent (alirocumab) are quite expensive, HCPs have struggled to figure out when they should be prescribed (i.e. which patients will benefit the most?) and to secure coverage from payers (insurers have rather complicated and lengthy prior authorization processes in place for the PCSK9 class). This analysis, especially if a similar effect can be observed in ODYSSEY Outcomes (the CVOT for Praluent), could offer valuable guidance to HCPs on which patients could benefit most from this class. That said, 60% of participants in FOURIER had metabolic syndrome, which isn’t the most precise refinement of the ideal patient population for PCSK9s ­– we imagine a similar proportion of ODYSSEY Outcomes’ enrollment would meet the metabolic syndrome criteria, as this CVOT also enrolled a high-risk population. Ultimately, Dr. Deedwania advocated for the use of clinical judgment above all else: Patients without metabolic syndrome shouldn’t be automatically ruled out for PCSK9 therapy. Individualized risk, he noted, is always going to trump all else – factors like prior events, surgeries, and comorbidities like CKD mean patients can be at very high risk for LDL-related adverse events without having metabolic syndrome per se. Presumably this is theoretical – what trumps everything in terms of US sales seems to be formulary status.

    • For comparison, the overall FOURIER analysis identified a 59% mean drop in LDL with evolocumab vs. placebo, from a baseline 92 mg/dl to 30 mg/dl by the end of the study. This translated to a 15% relative risk reduction (HR=0.85, 95% CI: 0.79-0.92) on the primary endpoint and a 20% relative risk reduction (HR=0.80, 95% CI: 0.73-0.88) on the key secondary endpoint. These analyses were adjusted for age, race, sex, history of diabetes, history of MI, heart failure, smoking status, eGFR >60 ml/min/1.73 m2, LDL, and high-intensity statin use at baseline.

  • Repatha was safe with respect to glycemic and safety outcomes. Among participants with metabolic syndrome at baseline, new-onset diabetes occurred in 12% of those on evolocumab vs. 11% on placebo (NS). There were no changes in fasting blood glucose. In light of still-lingering concerns related to type 2 diabetes risk stemming from statin therapy, it’s encouraging to see another lipid-lowering agent without this risk, though the novelty of PCSK9 inhibitors means long-term data is limited. The only imbalance in safety endpoints for this metabolic syndrome subgroup was on injection site reactions, which occurred in 2% of the evolocumab group vs. 1.5% of the placebo group (p=0.02).

  • At baseline, patients with metabolic syndrome – who comprised 60% of the FOURIER population – were a higher-risk group overall: Significantly more patients in the metabolic syndrome cohort had hypertension (88% vs. 68%) and mean baseline BMI was higher (31 vs. 27 kg/m2, p<0.01 for both). All patients enrolled in FOURIER had either prior stroke, prior MI, or symptomatic PAD (i.e. established CVD). Patients who met ≥three of five established NCEP III criteria were designated to have metabolic syndrome at baseline; these include (i) FPG ≥100 mg/dl (equivalent to prediabetes), (ii) waist circumference >102 cm (~40.2 in) in men and >88 cm (~34.6 in) in women, (iii) systolic blood pressure ≥135 and diastolic blood pressure ≥85 mmHg or history of hypertension, (iv) triglycerides >150 mg/dl, and (v) HDL-C <40 mg/dl in men and <50 mg/dl in women. Diabetes was defined as either A1c ≥6.5% or FPG ≥126 mg/dl. By criteria and metabolic syndrome groups, 77% vs. 28% met the FPG metric, 77% vs. 24% the waist circumference metric, 96% vs. 80% the hypertension metric, 61% vs. 9% the triglyceride metric, and 63% vs. 14% the HDL metric. 

4. Dr. Verma: CV Benefits of SGLT-2s Should Extend Outside of Diabetes; Outlines Multimodal Mechanism Involving Interstitial Volume Reduction Plus Improved Myocardial Function and Glucose Utilization

In a diabetes-meets-cardiology master class (sponsored by Lilly/BI), Dr. Subodh Verma argued that SGLT-2 inhibitors should confer cardioprotection in patients with heart failure who do not have diabetes. As Dr. Verma explained, patients with HF and patients with diabetes have a similar pathophysiology, CV benefits observed with the class to-date occur independently of glucose improvements, and the glucosuria, natriuresis, and other metabolic effects of SGLT-2s are maintained in patients without type 2 diabetes. In addressing potential mechanisms of SGLT-2 inhibitor-mediated cardioprotection, Dr. Verma gave his take on the near-ubiquitously referenced “volume effects.” He focused on a 2017 analysis contrasting dapagliflozin (AZ’s Farxiga) with the loop diuretic bumetanide to explain that the effects of SGLT-2s go far beyond diuresis (excess urine production): In contrast to bumetanide, dapagliflozin actually had little impact on circulating blood volume. Rather, he detailed, SGLT-2s differentially regulate interstitial volume; relative to loop diuretics, SGLT-2 inhibition reduces interstitial fluid more than it reduces blood volume, improving the interstitial edema present in congestive heart failure – see Dr. Verma’s recent paper with Dr. John McMurray in Diabetologia for more (Dr. Naveed Sattar later cited this as one of the two best SGLT-2 reviews he’s seen to-date!). Dr. Verma also tackled potential direct effects on the myocardium, also detailed in that manuscript, and suggested that SGLT-2 inhibitor molecules – through an entirely different cellular receptor – can reduce intracellular sodium and calcium levels by inhibiting the sodium-hydrogen exchanger in myocytes, which drives a downstream sodium-calcium antiporter. He emphasized that this has nothing to do with diabetes or the presence of glucose, and represents a mechanism by which SGLT-2s might improve myocardial function in patients even without type 2 diabetes. Finally, addressing the other leading SGLT-2 hypothesis – energetics and ketone metabolism – Dr. Verma gave a sneak-peak of an abstract presented the next day and just published in JACC BTS demonstrating that, in mice, empagliflozin improves cardiac energy production, but through improvements in glucose oxidation rather than a shift to ketone metabolism. In his assessment, “It seems like part of the energetics hypothesis is true, but through other mechanisms than ketone metabolism.” Finally, Dr. Verma cited evidence that SGLT-2s give both glucosuria and sodium excretion in patients without diabetes and also presented unpublished data that empagliflozin improves cardiac fibrosis in the absence of both diabetes and SGLT-2 receptors.

  • Regardless of mechanism, the field will have much more robust evidence on the breadth of the SGLT-2 class’ cardioprotection within the next few years. As Dr. Verma detailed, DECLARE (for AZ’s Farxiga), VERTIS-CV (for Merck/Pfizer’s Steglatro), and SCORED (for Sanofi/Lexicon’s sotagliflozin) will offer additional detail on the benefits (and risks) of SGLT-2s for preventing HF in patients with type 2 diabetes (including, through DECLARE, in a majority primary prevention population). But even more outcomes trials are investigating SGLT-2s for the treatment of heart failure, including the EMPEROR-Preserved and EMPEROR-Reduced trials for Lilly/BI’s Jardiance, Dapa-HF and the newly-announced DELIVER for Farxiga, and SOLOIST-WHF for sotagliflozin. Finally, trials in CKD are also evaluating heart failure, including CREDENCE for J&J’s Invokana (now complete), Dapa-CKD for Farxiga, and EMPA-KIDNEY for Jardiance. All in all, this totals over a dozen largescale outcomes trials of SGLT-2 inhibitors across the cardio-renal spectrum, in both HFpEF and HFrEF as well as worsening heart failure, and in people with and without diabetes.

5. Dr. McGuire Promotes SGLT-2/GLP-1 Combo Therapy for Patients with CV Disease; Challenges Continued Use of Metformin as First-Line Therapy, Based on Lack of CV Benefit

During the same Lilly/BI-sponsored session, Dr. Darren McGuire posited that the standard type 2 treatment algorithm should be flipped for patients with ASCVD: That is, either an SGLT-2 (empagliflozin or canagliflozin, based on current evidence) or a GLP-1 (liraglutide or semaglutide) should be initiated immediately, for CV risk mitigation and independent of glycemic control. Once the patient is stable on monotherapy, he explained, the other class should be added if at all possible. In other words, he promoted SGLT-2/GLP-1 co-administration. While there isn’t any long-term cardiovascular outcomes data on this combination, it seemed that Dr. McGuire suspects there is additional CV benefit to be gained from adding one on top of the other. In our view, additive CV benefit seems plausible given the distinct aspects of CVD improved with SGLT-2s vs. GLP-1s – one class primarily acting on heart failure, the other on atherosclerosis.

After SGLT-2 + GLP-1 or vice versa, Dr. McGuire recommended the addition of a low-dose of the TZD pioglitazone for people with diabetes and ASCVD. In his assessment, this algorithm is what the evidence supports. If, after all of this, further glucose-lowering is necessary, basal insulin (glargine or degludec) or other agents can be added; however, Dr. McGuire underscored that both the SGLT-2 and GLP-1 should be maintained. In many ways, this seems to be the direction that treatment algorithms are headed: ADA/EASD’s 2018 draft consensus statement recommends that patients with ASCVD who aren’t meeting glycemic goals receive a GLP-1 agonist or SGLT-2 inhibitor as second-line therapy, after metformin; if a patient is meeting his/her A1c target, a softer recommendation to consider adding one of these agents is given. More specifically, the guidelines will recommend GLP-1 if atherosclerosis predominates and an SGLT-2 if heart failure predominates, but the writing committee specifically noted that no CVOT supports the use of these classes in combination for CV risk reduction. Dr. McGuire has suggested a distinctly more aggressive approach – still grounded in data – to treating patients with ASCVD; of course, we recognize that guideline-writing committees face a set of restraints in defending their recommendations. For now, it seems patients that are sicker are much more likely to get an SGLT-2 or GLP-1 – from our view, this is not completely fair to patients.

  • We appreciated Dr. McGuire’s enthusiasm for SGLT-2s in particular. In discussing the 38% relative risk reduction for CV death observed with Lilly/BI’s Jardiance (empagliflozin) in EMPA-REG OUTCOME, he emphatically stated, “there is no therapy in all of cardiology that reduces CV death by this magnitude. Everyone looked for a reason not to believe this data, but the finding has withstood the most rigorous statistical assessments and I believe it is both true and robust.” In fact, at every ESC conference since EMPA-REG read out (at EASD 2015), we’ve noticed abundant enthusiasm from cardiologists for the SGLT-2 inhibitor class (and a perceived willingness to prescribe these agents primarily as CV drugs, at least from KOLs - while the mechanism of cardioprotection remains elusive, they all seem to be acknowledging it more than they previously did). Exhibit A: An audience comment from a Lilly/BI-sponsored corporate symposium at ESC 2016 – “cardiologists want to look at empagliflozin as a cardiovascular drug with a glucose-lowering effect.” From our view, there is no problem with this except perhaps to acknowledge that some may not need the glucose-lowering effect.

  • Dr. McGuire aggressively challenged the claim that, because CVOTs were conducted in a majority of patients on background metformin, all patients should be on first-line metformin. He argued that the primacy of metformin has not been seriously challenged due to its availability, tolerability, and affordability. As he put it, a dearth of compelling evidence on CV outcomes makes it unnecessary to keep the vast majority of patients on metformin. In Dr. McGuire’s view, plenty of patients in any given diabetes CVOT were not on metformin, including 24% in LEADER, 27% in SUSTAIN 6, 26% in EMPA-REG OUTCOME, and 23% in CANVAS. We do think this is an important question – one highly relevant to patients – that beckons further investigation. If a patient is on an SGLT-2 or GLP-1, does metformin actually add any benefit, glycemic or otherwise? To be sure, the medication has garnered a remarkable amount of attention lately, from the likes of ADA Chief Scientific, Medical, and Mission Officer Dr. Will Cefalu and Dr. James Gavin. We’d be keenly interested in post-hoc analyses looking at CVOT participants by metformin use.

  • Notably, during Q&A, Dr. McGuire explained that most cardiologists find it far easier to prescribe SGLT-2s than GLP-1s. His clinic routinely prescribes Jardiance to patients with ASCVD, and he acknowledged that there are some politics at play with the patient’s diabetes care provider: They use a templated message to explain that the patient is being started on an SGLT-2 for CV risk mitigation, but management of the patient’s glucose is still being deferred to the lead diabetes care provider; this is a sentiment and strategy we’ve heard before from Dr. Mikhail Kosiborod. On the other hand, Dr. McGuire explained, there are still sizable barriers to starting GLP-1 agonists in the cardiology clinic, largely due to the injection training required. While Dr. McGuire emphasized that the needles are very small and injection devices increasingly patient-friendly, Dr. Naveed Sattar noted that adjustments to insulin and SU doses are necessary when patients are given GLP-1, so depending on the patient, it’s often preferable to have the diabetologist run point – though cardiologists often do send recommendations that a patient start GLP-1.

6. Neutral Results from CAMELLIA CVOT for Obesity Therapy Belviq (Lorcaserin) Raise Questions About Relationship Between Weight Loss and CV Outcomes – But with Only ~4 Lbs Treatment Difference at 3.3 Years; Small Impact on CV Risk Factors

Brigham and Women’s Dr. Erin Bohula presented cardiovascular results from the CAMELLIA CVOT (n=12,000) of Arena/Eisai’s obesity therapy Belviq (lorcaserin); the results were simultaneously published in NEJM. This full results presentation followed topline data released by Eisai in July, which indicated non-inferiority vs. placebo on three-point MACE (non-fatal MI, non-fatal stroke, CV death). The hazard ratio for this primary endpoint was 0.99 favoring Belviq (95% CI: 0.85-1.14, 364 vs. 369 events) – resoundingly neutral, over a median follow-up of 3.3 years. On a secondary composite endpoint of CV death, MI, stroke, hospitalization for heart failure or unstable angina, or coronary revascularization, Belviq was similarly non-inferior but not superior to placebo (HR=0.97, 95% CI: 0.87-1.07, 707 vs. 727 events). Despite these neutral findings, Dr. Bohula noted that Belviq was associated with “modest” improvements in CV risk factors: Compared to placebo, the study drug lowered systolic and diastolic blood pressure by a mean 1 mmHg, reduced heart rate by a  mean 1 BPM, lowered LDL cholesterol by a mean 1 mg/dl, and reduced triglycerides by a mean 12 mg/dl. Among participants without diabetes (everyone enrolled had overweight/obesity plus atherosclerotic CV disease or multiple CV risk factors), Belviq gave a 0.2% placebo-adjusted drop in A1c. In participants with diabetes at baseline, a 0.3% placebo-adjusted drop was observed. In the CAMELLIA population overall, FPG also fell by 5 mg/dl with Belviq vs. placebo. While these biomarker changes were all statistically significant, we suspect many providers would not consider these clinically meaningful; after the topline release highlighted significant improvements in CV risk factors, we found these effect sizes somewhat disappointing. Moreover, given the safety issues surrounding the first generation of obesity drugs, it would have been a substantial win for Belviq to demonstrate CV efficacy that puts patient/provider concerns definitively to rest, also linking weight loss to improvements in hard outcomes. Our sense is that most were underwhelmed by these CVOT results, although Dr. Peter Nilsson offered a slightly more positive perspective in the discussant and clearly the NEJM simultaneous publishing was a major win.

  • Because the high-risk population in CAMELLIA was so well-treated with other therapies (e.g. 85% on statins, 85% on aspirin, 63% on beta blockers, 75% on ACE/ARBs), Dr. Nilsson suggested that it was simply too tall of an order for Belviq to demonstrate a further CV improvement. He cited a similar issue apparent in the Look AHEAD trial. Moreover, as we discuss below, the average treatment difference with Belviq vs. placebo at the end of the trial was only a little over four lbs – we struggle to see this as enough to show a CV benefit, especially since most thought leaders stick to 5% weight loss as the threshold for clinical benefits. Dr. Bohula commented that the lack of statistically significant superiority could be the result of either study duration or magnitude of the weight loss effect.

    • While we maintain that CV efficacy would have been a tremendous win for Eisai and for the obesity field, we’re still reassured by compelling CV safety. That 39% of participants achieved ≥5% body weight loss (also discussed further below) is encouraging, and we’re intrigued to learn more about who responds best to Belviq. We’re also optimistic that reassurance on safety can knock down at least one of the barriers to greater prescription and utilization of this obesity pharmacotherapy, though we note that under-diagnosis, undertreatment, cost, access, and stigma remain highly prevalent in this field.

    • In our view, CAMELLIA results in no way bode poorly for Novo Nordisk’s highly-anticipated SELECT CVOT for semaglutide in obesity, both because (i) the GLP-1 agonist is considered a far more efficacious weight loss agent and (ii) as a GLP-1, semaglutide carries the promise of cardioprotection via putative anti-atherosclerotic mechanisms.

  • Incident diabetes was the only individual outcome presented by Dr. Bohula that was significantly impacted (see picture): Belviq conferred a 19% relative risk reduction for new-onset diabetes vs. placebo (HR=0.81, 95% CI: 0.66-0.99). The Belviq group saw a 3.1% progression (174 individuals, or 8.5%, total) to diabetes per year, compared to a 3.8% rate (204 individuals, or 10.3%, total) in the placebo group. With an upper bound of the 95% confidence interval at 0.99, this result was only marginally statistically significant. Nonetheless, this is encouraging evidence that even modest weight loss can reduce type 2 diabetes incidence in a high-risk population. Going forward, we’d be curious to see if the benefit on this endpoint was affected by the degree of individual weight loss. In other words, did more weight loss correlate with more relative risk reduction for new-onset diabetes? Brigham and Women’s Dr. Benjamin Scirica will present full metabolic data from the CAMELLIA on October 4 at EASD 2018 in Berlin.

  • CAMELLIA also provides valuable evidence on Belviq’s long-term weight loss efficacy: At one year, the placebo group lost a mean of 1.4 kg (~3.1 lbs) vs. 4.2 kg (~9.3 lbs) weight loss in the Belviq group, for a treatment difference of 2.8 kg (~6.2 lbs, p<0.001). However, this treatment difference fell down to 1.9 kg (~4.2 lbs) at the end of the ~three-year trial, which one attendee said he struggled to categorize as “weight loss.” Indeed, CAMELLIA reflects an ongoing challenge with obesity pharmacotherapy: People respond differently to different medications, depending on a host of factors including the etiology of their obesity. In CAMELLIA, 39% of participants in the Belviq group achieved ≥5% weight loss (vs. 17% with placebo, p<0.001), and 15% reached ≥10% weight loss (vs. 5% with placebo, p<0.001). The flip side of this is that as many as 60% of patients treated with lorcaserin did not achieve the minimum 5% weight loss that’s generally considered necessary for other metabolic benefits. We’re very much looking forward to future responder analyses, which are indeed planned, Dr. Bohula confirmed. We can imagine that the modest weight loss achieved with Belviq was simply not enough to confer significant cardioprotection in a little over three years in the overall cohort. Is this different among those who achieved greater weight loss? We’d also add that a remarkable 83% (!) of patients randomized to semaglutide in Novo Nordisk’s phase 2 obesity study achieved ≥5% weight loss; what’s more, 27% of semaglutide-treated participants achieved ≥20% weight loss (we’re still astounded by that figure). That is to say, we remain optimistic that the data will reveal a strong link between meaningful weight loss and reduced CV risk. And again, CAMELLIA results do not shake our confidence in the potential for SELECT to be an important, positive obesity CVOT.

  • Adverse events were balanced across treatment arms (31% with Belviq vs. 32% with placebo), but lorcaserin did lead to more treatment discontinuations. In the Belviq arm, 7.2% of participants discontinued from the study due to an adverse event, compared to 3.7% in the placebo group. Additionally, Belviq was associated with higher rates of neurological side-effects, including dizziness (1.3% vs. 0.3%), fatigue (1.1% vs. 0.1%), headache (0.6% vs. 0.3%), and nausea (0.6% vs. 0.3%). As Dr. Bohula noted, weight loss drugs have a history of association with psychiatric side-effects; because lorcaserin is a centrally-acting serotonergic agonist, we imagine many providers are hesitant to prescribe the therapy for fear of inducing psychiatric or neurological side-effects, so this is encouraging long-term safety data. There was no significant signal for FDA-defined valvulopathy at one year, nor for pulmonary hypertension.

    • There was a significant signal for severe hypoglycemia with Belviq, though rates overall were low: Severe hypoglycemia with complications was experienced by 0.2% of the Belviq group vs. 0.1% of the placebo group. Numerically, this translated to 13 events in the Belviq group vs. four in the placebo group – so, a small signal but present nonetheless (any severe hypoglycemia is a big deal for patients and it should not be overlooked). It was clarified that this signal occurred in the group of patients with diabetes, specifically in patients on insulin and sulfonylureas.

  • CAMELLIA enrolled people with overweight/obesity (BMI ≥27 kg/m2), plus established CV disease OR type 2 diabetes with other CV risk factors. However, only 13% of participants had a BMI <30 kg/m2, which Dr. Nilsson was quick to criticize. Dr. Bohula emphasized the high comorbidity burden of the CAMELLIA study population: In both groups, mean BMI was 35 kg/m2, 75% had established CVD and the other 25% had multiple CV risk factors, 90% had hypertension, 94% hyperlipidemia, 57% type 2 diabetes, and 33% prediabetes. One-fifth had an eGFR <60 ml/min/1.73 m2, and median age was 64. Follow-up visits were conducted every three months for two years, then every four months until the end of the trial; median follow-up was 3.3 years. All participants were prescribed exercise and a reduced-calorie diet (i.e. lifestyle therapy). At 98% and 97% in the Belviq and placebo groups respectively, completion rates were high, though 12% and 13% did discontinue study drug on an annualized basis.

  • In his discussion, Dr. Nilsson focused on the importance of examining incident depression, smoking, and quality of life. Dr. Bohula explained that psychiatric status was not a prespecified adverse event in the trial design, but this data was caught through the adverse event database and will be reported over time. Data on smoking cessation was captured at baseline and then at multiple time points, and this will be evaluated by treatment group and overall (lorcaserin has been studied for smoking cessation). Finally – and unfortunately – no quality of life or patient-report metrics were incorporated into the CVOT.

Exhibit Hall


Amgen’s booth sprawled over the back corner of one exhibit hall, standing as a large monument to PCSK9 inhibitor Repatha (evolocumab). Two massive signs lit up the area, one a glowing red neon arrow that beckoned “Lower LDL-C / Reduce CV Risk” and another even more substantial sign with “LDL-C” printed to look like a crumbling stone. Across the aisle from Sanofi’s bright white booth for Praluent (alirocumab), Amgen’s booth brought more drama and a heavier tone, but the message was clear throughout: Repatha offers strong LDL-C lowering on top of statin therapy, and as evidenced in the FOURIER CVOT, this translates to a reduction in CV events among high-risk patients. Expectedly, Amgen is heavily leveraging positive FOURIER data and Repatha’s new indication for preventing heart attack, stroke, and coronary revascularization (approved May 2018 in Europe, December 2017 in the US) in order to appeal to cardiologists. Within the booth itself, Repatha pens were heavily showcased to cardiologists, and reps emphasized their ease of use and different dosing schemes (either twice or once a month); unfortunately, we didn’t see any demo pens floating around. Also disappointingly, we couldn’t find any info on Repatha access, which was readily available on a country-by-country basis in last year’s Amgen exhibit. Our sense is that Repatha’s CV indication has boosted sales and translated to substantially lower utilization management criteria in the US (e.g. from 45 to 33 questions on the CVS Health formulary), but the company did indicate during its 2Q18 update that it has started offering higher rebates, seemingly in exchange for better access in the US. Sanofi/Regeneron have publicly offered lower list prices in exchange for expanded access and inked a deal with Express Scripts to this end. Certainly, PCSK9 access and reimbursement still aren’t “good” but we’re encouraged by the commitment we’ve seen from both Amgen and Sanofi/Regeneron on improving the situation.


AZ boasted a mammoth booth, divided into two large sections focused on three drugs: SGLT-2 inhibitor Forxiga (Farxiga), antiplatelet Brilique (Brilinta), and hyperkalemia therapy Lokelma. Most of the square footage was dedicated to round sofas and tables for those taking advantage of two coffee stations and presumably conversation on these drugs, but there were a few sets of interactive technology where attendees could learn more about AZ’s cardiovascular, renal, and metabolism offerings. We used these touchscreen stations to delve into the various clinical trials studying Farxiga, currently underway in AZ’s DapaCare program: This includes the CVD-REAL real-world evidence program, the DECLARE CVOT, Dapa-HF in heart failure, Dapa-CKD in CKD, the DELIGHT trial for Qtern (dapagliflozin/saxagliptin), plus a few mechanistic studies (DEFINE-HF, PRESERVED-HF, and DAPASALT) – and, as of just last week: DELIVER (n=4,700), a brand new CVOT in patients with heart failure with preserved ejection fraction (HFpEF). DELIVER will complement Dapa-HF, which has just finished enrolling 4,500 patients with heart failure with reduced ejection fraction (HFrEF) and should complete by December 2019. We discuss this trial more fully above, but suffice it to say we think addition this is a very smart move by AZ, and we’re thrilled for another robust dataset on Farxiga and SGLT-2s.

There was an almost palpable excitement around the booth regarding DECLARE, the CVOT for Farxiga in type 2 diabetes, for which a topline readout is expected by the end of 2018. We’ve heard speculation, both at this meeting as well as ADA 2018, that results may be presented at AHA 2018 (Chicago, IL; November 10-12), though this has not been confirmed by AZ nor AHA. In the exhibit hall, reps emphasized that DECLARE’s ~60% primary prevention cohort and dual primary endpoints (traditional three-point MACE and CV death or hospitalization for heart failure) set the trial apart. They also explained that the trial will be “positive” if either endpoint is met, and AZ is optimistic that DECLARE can support an indication for both MACE and HF, in any patients with CV risk (but not necessarily CVD). This would be a first for the SGLT-2 class and a huge win for both AZ and patients with type 2 diabetes.


A large and prominently-located Lilly/BI exhibit was dedicated chiefly to Jardiance, with a smaller corner for BI’s anticoagulant Pradaxa (and accompanying coagulant, Praxbind). Augmented reality stations for these two therapies gave an interactive look at product information and data, and a large screen promoted the use of Jardiance for reduction of both CV death and hospitalizations for heart failure. In particular, it was emphasized that ESC guidelines formally recommend Jardiance for both of these purposes, and we note that the ADA/EASD 2018 consensus statement is also set to recommend the use of SGLT-2 inhibitors in patients with ASCVD, particularly when heart failure is the dominant pathophysiology (Jardiance will be preferred over J&J’s Invokana, as per the ADA 2018 presentation on these guidelines). The 38% risk reduction for CV death associated with Jardiance, as identified in EMPA-REG OUTCOME, has been on prominent display in Munich, and we were struck by a section of the booth highlighting that a patient with type 2 diabetes dies of CV causes every 18 seconds. While the ticking clock driving this point home may have been a bit blithe, it’s certainly clear that Lilly/BI want cardiologists to know about Jardiance’s impact on CV death. Encouragingly, the field does seem to be listening: Talks on SGLT-2s here have been remarkably well-attended and HCPs have been eager to ask questions. In some ways, cardiology seems a bit apprehensive about the unfamiliar mechanism (“is this just a diuretic?), what to do with other diuretic/anti-hypertensive medications, and use in patients with lower renal function. On the other hand, we also think some cardiologists may even be a bit incredulous about the benefits; as Dr. Darren McGuire put it on Saturday, “there is no therapy in all of cardiology that reduces CV death by this magnitude.” Ultimately, we can only imagine the use of SGLT-2s in cardiology will continue to increase dramatically, particularly as HF and renal benefits become better established.

Novo Nordisk

Novo Nordisk’s booth, while somewhat on the smaller side, was well-attended and dedicated entirely to promotion of GLP-1 agonist Ozempic (semaglutide). We were most struck to see the promotion of Ozempic for CV risk reduction: Large screens and the company’s hanging banner (which also featured the Victoza logo) highlighted the 26% reduction on three-point MACE observed with semaglutide in SUSTAIN 6 as evidence of the product’s “proven CV benefits.” This stands in stark contrasts to promotion and advertising in the US, where Novo Nordisk is strictly limited to discussing only CV safety when promoting Ozempic. These dissimilarities, of course, reflect the substantial differences between Ozempic’s FDA- and EMA-approved labels. The EU label for Ozempic acknowledges a statistically significant CV benefit as well as significant risk reduction for nephropathy with semaglutide vs. placebo, without granting a formal indication. On the other hand, the FDA presents SUSTAIN 6 only as a CV safety trial, with no allusion to cardioprotection or renal protection. For more on CVOT plans for semaglutide, see our Novo Nordisk 2Q18 report. Unfortunately, we couldn’t get any more color on how the Ozempic launch, access, and promotion are progressing in Europe; Novo Nordisk reps cited German regulations against advertising to non-prescribers.

Additionally, three different pens were on display in the booth, as per EMA’s 2Q18 decision on Novo Nordisk’s post-approval variation application to alter the intended pen offerings. EMA initially, in February, approved a single FlexTouch pen with all three doses (0.25 mg for titration, plus 0.5 and 1.0 mg therapeutic doses), but Novo Nordisk quickly followed that approval with a request to market two pens: one with 0.25 and 0.5 mg dosing, and another with 1.0 mg dosing, as in the US, as we understand it. However, EMA ended up approving three separate pens, one at each dose, and it’s not clear if Novo Nordisk or EMA prompted this final change; we imagine this final three-pen offering could actually prevent a fair amount of confusion and accidental mis-dosing, both too high and too low, among patients. Attendees could test out the pen for themselves on a pad mimicking human skin – in our view, a smart move at a conference full of clinicians less comfortable with prescribing injectable therapies (the booth also offered A1c tests, another cool strategy to improve this audience’s familiarity with diabetes). A few weeks ago, Novo Nordisk said Ozempic would soon be launched in EU markets (we expect launch in the first countries any day now), but we couldn’t confirm any more precise timing with reps.


Sanofi/Regeneron’s bright white and teal booth – dedicated entirely to PCSK9 inhibitor Praluent – stood across the aisle from Amgen’s more dramatic gray and red space, an intriguing contrast between the two PCSK9s on the market. An unusually inviting medical information section offered attendees info on ODYSSEY Outcomes, presented at ACC 2018 in March. Currently, a CV indication based on these data is under review (submitted 2Q18) at both FDA and EMA, and a decision is expected anytime from 4Q18 to 2Q19, depending on whether priority review is received.

A corner of the exhibit was dedicated to Sanofi’s new Life Companion app, an in-house development expected to launch in Greece, Slovakia, and Canada by the end of 2018 with a more expansive launch to follow in 2019. The app will offer dose reminders and alerts for Praluent, but patients will also be able to add reminders for any other medications they take, and alerts are apparently highly customizable; additionally, there’s a feature to track LDL-C over time (some features apparently require HCP authorization). Life Companion is also a more comprehensive lifestyle and wellness app, offering videos for healthy recipes, fitness tracking, and more. This isn’t Sanofi’s first foray into apps: The company launched GoMeals, a food and fitness tracker, way back in 2009, and Sanofi is also collaborating with Voluntis on a basal insulin titration app. Ultimately, we can see Life Companion offering some consolidation and convenience for a number of patients, but it doesn’t seem that the app actually brings any novel features or functionality, in diabetes, CV health, or otherwise – nonetheless, we’re always glad to see investment in patient support and hope it can be as well-directed as possible.

Additionally, we were excited to get a better sense of European access for Praluent, in light of the highly-restrictive utilization management criteria insurers in the US have put in place and the fact that commercial growth in Europe has been much stronger than in the US recently (albeit from a lower base). We spoke to a rep from Denmark, who indicated that access in EU countries has, unfortunately, been similarly limited in patient population: In Denmark, the Danish Medicines Agency apparently only covers Praluent for patients with an LDL-C of 4.0 mmol/L (~155 mg/dl) who have had a prior CV event – which she characterized as a very small number of patients. The LDL threshold in particular seems quite restrictive, as we imagine few patients who tolerate statins still have an LDL so high, and 155 mg/dl certainly wouldn’t be “at target” for any patient with CV disease, under any guidelines we’re aware of. However, the rep did seem optimistic that, if Praluent gains a CV indication and as the class becomes more established, better coverage will gradually follow. Indeed, Amgen’s Repatha has seen relatively stronger growth and a lowering of utilization management criteria following approval of its CV indication, though in our view the barriers still remain substantial. 


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