March 10-12, 2018; Orlando, FL; Full Report - Draft

Executive Highlights

The American College of Cardiology hosted its 67th Annual Scientific Session & Expo in Orlando, FL, from March 10-12, drawing more than 18,000 attendees from 100+ countries. Things kicked off bright and early on Saturday with late-breaking results from Sanofi/Regeneron’s ODYSSEY Outcomes CVOT, demonstrating risk reductions on both a composite CV endpoint and all-cause death with PCSK9 inhibitor Praluent. In fact, ACC was rife with new data, including a CANVAS post-hoc showing that those with baseline heart failure benefitted more from J&J’s SGLT-2 inhibitor Invokana on certain CV endpoints, as well as very positive real-world CVD-REAL 2 results showing a 49% risk reduction on all-cause death with SGLT-2 inhibitors across diverse geographies. We were glad to note particularly strong attendance at a number of diabetes-focused sessions, plus a high level of excitement for SGLT-2 inhibitors. Below, you’ll find intriguing commentary on CVOT design/results from thought leaders including Drs. Mikhail Kosiborod, Lawrence Leiter, and Benjamin Scirica, alongside some compelling posters on SGLT-2 mechanisms, CVOT enrollment criteria, and more. Read on for our full report, beginning with the major themes (brand new!) we drew from this outstanding meeting.

Themes

Enthusiasm Abounds for SGLT-2 Inhibitors, Especially in Heart Failure, as Attention Shifts to Cardioprotection in Diabetes

• Based on what we saw at ACC, the cardiology community’s enthusiasm for SGLT-2 inhibitors is only growing with time, and reasonably so: A new CANVAS post-hoc and announcement of CVD-REAL 2 results highlighted this class’s remarkable benefits on heart failure and mortality. In a stellar late-breaking session, Dr. Mikhail Kosiborod presented the second iteration of AZ’s real-world evidence program, CVD-REAL 2, which included patients from often under-represented Asian countries. In this observational study, patients initiated on any SGLT-2 inhibitor saw a 49% risk reduction on all-cause death (HR=0.51, 95% CI: 0.37-0.70) and a 36% risk reduction on hospitalization for heart failure (HR=0.64, 95% CI: 0.50-0.82) compared to those started on any other glucose-lowering agent. These results weren’t a huge surprise given similarly positive outcomes in CANVAS and EMPA-REG OUTCOME, not to mention the first CVD-REAL readout, which found nearly identical risk reductions for the same endpoints. Notably, CVD-REAL 2 featured some lesser-known SGLT-2 agents (some only available in Asia) like ipragliflozin, tofogliflozin, and luseogliflozin – this lends further evidence to the notion that CV benefits are an SGLT-2 class effect. The second iteration of this real-world study also measured MI and stroke events, and found that SGLT-2s reduced risk by 19% and 32%, respectively. The magnitude and consistency of outcomes across investigations of SGLT-2s have been received with great positivity from the cardiology community, with very good reason. As Dr. Kosiborod laid out in a separate talk, the gap in CV outcomes between people with and without diabetes has not narrowed over the last 20 years, and ischemic heart disease remains the #1 killer of people with diabetes. In his view, CV death and heart failure are the two most important CV events for people with diabetes, given the latter’s high morbidity, which perfectly underlies his and the field’s enthusiasm for SGLT-2 inhibitors. Indeed, cardioprotection with diabetes drugs has emerged as a key theme of nearly every conference we’ve attend of late, and Dr. Kosiborod has played a leading role in bridging the fields of cardiology and endocrinology.

• New analyses and basic science are starting to shed light on which patient populations might benefit the most from SGLT-2 inhibitors, though definitive answers are still far away. Dr. Gemma Figtree presented a new CANVAS post-hoc indicating that patients with heart failure at baseline may have received even greater CV death/heart failure hospitalization benefit from canagliflozin than those who did not have prior heart failure (with the caveats that the study was not powered for this analysis, and that heart failure was physician-reported, not measured). While it’s encouraging to know that the sickest patients may experience an amplified benefit from SGLT-2 inhibitors, questions still abound as to which patients can reap CV benefits from both SGLT-2s and GLP-1s (more on this below). Along these lines, we noticed serious interest in SGLT-2 inhibitor mechanisms, which might aid understanding of how broadly applicable CV benefits are; mechanistic considerations surfaced during Dr. Lawrence Leiter's presentation and in at least three posters exploring ketone metabolism, inflammation, and proteomics. We also see a need for greater work to increase both SGLT-2 and GLP-1 uptake among cardiologists, PCPs, and even endocrinologists. To this end, we were thrilled to attend a 100% full, standing-room-only dinner symposium featuring Drs. Kosiborod, Robert Eckel, Laurence Sperling, and Ms. Melissa Magwire, offering insight and practical advice on diabetes management to an audience of primarily cardiologists.

Limitations of CVOT Design & Role of Real-World Evidence: What About Low-Risk and Primary Prevention Populations?

• While there was no shortage of enthusiasm at ACC for the cardioprotective benefits of diabetes drugs, commentary and analyses also pointed to the limitations of completed diabetes CVOTs, particularly with regard to their generalizability. In a standout talk on selecting diabetes drugs for CV risk reduction, Dr. Lawrence Leiter called current CVOTs “crash tests” for safety due to their relatively short duration and enrollment of predominantly high-risk patients. More than 80% of participants in LEADER and SUSTAIN 6 had baseline CV disease, compared to 73% in EXSCEL, one reason that has been cited for the latter study’s narrow miss on CV superiority. EMPA-REG OUTCOME and CANVAS enrolled >99% and 66% participants with baseline CV disease. Things are about to swing a bit in the other direction: Lilly’s REWIND CVOT for GLP-1 Trulicity and AZ’s DECLARE CVOT for SGLT-2 Farxiga have enrolled only 31% and 41% secondary prevention cohorts, though we wouldn’t call these a new generation of trials – they still enrolled all participants with relatively high CV risk. Indeed, a poster presented at ACC revealed that at least ~half of patients with type 2 diabetes would not have been eligible for any SGLT-2 inhibitor CVOT based on A1c, CKD, and CV status: Only 21% would have been eligible for EMPA-REG OUTCOME, 30% for CANVAS, and 46% for DECLARE. Dr. Leiter suggested a much longer duration CVOT that starts with patients at lower CV risk (although with a diabetes diagnosis, an individual’s risk for CV events is already significantly elevated) and exposes them to an SGLT-2 or GLP-1 for more time. We are extremely intrigued by this idea, are wondering who might fund this, because we see enormous potential to do public health good (reducing the prevalence of heart attacks, strokes, other diabetes complications, and death). What we would really love to see is a CVOT with all SGLT-2s and GLP-1s in prediabetes or very recently-diagnosed diabetes. We can imagine mechanisms by which each could give long-term cardioprotection (more on this below), and we also think it’s reasonable to expect even greater magnitude of benefit in the longer term when these agents are initiated in “less sick” patients to start. We acknowledge that this study would be quite expensive and long, likely even running out the patent expiry of SGLT-2s and GLP-1s, but we also think the lifesaving impact would be immense and well worth the investment.

• Dr. Mikhail Kosiborod outlined the ongoing “paradigm shift” in diabetes care, away from a hyper-focus on glucose control and toward a central focus on CV risk reduction. That said, we were also continually reminded at ACC that good CV outcomes data doesn’t exist for most of the patients HCPs will see in the real world. While post-hoc analyses offer some clues as to how primary prevention populations fare with cardioprotective diabetes drugs, these analyses are always underpowered. Moreover, post-hocs can’t say anything about those who were ineligible for being too “healthy” or too “unhealthy” to enroll in the trial. On the bright side, this evidence gap is receiving more attention within the diabetes and cardiology fields, and we expect to move toward answers over the next few years: Should SGLT-2s and/or GLP-1s be initiated earlier in the course of diabetes development, before patients have a CV event? What defines an ideal candidate for SGLT-2 therapy vs. GLP-1 therapy vs. both? We see a key role for real-world evidence to play (e.g. AZ’s CVD-REAL program), and down the line, trials of SGLT-2 inhibitors from AZ and Lilly/BI in patients with and without diabetes will shed valuable light on the breadth of SGLT-2s’ impact on heart failure and renal outcomes. Finally, we think it’s worth noting that Dr. Kosiborod very recently endorsed GLP-1 agonists for primary prevention of CV events – though this is certainly not commentary we hear often.

PCSK9 Inhibitors: ODYSSEY Outcomes Solidifies Cardioprotection as a Class Effect; Questions Remain as to Who Can – and Should – Access Them

• Positive results from the ODYSSEY Outcomes CVOT for Sanofi/Regeneron’s Praluent (alirocumab) were a meeting-wide highlight, but the readout introduced just as many questions as answers. We also heard continued commentary on the difficulty patients/providers have faced in securing reimbursement for these impressive but expensive agents. Full results are below; in short, alirocumab met its primary endpoint, giving a 15% relative risk reduction (HR=0.85, 95% CI: 0.78-0.93) on the composite primary endpoint of CV death, non-fatal MI, non-fatal stroke, or unstable angina requiring hospitalization. The FOURIER trial also found a 15% relative risk reduction with evolocumab (Amgen’s Repatha) on a similar endpoint, and our sense is that the field agrees that this class certainly offers additional cardioprotection in patients with CV disease when added to maximally-tolerated statin therapy. The most buzz at ACC, though, centered on an interaction analysis indicating that the highest baseline LDL tertile drove the CV and all-cause mortality benefits in ODYSSEY Outcomes overall. These patients, with baseline LDL >130 mg/dl, saw a 24% relative risk reduction on the primary endpoint (HR=0.76, 95% CI: 0.65-0.87). The lower two tertiles did not see significant benefit on their own, though the trial was not powered to show this. Will payers only cover a PCSK9 inhibitor for patients with very high LDL levels? Should they? Is risk stratification a useful tool for PCSK9 manufacturers to use right now in improving reimbursement? These were the immediate questions that arose following the ODYSSEY Outcomes results presentation at ACC. We do note that Dr. Steven Nissen argued the next week at ENDO 2018 that this effect was a result of dose confounding: Because trial protocol included titration down to a lower alirocumab dose when LDL was ≤25 mg/dl as well as switching to placebo once LDL fell to ≤15 mg/dl for two consecutive measurements, those who started at a higher LDL were more likely to receive the max dose of alirocumab, for longer. Dr. Paul Jellinger has also pointed out this design quirk as a limitation of the trial. We think this critique makes sense, particularly as no similar effect was seen in FOURIER. Nevertheless, the interaction exists in the data, and the implications for clinical practice, access, and reimbursement are real, though it remains to be seen how strongly payers, PBMs, and clinicians will latch onto this analysis – will they reserve Praluent for those with the very highest LDL, or will they recognize this confounding and the FOURIER data as evidence that the benefit exists across a wide spectrum of baseline LDL?

• Similar questions were the focus of two posters, showing that (i) the utilization management criteria used to gate access to PCSK9s are not actually associated with an increased risk of CV events and (ii) outcomes-based reimbursement for Repatha would not alter the drug’s cost-effectiveness. We found this first analysis particularly concerning; it demonstrated that four criteria (focused on statin and ezetimibe use) commonly used by commercial payers to decide who they’ll reimburse for PCSK9s actually had no association with occurrence of CV events. Theoretically, payers have been reserving PCKS9 inhibitors for patients at highest risk of CV events, but it now seems quite possible that they’re making coverage decisions based on irrelevant criteria. That said, just what PCSK9 reimbursement should look like remains a fairly open question. How should HCPs be prescribing PCSK9s to maximize their cost-benefit? Another analysis presented at ACC showed that, even with certain types of outcomes-based contracting, it’s hard to make Repatha as cost-effective and appealing to payers as statins. This echoed a more qualitative sentiment we heard throughout the meeting that, at their current price, PCSK9s are simply not an effective population-level solution, with much discussion focusing on identifying the “best candidates” for Praluent and Repatha. We do feel that this philosophy of reserving PCSK9s for the “highest risk” patients sets up a counterintuitive and even dangerous paradigm in which one has to be sicker to get a better, preventative drug – though we’re not certain ourselves what the best system is, because risk stratification could be a great way to get the ball rolling on better reimbursement. Interestingly, Sanofi/Regeneron have agreed to lower the list price of Praluent for payers/PBMs willing to reduce access barriers (i.e. implement simplified paperwork and fewer prior authorization requirements) – Express Scripts has already taken them up on the offer. We’re glad to think that more patients will soon be able to access the impressive LDL-lowering effects of Praluent, particularly as we hear more about the criticality of lipid-lowering and about the benefit of lowering LDL as much as possible.

Detailed Discussion and Commentary

Joint American College of Cardiology/Journal of American College of Cardiology Late-Breaking Clinical Trials

Cardiovascular Outcomes with Alirocumab after Acute Coronary Syndrome: Results of the ODYSSEY Outcomes Trial

Philippe Steg, MD (Paris Diderot University, Paris, France)

Dr. Philippe Steg presented the very first ODYSSEY Outcomes (n=18,924) results on Sanofi/Regeneron’s Praluent (alirocumab), revealing that treatment with the PCSK9 inhibitor led to significant 15% relative risk reductions on a primary composite MACE endpoint and on an observational endpoint of all-cause death. In the CVOT (median follow-up time 2.8 years), alirocumab gave a 15% risk reduction (HR=0.85, 95% CI: 0.78-0.93, p=0.0003 vs. placebo) on the composite primary outcome of coronary heart disease-related death, non-fatal MI, non-fatal stroke, or unstable angina requiring hospitalization, which translated to a 1.6% absolute risk reduction at four years. In the FOURIER trial, which read out at last year’s ACC meeting, Amgen’s PCSK9 inhibitor Repatha (evolocumab) gave a similar 15% relative risk reduction (p<0.0001) for the primary endpoint (CV death, non-fatal MI, non-fatal stroke, hospitalization for unstable angina, or coronary revascularization). The alirocumab group experienced 903 MACE events in total (occurring in 9.5% of participants) vs. 1,052 MACE events in the placebo group (11.1% of participants). The positive primary endpoint result was driven by significant decreases in non-fatal MI (HR=0.86, 95% CI: 0.77-0.96, p=0.006), ischemic stroke (HR=0.73, 95% CI: 0.57-0.93, p=0.01), and unstable angina (HR=0.61, 95% CI: 0.41-0.92, p=0.02), but not CHD death (HR=0.92, 95% CI: 0.76-1.11, p=0.38) with Praluent vs. placebo. Moreover, there was a 15% relative risk reduction for all-cause death in the trial (HR=0.85, 95% CI: 0.73-0.98, p=0.026), though this finding is considered observational because of the hierarchical testing structure. Numerically, there were 334 deaths in the alirocumab arm vs. 392 in the placebo arm. Dr. Steg seemed particularly enthused by the all-cause mortality results, and he opened his presentation by noting that neither high-intensity statins (compared to moderate intensity) nor ezetimibe on top of a statin affect mortality, despite reducing LDL levels. FOURIER actually trended in the wrong direction on all-cause death (HR=1.04, p=0.54), favoring placebo over Repatha, but this result was far from statistically significant. Dr. Steg referred to the p-value for all-cause death in ODYSSEY Outcomes as “nominal,” and we’ve heard hallway chatter that Sanofi/Regeneron will try to “massage” the mortality data in positioning this CVOT to guideline-writing committees, payers, and regulators. We’re very positive on these results overall, and we hope healthcare stakeholders view ODYSSEY Outcomes through the same lens. We have our fingers crossed that payers, in particular, appreciate cardioprotection as a class effect of PCSK9s now that FOURIER and ODYSSEY Outcomes have both reported; leading up to ACC, Amgen management was optimistic that a second positive CVOT would be good news for the class overall in terms of reimbursement prospects. Sanofi plans to file with FDA for Praluent’s CV indication in 3Q18.

• Additional secondary endpoints were also tested hierarchically in ODYSSEY Outcomes. Dr. Steg reported a 12% drop in coronary heart disease events (CHD death, non-fatal MI, unstable angina requiring hospitalization, ischemia-driven coronary revascularization) with alirocumab vs. placebo (HR=0.88, 95% CI: 0.81-0.95, p=0.001). Risk for major CHD events, including CHD death or non-fatal MI, fell by 12% with alirocumab (HR=0.88, 95% CI: 0.80-0.96, p=0.006), while risk for CV events (CV death, non-fatal CHD event, non-fatal stroke) fell by 13% (HR=0.87, 95% CI: 0.81-0.94, p=0.0003). The triple endpoint of all-cause death, non-fatal MI, and non-fatal stroke showed a 14% risk reduction with alirocumab vs. placebo (HR=0.86, 95% CI: 0.79-0.93, p=0.0003). Neither CHD death (HR=0.92, 95% CI: 0.76-1.11, p=0.38) nor CV death (HR=0.88, 95% CI: 0.74-1.05, p=0.15) were significantly reduced, though this data trended in the right direction, favoring alirocumab over placebo. Risk for ischemia-driven coronary revascularization was reduced by 12% with alirocumab vs. placebo (HR=0.88, 95% CI: 0.79-0.97, p=0.009), while risk for congestive heart failure requiring hospitalization was not impacted (HR=0.98, 95% CI: 0.79-1.20, p=0.84); both of these analyses were pre-specified.

• An interaction analysis showed that the ~one-third of patients with the highest LDL levels at baseline drove positive CV results in ODYSSEY Outcomes. Importantly, the magnitude of benefit in these patients is much higher (i.e. there’s more room for improvement when starting at higher LDL), which has implications for clinical practice, access, and reimbursement. Primary endpoint efficacy in pre-specified subgroups showed no interaction with age, sex, region, or time from event to randomization; however, there was a borderline significant interaction (p=0.09) with baseline LDL. Participants with LDL <80 mg/dl (n=7,164) or between 80-100 mg/dl (n=6,128) at baseline did not experience significant reductions in the primary composite endpoint with alirocumab vs. placebo (HR=0.86, 95% CI: 0.74-1.01 and HR=0.96, 95% CI: 0.82-1.14, respectively). Note that the 95% confidence interval spans unity, or 1.00, for both subgroups, indicating a non-significant result. On the other hand, participants with baseline LDL ≥100 mg/dl (n=5,629) experienced an impressive 24% risk reduction for the primary composite endpoint (HR=0.76, 95% CI: 0.65-0.87), although no subgroup was powered to show superiority alone, hence the wider confidence intervals. An analogous pattern was seen for all-cause death, where the subgroup with highest baseline LDL experienced a significant 29% relative risk reduction (HR=0.71, 95% CI: 0.56-0.90) and the overall interaction had a p-value of 0.12. During Q&A, Dr. Steg emphasized that this interaction was indeed borderline. He underscored that the trial was positive overall, with all baseline LDL subgroups trending toward benefit with alirocumab. The heterogeneity is somewhat disappointing – it would have been a tremendous victory if Praluent demonstrated cardioprotection regardless of a patient’s starting LDL, because this could have positioned PCSK9 therapy as broadly applicable and beneficial (Repatha conferred similar risk reduction for the primary CV endpoint across all baseline LDL subgroups in FOURIER). That said, we don’t want to undercut the importance of ODYSSEY Outcomes results, which do highlight cardioprotection as a class effect of PCSK9 inhibitors, showing a significant impact on hard outcomes with effective lipid-lowering. Dr. Steg concluded that these data support the long-term safety/efficacy of alirocumab, and that patients with high (≥100 mg/dl) LDL may benefit the most from Praluent.

  • We expect ODYSSEY Outcomes to have big implications for PCSK9 inhibitor reimbursement. In our view, the combination of FOURIER and ODYSSEY Outcomes absolutely points to a cardioprotective class effect. The borderline significant interactions in ODYSSEY related to baseline LDL could lead payers to reserve PCSK9 inhibitor coverage for only the highest-risk patients – but this isn’t necessarily bad news. Considering the abysmal state of reimbursement right now, Sanofi/Regeneron and Amgen have to start somewhere in improving coverage for their products, and risk stratification is a useful strategy to get the ball rolling and convince payers of the potential cost-savings. We can’t ignore that Praluent and Repatha are expensive drugs, and widespread use would levy a substantial cost on the healthcare system; Dr. Naveed Sattar elaborated on this point at ADA 2016. Eventually, we do hope to see PCSK9 inhibitors reimbursed for everyone who could benefit from a more efficacious lipid-lowering agent, but it will likely be a slow-and-steady climb to “ideal” coverage. Notably, given that diabetes also confers residual CV risk (on top of dyslipidemia), many type 2s may fall into the category of “highest-risk.” We’ve certainly heard thought leaders advocate for more consideration of PCSK9 inhibitor therapy in diabetes care. Look out for much more on the topics of patient selection and cost in the rest of our ACC coverage.

  • Very notably, ICER (Institute for Clinical and Economic Review) immediately issued a Preliminary New Evidence Update after Dr. Steg’s ACC presentation – this is terrific news for Sanofi/Regeneron. ICER will continue to refine its cost-effectiveness analysis on Praluent, and a final New Evidence Update is promised by May 3, 2018. In ICER’s interpretation, ODYSSEY Outcomes does confirm a mortality benefit with alirocumab treatment. Per the comment on page 7, ICER will not extend a more favorable cost-effectiveness calculation to Amgen’s Repatha (evolocumab).

• As in FOURIER, both the MACE and all-cause mortality curves started to diverge around one year. Both CVOTs showed a similar 15% risk reduction on their primary endpoint, although some expected a greater magnitude of benefit in ODYSSEY Outcomes due to the longer duration of follow-up. In presenting FOURIER at ACC 2017, Dr. Marc Sabatine asserted that it takes time for LDL-lowering to translate to clinical benefit. Median follow-up time in FOURIER was 26 months, compared to 34 months in ODYSSEY Outcomes. Dr. Steg highlighted this as a key difference between the studies. He also addressed population differences: FOURIER enrolled stable patients with atherosclerosis in contrast to ODYSSEY Outcomes’ recent ACS (acute coronary syndrome) population; Dr. Steg characterized background statin therapy as truly maximal in ODYSSEY (and perhaps less so in FOURIER). One panelist inquired during Q&A, should we have expected an even greater risk reduction with a sicker population and longer follow-up, rather than these results which are nearly identical to FOURIER? This remains an open question, although Dr. Steg did advance a couple hypotheses. First, a large fraction of early events in ACS are thrombosis-related rather than plaque-related, he explained, and it’s uncertain how a PCSK9 inhibitor would affect thrombosis. Second, maximizing statin therapy may hamper further LDL-lowering, hiding some of the impact of alirocumab vs. a placebo group also on high statin doses.

  • Following the ODYSSEY Outcomes presentation, we’ve heard the divergence of Kaplan-Meier curves characterized as slightly earlier in FOURIER. That said, during a Sunday session, Dr. Jennifer Robinson was quick to fight the suggestion that evolocumab and alirocumab are inherently different molecules. Rather, she attributed the divergence timing to differences in study population and trial design.

• ODYSSEY Outcomes provides excellent long-term data showing the durability of Praluent’s lipid-lowering effect. Average LDL was lowest in the alirocumab group at four months, when mean LDL was 38 mg/dl for participants on treatment vs. 93 mg/dl in the placebo group (treatment difference of 56 mg/dl, 63% lower). Over 48 months, mean LDL level drifted upward to 53 mg/dl with alirocumab, but this was still 48 mg/dl (or 55%) lower than the 101 mg/dl seen in the placebo group. During Q&A, Dr. Steg attributed most of this upward drift to participants stopping or de-intensifying background statin therapy.

• Study design: ODYSSEY Outcomes enrolled 18,924 patients who had experienced acute coronary syndrome (acute MI or unstable angina) in the past 1-12 months. All participants also had elevated levels of LDL cholesterol despite being on high-intensity statin therapy (or, they had documented statin intolerance). After a run-in period of 2-16 weeks on high-intensity/maximum-tolerated atorvastatin or rosuvastatin, participants who still met one of three high-lipid criteria were randomized to alirocumab 75 mg twice-monthly (n=9,462) or matched placebo (n=9,462). The study used a treat-to-target design, meaning HCPs were instructed to aim for a target LDL of 25-50 mg/dl in all patients. To maximize the number of participants reaching target range, investigators blindly titrated up or down, but also blindly switched to placebo if two consecutive LDL readings were below 15 mg/dl. A striking 7.7% of participants (730 individuals) were blindly switched to placebo – to our understanding, this could have partially muted the effect seen in the intention-to-treat analysis, although it also signals impressive lipid-lowering efficacy with Praluent (since so many people were experiencing consecutive readings <15 mg/dl). Median follow-up was 2.8 years in ODYSSEY Outcomes (compared to 2.2 years in FOURIER). Treatment was prematurely discontinued in 14.2% of alirocumab participants and in 15.8% of placebo participants. Only 23 people (~0.1%) were lost to follow-up. Dr. Steg also alluded to the possibility of artifacts related to titration, which may have caused patients to receive slightly varying doses of alirocumab.

• Baseline characteristics: Average age in both arms was 58, and 25% of participants were female. Diabetes was present in 29% of participants at baseline, 24% were current smokers, 65% had hypertension, and 19% had prior MI. We’re extremely eager for subgroup analysis focused on the patients with baseline diabetes, and we imagine these are coming soon from Sanofi/Regeneron (Sanofi management shared on the company’s 4Q17 earnings call that ODYSSEY Outcomes data was coming in the nick of time for ACC 2018, so it’s understandable that post-hocs won’t be ready until subsequent scientific meetings). Median time from acute coronary syndrome (ACS) to randomization was 2.6 months; of those ACS events, 49% were non-ST-elevated MI (NSTEMI), 35% were ST-elevated MI (STEMI), and 17% were unstable angina. Median LDL at baseline was 87 mg/dl, and 93% of patients met lipid criteria based on LDL cholesterol alone. At baseline, 89% of participants were on high-dose atorvastatin/rosuvastatin, and another 9% were on low to moderate doses; only 1% were on no lipid-lowering therapy. Drawing a contrast to FOURIER, Dr. Steg emphasized that this was truly a maximally-treated population in terms of statin therapy.

• Safety: There was no overall imbalance in treatment-emergent adverse events, including worsening diabetes or complications and new-onset diabetes. There were more local injection site reactions with alirocumab (3.8%) vs. placebo (2.1%), but this was expected, and we’re quite impressed by how safe/tolerable the PCSK9 inhibitor class has turned out.

Featured Clinical Research II: Interventional

Lower Risk of Cardiovascular Events and Death Associated with Initiation of SGLT-2 Inhibitors versus Other Glucose
Lowering Drugs - Real World Data Across Three Major World Regions with More Than 400,000 Patients: The CVD-REAL 2 Study

Mikhail Kosiborod, MD (Saint Luke's Hospital, Kansas City, MO)

Dr. Mikhail Kosiborod presented very positive CVD-REAL 2 results (just published in JACC), demonstrating that the impressive real-world CV benefit of SGLT-2 inhibitors observed in the first iteration of CVD-REAL extends into a population representing South Korea (n=336,644), Japan (n=67,780), Australia (n=27,442; only included in all-cause death analysis), Israel (n=19,472), Canada (n=16,064), and Singapore (n=2,726). In the multinational cohort (n=470,128), SGLT-2 inhibitors were associated with a 49% risk reduction for all-cause death (HR=0.51, 95% CI: 0.37-0.70, p<0.001 vs. other glucose-lowering drugs); 5,216 events were observed in total. SGLT-2s reduced risk for heart failure hospitalization by 36% (HR=0.64, 95% CI: 0.50-0.82, p=0.001 vs. other glucose-lowering drugs); 5,997 events were observed in total. For the composite outcome of all-cause death/heart failure hospitalization, SGLT-2s reduced risk by 40% (HR=0.60, 95% CI: 0.47-0.76, p<0.001 vs. other glucose-lowering drugs); 9,788 events were observed in total. Dr. Kosiborod also presented original CVD-REAL results at last year’s ACC meeting, reporting 51% relative risk reduction for all-cause death with SGLT-2s, 39% relative risk reduction for heart failure hospitalization, and 46% relative risk reduction for the composite (all p<0.001) – we note similar findings from CVD-REAL and CVD-REAL 2, adding to the robustness of this real-world evidence. Dr. Kosiborod continued by showing heterogeneity p-values of <0.001 for all three endpoints in CVD-REAL 2 across regions; the only individual region to cross the line of unity (into non-significance) was Singapore for all-cause death (HR=0.75, 95% CI: 0.38-1.47) and heart failure hospitalization (HR=0.62, 95% CI: 0.38-1.02), and we note the wide confidence intervals due to small sample size; with n=2,726, Singapore had the lowest number of participants among all countries. We continue to be very positive about the results coming out of this AZ-sponsored program, which supports both a cardioprotective class effect for SGLT-2 inhibitors and the real-world benefit of these agents outside the rigorous clinical trial setting. CVD-REAL and CVD-REAL 2 additionally show the advantages of SGLT-2s in a broad, diverse population. Even with different event rates between populations with and without CV disease at baseline, similar benefit was observed with SGLT-2s in CVD-REAL 2 across all endpoints and regions. Dr. Kosiborod acknowledged that real-world studies come with inherent risk of residual confounders, though sensitivity analysis minimizes this risk. Events were not adjudicated, and safety was not evaluated. It’s important to keep in mind that our experience with SGLT-2 inhibitors in the real world is still relatively short – will these risk reductions hold up over decades? In the shorter-term, these results have only made us more excited for the DECLARE CVOT on AZ’s Farxiga (dapagliflozin), expected to read out in 2H18, and for the now-underway Dapa-HF trial in chronic heart failure. Many thanks to AZ for sponsoring such a groundbreaking real-world program!

• Of particular note, CVD-REAL 2 also found a significant 19% risk reduction for MI with SGLT-2 inhibitors (HR=0.81, 95% CI: 0.74-0.88, p<0.001) and a 32% risk reduction for stroke (HR=0.68, 95% CI: 0.55-0.84, p<0.001). All five individual regions saw significant reductions in stroke with SGLT-2s, compared to only two regions for MI. Dr. Kosiborod positioned the stroke benefit as all the more notable considering a higher prevalence of stroke in Asia, and we’re certainly excited about the implications of reducing stroke risk in a population where diabetes is becoming so much more prevalent. Dr. Kosiborod characterized these athero-thrombotic events as “not well studied” in previous CVD-REAL datasets, and he thus explained that these results should be seen as more reassuring than anything. Following EMPA-REG OUTCOME results at EASD 2015 (this was the first positive diabetes CVOT, for Lilly/BI’s SGLT-2 inhibitor Jardiance), there was some talk about the above 1.00 hazard ratio for non-fatal stroke (HR=1.24, 95% CI: 0.92-1.67; p=0.16), even though this did not meet statistical significance. To this end, we think it’ll be valuable to have a large real-world dataset showing far fewer strokes among patients starting SGLT-2 therapy vs. other diabetes drugs, to address any lingering concerns among real-world HCPs, patients, or payers.

• CVD-REAL 2 compared new users of SGLT-2 inhibitors vs. new users of other glucose-lowering drugs using propensity scores to match 235,065 sets of patients 1:1 in each participating country. National registries, insurance claims, and EMR records combined for the total study population of 470,128. Dr. Kosiborod explained that an intent-to-treat analysis was used, so treatment discontinuation was not considered. He shared that mean follow-up was ~400-500 days. He reviewed baseline characteristics to show that the cohorts were well-matched, with the biggest differences being CV disease prevalence (2.7% higher in the SGLT-2 inhibitor group) and anti-hypertensive therapy (1.9% higher in the SGLT-2 inhibitor group); this makes sense, since real-world HCPs are more likely to prescribe an SGLT-2 to a patient with previous CV events or high blood pressure. Dr. Kosiborod also highlighted high metformin (74%-75%) and, unfortunately, high sulfonylurea (51%-52%) utilization across the entire real-world study population.

• Dapagliflozin accounted for 75% of SGLT-2 exposure time in CVD-REAL 2, which reflects the ex-US dominance of AZ’s Farxiga. Empagliflozin (Lilly/BI’s Jardiance) accounted for 9% of SGLT-2 exposure time, while ipragliflozin accounted for 8%, canagliflozin (J&J’s Invokana) for 4%, tofogliflozin for 3%, and luseogliflozin for 1%. This stands in contrast to the original CVD-REAL conducted in the US, UK, Norway, Denmark, Sweden, and Germany, in which 53% of participants were on canagliflozin (first-to-market in the US), 42% on dapagliflozin (first-to-market in Europe), and 5% on empagliflozin. Even in that dataset, 92% of EU patients were taking dapagliflozin and 76% of US patients were on canagliflozin, mirroring the first-to-market therapies in those geographies.

• Dr. Kosiborod also shared hopes to use the CVD-REAL dataset to analyze amputations in the future. He does think this is a good venue in which to do so, but understandably stipulated that this must be undertaken with a great deal of care. We remain convinced that amputation risk should be manageable in the real world, primarily with careful patient selection (e.g. Dr. Lawrence Leiter suggested on ACC day #1 that HCPs avoid prescribing canagliflozin to people with peripheral artery disease), diligent monitoring of the feet, and strong patient education. Given the low base rate of amputations, even in the canagliflozin arm of CANVAS, we don’t think this risk holds a candle to the tremendous benefits of Invokana (and other SGLT-2s) on heart failure, other CV events, and overall mortality. That said, the J&J’s sponsored real-world study EASEL did find a higher amputation rate in patients on SGLT-2s vs. other diabetes drugs, so we can’t ignore that this is a safety issue that needs to be concertedly addressed.

Canagliflozin for Prevention of Heart Failure In Type 2 Diabetes: Results From The CANVAS Program

Gemma Figtree, PhD (University of Sydney, Sydney, Australia)

In an absolutely packed late-breaking session, Dr. Gemma Figtree presented a new post-hoc analysis of CANVAS showing that certain CV benefits with SGLT-2 Invokana (canagliflozin) were greater in patients with a history of heart failure. These data were simultaneously published in Circulation, and were also called out in a dedicated press release from J&J. For the composite endpoint of CV death or hospitalization for heart failure, CANVAS participants with a history of heart failure at baseline (n=1,461) experienced a 39% relative risk reduction (HR=0.61, 95% CI: 0.46-0.80) with 203 observed events, while participants without prior heart failure (n=8,681) did not experience a significant risk reduction (HR=0.87, 95% CI: 0.72-1.06) with 449 observed events. The p-value for this interaction was significant at 0.021. For reference, the full CANVAS population experienced 22% risk reduction for CV death or heart failure hospitalization with canagliflozin vs. placebo (HR=0.78, 95% CI: 0.67-0.91). It’s encouraging that the effect in the no heart failure group trended in the right direction, favoring Invokana, despite being underpowered. Moreover, these are exciting results for patients with diabetes/heart failure (which overlap substantially in the real world), who may garner even greater benefit from canagliflozin than initial results indicated. Heart failure has been too-long ignored as a CV complication of diabetes, but it’s now gaining more attention because of the protective effects of SGLT-2 inhibitors, including Invokana. We certainly view heart failure in diabetes as an unmet need, and it would be a huge win to have a therapy class (SGLT-2s) that demonstrates convincing benefits on hard outcomes (!) in this patient population.

• There were no significant, or even borderline significant, interactions between heart failure history and three-point MACE, MI, stroke, all-cause death, or serious decline in renal function. When splitting the study population by heart failure status, enough power was lost such that few significant effects were seen on these endpoints in either cohort individually, but the group with baseline heart failure did retain a statistically significant benefit on all-cause mortality (30% relative risk reduction, HR=0.70, 95% CI: 0.51-0.96), while the group without baseline heart failure retained a statistically significant benefit on the composite renal endpoint (48% relative risk reduction, HR=0.52, 95% CI: 0.37-0.72).

• Dr. Figtree also presented new data showing that canagliflozin decreased risk for the composite of fatal heart failure or heart failure hospitalization by 30% vs. placebo (HR=0.70, 95% CI: 0.55-0.89). For comparison, canagliflozin reduced risk for heart failure hospitalization alone by 33% vs. placebo (HR=0.67, 95% CI: 0.52-0.87), according to the initial full results presentation at ADA, so this post-hoc finding lends additional evidence for the SGLT-2 inhibitor’s heart failure benefit. Experts in diabetes and cardiology alike are enthusiastic about the SGLT-2 class and its applications in heart failure. Although J&J isn’t currently pursuing a dedicated heart failure indication for Invokana, to our knowledge, both Lilly/BI and AZ are – for Jardiance (empagliflozin) and Farxiga (dapagliflozin), respectively.

• Dr. Figtree emphasized that the heart failure cohort was fairly small, comprising only 14.4% of the sample (n=1,461). At baseline and compared to the cohort without a history of heart failure, the heart failure cohort had significantly more participants who were female, a shorter mean duration of diabetes, a higher mean A1c, and more hypertension, CV disease, atrial fibrillation, and atherosclerotic CV disease, suggesting an overall sicker subgroup of patients (p<0.001 for all comparisons). Despite this, there were no significant differences on safety endpoints. The amputation signal was maintained across the split cohorts (HR=2.32, 95% CI: 1.10-5.51 for those with prior heart failure, and HR=1.90, 95% CI: 1.33-2.78 for those without prior heart failure).

• These analyses affirmed that canagliflozin reduced risk for CV death and heart failure hospitalization across most baseline characteristics, which is an important finding in terms of expanding the spectrum of type 2 patients who take the drug. Dr. Figtree reviewed various subgroups to show that no significant interactions existed for region, CANVAS vs. CANVAS-R, age (split at 65 years, though p=0.09 suggesting borderline significance), blood pressure (split at 140/90 mmHg), diabetes duration (split at 10 years), baseline eGFR, insulin use, atrial fibrillation, DPP-4 inhibitor use, TZD use, or history of CV disease. There were borderline significant interactions with BMI (p=0.03) when split at 30 kg/m², baseline A1c (p=0.04) when split at 8%, and baseline metformin use (p=0.03); these indicate that those with higher BMI, higher baseline A1c, and who aren’t using metformin may benefit more from canagliflozin in terms of CV death and heart failure hospitalization – we’d be curious for additional investigations into these effects, and we’re intrigued by the possibility that SGLT-2s could be effective first-line agents for people with higher starting A1c and/or BMI.

• Neither CANVAS nor EMPA-REG OUTCOME (the CVOT for Lilly/BI’s Jardiance) measured baseline heart failure status with a cardiac echo or another formal diagnostic assessment; rather, this was an investigator-reported comorbidity. As such, it’s possible that far more patients had heart failure at baseline (i.e. the 14.4% of CANVAS participants is an underestimate). Indeed, Dr. David Aguilar asserted that heart failure is comorbid with diabetes in 25%-40% of patients, depending on the population studied, so the 14% prevalence in CANVAS does seem low. Dr. Figtree shared that, for the hospitalizations that occurred in CANVAS, only ~60% have echo data; she pointed to EMPEROR HF as an important prospective trial program for clearing things up. Also notable, DECLARE (CVOT for AZ’s Farxiga) is recording baseline cardiac echo or other formal assessments of heart failure if they were conducted, and the trial features a co-primary endpoint that includes heart failure. As such, DECLARE seems poised to give us the best evidence yet on SGLT-2 inhibitors and heart failure in diabetes, and we look forward to this readout anticipated in 2H18.

Joint American College of Cardiology/New England Journal of Medicine Late-Breaking Clinical Trials

Anti-Inflammatory Therapy with Canakinumab and Incident Type 2 Diabetes: A Pre-Specified Key Secondary Endpoint
of the CANTOS Trial

Brendan Everett, MD (Brigham and Women's Hospital, Boston, MA)

While baseline biomarkers of inflammation can predict the progression from prediabetes to diabetes, anti-inflammatory drug canakinumab (an interleukin-1ß inhibitor) does not prevent that progression in patients with prior MI and high baseline inflammation – these findings from a pre-specified analysis of the CANTOS trial (n=10,061) were presented by Dr. Brendan Everett. He showed that baseline hsCRP and IL-6 were both significantly associated with incident type 2 diabetes (p=0.0002 and p<0.001) at five years, even after multivariable adjustment. However, only the IL-6 association remained significant after adjusting for baseline A1c. Treatment with canakinumab did significantly reduce both hsCRP (at 50 mg, 150 mg, and 300 mg doses) and IL-6 (at 150 mg and 300 mg doses), but this drop in inflammatory biomarkers did not translate to a drop in incident diabetes at five years. Adjudicated diabetes incidence per 100 person-years across all three canakinumab arms of the trial was 4.23 vs. 4.2 in the placebo arm (p=0.86). Incidence was 4.24/100 person-years in the 50 mg canakinumab group, 4.35 in the 150 mg canakinumab group, and 4.12 in the 300 mg canakinumab group (p=0.70-0.80 vs. placebo). That said, canakinumab did reduce risk for three-point and four-point MACE in this population of high-risk patients with pre-existing atherosclerotic CV disease, and this benefit was seen regardless of baseline glycemic status (p=0.86 for heterogeneity). In a somewhat surprising effect, canakinumab provided a significant (p<0.0001) but fleeting A1c reduction in people with prediabetes, in dose-dependent fashion. This benefit rapidly attenuated over ~9-12 months, and was not clinically-meaningful in our opinion. Overall, it seems that canakinumab is likely effective in reducing CV but not diabetes risk, despite an effect on inflammation. In CANTOS (n=10,061), 4,057 participants had baseline diabetes, 4,960 had baseline prediabetes, and 1,044 had normal glucose.

• Dr. Laurence Sperling highlighted this CANTOS analysis in a talk on the top three presentations in diabetes most likely to impact clinical or research practice. Despite seemingly negative results overall, Dr. Sperling was very positive about the fact that glycemia didn’t interfere with canakinumab’s MACE benefit in CANTOS. He also emphasized that this trial contributes to our ability to predict diabetes. Indeed, we think this analysis supports the importance of inflammation as a risk factor for diabetes, even if modification of that risk factor with canakinumab didn’t reduce diabetes incidence per se. To our understanding, the inflammation predicting development of diabetes is possibly more of a symptom than a cause of the pathophysiology underlying diabetes. More work certainly need to be and is being done to puzzle apart the role of inflammation in cardiometabolic disorders – it’s been implicated in insulin resistance and impaired insulin production, according to Dr. Everett – and Dr. Sperling urged the audience not to abandon the idea of anti-inflammatory therapy.

Ensuring a Bright Future for Patients with T2DM: Comprehensive Risk Management

Is T2DM Really a CHD Risk Equivalent? The Case for Formal Risk Assessment

Peter Wilson, MD (Emory University, Atlanta, GA)

In a very well-attended session on diabetes risk management, Emory’s Dr. Peter Wilson argued against the commonly-used heuristic that diabetes confers equivalent CV risk as a prior MI in someone without diabetes; instead, he illustrated the importance of considering both individual risk and atherosclerotic CV disease heterogeneity. Dr. Wilson traced the commonly-cited assertion back to a 1998 Finnish population-based study, published in NEJM, which found equivalent MI risk among 1,373 people without diabetes and 1,059 people with diabetes. While he acknowledged that people with diabetes are often affected by CV disease (~two-thirds of people with diabetes will experience comorbid CV disease), he described the heuristic as too simplistic overall, since relative risk for CV events is also impacted by cholesterol, tobacco use, renal function, age, and blood pressure in addition to diabetes status. Moreover, it’s not just atherosclerotic events that HCPs must consider in treating diabetes: Intermittent claudication (pain in the legs caused by obstruction of arteries) and heart failure are at least as important. Dr. Wilson reviewed data showing a general doubling or tripling of CV risk with type 2 diabetes, but he ultimately emphasized that data on the magnitude of risk are mixed. In fact, in a 2009 meta-analysis, the 1998 Finnish study was the only one to support the conception of diabetes as a risk equivalent to prior MI. This meta-analysis (n=45,108) found that patients with diabetes and no prior MI actually have a 43% lower risk of overall coronary heart disease-related events vs. people with no diabetes and a previous MI. Ultimately, there’s no arguing that people with diabetes experience devastatingly higher rates of MI, stroke, and renal disease than the general population, but Dr. Wilson cautioned against blanket assumptions; he emphasized the importance of a range of factors when it comes to an individual’s CV risk and how that should be treated. Although he didn’t explicitly mention precision medicine, we felt this talk was a compelling argument for why we need better tools to personalize treatment plans that reduce cardiometabolic risk. As one example, BI and Geisinger have teamed up to develop a risk-prediction model for CV death, kidney failure, and heart failure in type 2, which could eventually become a decision-support tool for providers. Indeed, when session chair Dr. Mikhail Kosiborod (who very much agreed with Dr. Wilson’s argument) asked what providers should be using right now to identify people with diabetes who do have equivalent CV risk, Dr. Wilson acknowledged the lack of tools, data, and guidelines.

Moving Beyond A1C Control: Selection of Anti-Diabetic Agents for CVD Risk Reduction

Lawrence Leiter, MD (St. Michael's Hospital, Toronto, Canada)

Toronto’s Dr. Lawrence Leiter shared his thoughts on diabetes CVOT design, commenting that current trials are limited in their short duration and high-risk populations. As he put it, these “crash tests” for safety may not be the best way to demonstrate CV efficacy. Dr. Leiter noted that when early CVOTs (ELIXA for Sanofi’s GLP-1 lixisenatide and the DPP-4 studies like SAVOR-TIMI, EXAMINE, and TECOS) all showed neutrality, experts thought the populations being investigated were too late in the course of disease; thought leaders suggested that trials instead enroll lower-risk patients for a longer study duration. As we understand it, the main incentive in selecting an enriched, higher-risk population is that “sicker” participants have a higher CV event rate, allowing for shorter trials (and these outcomes studies are already a major investment of time and resources). Indeed, >80% of participants in LEADER and SUSTAIN 6 had baseline CV disease, and those trials demonstrated CV superiority of Novo Nordisk’s GLP-1 agonists liraglutide (Victoza) and semaglutide (Ozempic). In contrast, only 73% of patients in EXSCEL for AZ’s GLP-1 Bydureon (exenatide once-weekly) had baseline CV disease, and that study just missed the threshold for CV superiority. Moreover, only 31% of people in REWIND for Lilly’s GLP-1 Trulicity (dulaglutide) have baseline CV disease, which is one reason that Novo Nordisk CSO Dr. Mads Thomsen doesn’t believe this CVOT will show significant CV superiority (REWIND is expected to complete in July 2018). For SGLT-2 inhibitors, all of EMPA-REG OUTCOME and 66% of CANVAS participants had baseline CV disease, while DECLARE (the CVOT for AZ’s Farxiga, likely reading out in 2H18) has enrolled only ~41% of participants with baseline CV disease. Could a much longer study show a greater magnitude of benefit if the intervention is started in the early stages of disease? REWIND and DECLARE should help shed some light on this issue, with their larger primary prevention cohorts, but these trials are a fairly mild change compared to their predecessors. We often think about the potential for SGLT-2s and GLP-1s in diabetes prevention, and we’d be over the moon if a long outcomes trial investigated CV/renal effects with treatment beginning in prediabetes. We imagine SGLT-2s and GLP-1s may offer different advantages/disadvantages when started earlier in the course of hyperglycemia development, based on their distinct mechanisms of action. There is growing consensus that GLP-1 agonists exert their CV benefit via atherosclerotic effects, in which case a patient would need to have pre-existing plaques to experience appreciable CV risk reduction. Of course, there’s also the chance that a treatment prevents plaques from forming in the first place. Meanwhile, a post-hoc analysis of CANVAS presented at AHA 2017 found significant heart failure benefit with canagliflozin (J&J’s SGLT-2 inhibitor Invokana) in both the primary and secondary prevention cohorts, so perhaps SGLT-2s reduce risk for heart failure independent of CV or diabetes status – we’ll have a much more definitive answer on this in a couple years, now that AZ and Lilly/BI are investigating their SGLT-2 inhibitors in people with chronic heart failure, with or without diabetes (and we anticipate a subgroup of these studies will have prediabetes, allowing for interesting subgroup analyses).

• When asked if metformin should remain first-line therapy in type 2 diabetes, Dr. Leiter again characterized modern CVOT design as restrictive, an obstacle to change. He pointed out that most guidelines recommend adding drugs to background metformin because most clinical trial participants were on metformin at baseline. For guidelines to change, an advanced therapy would have to go head-to-head against metformin. Dr. Leiter suggested that there is significant interest for these trials within the medical community, but he also underscored that they will have to be done much less expensively than a standard CVOT – for example, by utilizing data sources like EMRs. If anyone has ideas on how to get pharma to focus on low-risk populations, , or on how to perform less expensive RCTs utilizing EMRs that are still compelling and convincing to all parties (including FDA and guideline-writers), we’d love to hear them.

• Dr. Leiter also commented on SGLT-2 inhibitor CVOTs and mechanisms of cardioprotection. He cited a recent post-hoc analysis of EMPA-REG OUTCOME (presented by Dr. David Fitchett at AHA 2017) demonstrating that hematocrit and hemoglobin were each associated with >50% mediation of heart failure benefit, while A1c and fasting plasma glucose showed 20%-30% mediating effect. According to Dr. Leiter, glucose-lowering takes a long time to exert an impact on CV risk, and the early separation of the Kaplan-Meier curves in this trial (and in CANVAS) are more likely to indicate a volume effect – but there is still significant discussion happening on this front. Combined with the fact that these trials were designed for glycemic equipoise, we can be almost certain that CV health isn’t being improved primarily through glycemic benefit. We’ve noticed that cardiologists are especially enthused by SGLT-2 inhibitors, which was reflected in an audience polling question: 63% of audience members increased their use of SGLT-2 inhibitors in clinical practice following CVOT readouts, compared to only 42% for GLP-1 agonists.

  • As for the amputation signal seen in CANVAS, Dr. Leiter explained that he would be cautious about prescribing canagliflozin to patients with peripheral arterial disease (PAD). He reviewed the overall two-fold increase in amputation risk observed in CANVAS, but also a 7-8-fold increase in those with PAD and a 20-fold increase in those with prior amputation. Pushing the question even further, Dr. Leiter asked whether you should ever use any SGLT-2 inhibitor in patients with PAD, noting EMA’s (conservative) class-wide warning for increased amputation risk (FDA has added a lower limb amputation warning only to the Invokana label, but not to other SGLT-2 products). While we’ve heard strong support in favor of a cardioprotective class effect for SGLT-2 inhibitors, the judge and jury are still out on class effects for amputation – another reason we’re looking forward to DECLARE results.

  • SGLT-2 inhibitors are particularly intriguing because of their demonstrated heart failure benefit in EMPA-REG and CANVAS alike (not to mention CVD-REAL). There have been suggestions that this class may exert a heart failure benefit regardless of patients’ baseline heart failure status (i.e. primary as well as secondary prevention), but Dr. Leiter pointed out that in both EMPA-REG and CANVAS, heart failure status was determined by the investigator simply checking a box, without a more formal assessment of cardiac function. After the impressive hospitalization for heart failure findings in EMPA-REG and CANVAS, AZ’s DECLARE trial for Farxiga (dapagliflozin) was modified to include collect assessments of cardiac function, e.g. a cardiac echo, if already preformed prior, so we’ll have a more precise picture of baseline cardiac function.

There's No Sugar Coating the New Reality: Changing the Paradigm for CVD Risk Reduction in T2DM

Setting the Stage: The Coming Paradigm Shift from Glycemic Control to CVD Risk Reduction

Mikhail Kosiborod, MD (Saint Luke's Hospital, Kansas City, MO)

“Does a cardiologist need to be a diabetologist, or does a diabetologist need to be a cardiologist?” Session chair Dr. Robert Eckel posed this question, and speaker Dr. Mikhail Kosiborod outlined a compelling case for both, advocating for greater collaboration in general between diabetes care specialists and CV medicine specialists. Dr. Kosiborod spoke to the paradigm shift taking place in diabetes, from a hyper-focus on glycemic control to a central focus on CV risk reduction: “No one wants people walking around with an A1c of 10% or 11%, but it’s more about how you do it [lower glucose] than just doing it.” At recent meetings, we’ve heard Dr. Kosiborod argue with increasing zeal that the ultimate goal of treatment in type 2 diabetes is to let patients live longer and feel better doing it, not to make lab values like A1c look better; in order to do this, treatment needs to target CV outcomes first and foremost, rather than glycemia. For us, this argument evokes the outcomes beyond A1c movement, highlighting the importance of hard outcomes that impact patient wellbeing, longevity, and quality of life – this includes hypoglycemia as well as CV events, and indeed, the two are likely related. Comparing people with and without diabetes, Dr. Kosiborod explained that both populations have experienced improvements in CV outcomes over the last 20 years, and yet the size of the gap between the two groups isn’t closing. Ischemic heart disease remains the #1 killer of people with diabetes by far. While glucose control unequivocally improves microvascular outcomes, Dr. Kosiborod noted that the same notion hasn’t panned out for macrovascular, CV outcomes: Aggressive, rapid lowering of A1c without paying attention to how you accomplish this has not and will not impact CV events. Dr. Kosiborod named CV death and heart failure the two most important CV events for diabetes patients, and he asserted that no evidence to-date has shown a correlation between better glycemic control and better prognosis for these outcomes. That’s why positive diabetes CVOTs represent such a major victory, because cardioprotection is inherent to molecules like liraglutide, empagliflozin, and canagliflozin, unrelated to A1c reduction per se. In Dr. Kosiborod’s words, EMPA-REG OUTCOME “changed the world” as the first positive read out, followed by IRIS, LEADER, SUSTAIN 6, CANVAS, and CVD-REAL. Suddenly, there was a new classification of glucose-lowering compounds – those that reduce CV risk, and even CV death – and we wholeheartedly share Dr. Kosiborod’s excitement about SGLT-2 inhibitors and GLP-1 agonists for their CV (and renal) benefits.

Recent Breakthrough Cardiovascular Outcomes Trials in T2DM

Benjamin Scirica, MD (Brigham and Women’s Hospital, Boston, MA)

Dr. Benjamin Scirica argued first that cardiovascular outcomes studies are absolutely necessary (offering insight above and beyond phase 2b/3 trials), and second that positive diabetes CVOTs are convincing evidence of the CV benefit associated with SGLT-2 inhibitors and GLP-1 agonist and should impact clinical practice. To the first point, Dr. Scirica presented pooled phase 2b/3 data for saxagliptin (AZ’s DPP-4 inhibitor Onglyza): 41 total MACE events translated into a 56% risk reduction vs. placebo at 76 weeks (HR=0.44, 95% CI: 0.24-0.82). On the other hand, the SAVOR-TIMI CVOT gave a resoundingly neutral hazard ratio of 1.00 for three-point MACE, out to 24 months. Without this RCT, our conception of the CV profile of saxagliptin would have been incredibly different. This evokes the point with which Dr. Scirica opened both of his talks: Before five years ago, the best randomized data on CV risk prevention in diabetes came from UKPDS, which is now 20 years-old and suggests a CV benefit for metformin with only 342 patients (a sample size that would never fly these days). He explained that the field needs to keep conducting large and lengthy CV outcomes trials in diabetes because there’s simply no other way to gain a more accurate or precise understanding of drug benefit. This echoed commentary from Dr. Darren McGuire at ESC 2017. Dr. McGuire described how prior to FDA’s 2008 CVOT guidance, diabetes therapies were approved for the market with as few as 250 patient-years of drug exposure, compared to the ~15,000 patient-years of drug exposure afforded through a CVOT. This is clearly a substantial investment for diabetes manufacturers, but we ultimately agree with Drs. Scirica and McGuire in their point that these outcomes studies paint a comprehensive picture on safety while also uncovering “surprise” CV benefits.

• Dr. Scirica divided diabetes CVOTs into two waves, the first including all three DPP-4 inhibitor trials (TECOS for sitagliptin, SAVOR-TIMI for saxagliptin, and EXAMINE for alogliptin) plus ELIXA (for GLP-1 lixisenatide). For the most part, these were reassuring safety trials, although Dr. Scirica did highlight the unexpected finding of a 27% increased risk for heart failure hospitalization with saxagliptin vs. placebo. Now, these studies are “quaint,” in his view. His second wave of diabetes CVOTs includes EMPA-REG (for SGLT-2 empagliflozin), LEADER (for GLP-1 liraglutide), SUSTAIN 6 (for GLP-1 semaglutide), CANVAS (for SGLT-2 canagliflozin), and also IRIS for TZD pioglitazone in patients with recent stroke. What these trials have in common is positive results, indicating cardioprotection. Dr. Scirica asserted that thought leaders would not have bet on CV superiority if someone asked them to put money down, meaning these were “surprise” findings for the field at first. Both empagliflozin and canagliflozin showed 14% risk reduction vs. placebo for three-point MACE (non-fatal MI, non-fatal stroke, CV death), while liraglutide gave a 13% relative risk reduction on the same endpoint and semaglutide gave a 26% relative risk reduction (with a much wider confidence interval). These results are “remarkable in a time we’re finding it harder and harder to reduce CV death,” Dr. Scirica said. For people who think the results are too good to be true, he pointed to the consistent signals across multiple trials and the fact that A1c reduction in CVOTs is much lower than in trials of glycemic efficacy, indicating glucose-independent mechanisms.

• To conclude, Dr. Scirica laid out a philosophy for cardiologists in prescribing diabetes therapies. With the explosion of new drugs classes in recent decades, how do cardiologists and other practitioners integrate them into clinical practice? As Dr. Scirica put it, providers need to fundamentally shift away from thinking that the goal of therapy is to lower A1c. HCPs should focus more on how a therapy lowers glucose rather than by how much, since SGLT-2s and GLP-1s have demonstrated CV benefit independent of A1c improvements.

• Dr. Scirica also listed some outstanding questions that he hopes will be addressed in the next, third wave of diabetes CVOTs. Should metformin still be first-line therapy? Dr. Scirica suggested a “modern UKPDS” that investigates A1c normalization in newly-diagnosed patients. He posited another question recently addressed in a JCI viewpoint: Can we still do placebo-controlled trials when we know SGLT-2 inhibitors and GLP-1 agonists are cardioprotective, helping patients avoid heart attacks, strokes, and CV death? See our take here. Lastly, Dr. Scirica considered class effects and their potential implications: At what point (if ever) can we say something is a class effect and that new class members don’t need full CVOTs? Separately, Dr. Mikhail Kosiborod explained that there are plenty of examples in medicine where members of the same class have very different safety or efficacy signals. Additionally, Dr. Scirica emphasized the need for more studies in people with obesity, CKD, and prediabetes, as well as more endpoints on eGFR and hospitalization for heart failure.

The New Reality is Here: The Why, Who, and How of Comprehensive Diabetes Care

Mikhail Kosiborod, MD (Saint Luke's Hospital, Kansas City, MO), Robert Eckel, MD (University of Colorado, Aurora, CO), Laurence Sperling, MD (Emory University, Atlanta, GA), Melissa Magwire (Shawnee Mission Health, Merriam, KS)

Drs. Robert Eckel and Mikhail Kosiborod gave a master class in diabetes management for the cardiologist during a packed dinner symposium. Dr. Eckel clarified that while glycemia is absolutely relevant in treating people with diabetes, HCPs can now lower glucose while simultaneously mitigating CV risk, with cardioprotective diabetes drugs like Novo Nordisk’s GLP-1 Victoza (liraglutide) and Lilly/BI’s SGLT-2 Jardiance (empagliflozin). This hit on an all-important conference theme – that it’s a tremendous victory to have multiple diabetes drugs with CV benefit, to have a choice (!) among glucose-lowering, weight loss-promoting, cardioprotective medicines, and that people with diabetes and comorbid CV disease need multidisciplinary care, coordinated among specialists. Drs. Eckel and Kosiborod addressed two safety signals, for amputations in CANVAS and retinopathy in SUSTAIN 6. Dr. Kosiborod argued that the amputation signal associated with canagliflozin (J&J’s SGLT-2 inhibitor Invokana) is very likely real: A near-doubling of risk is a big difference, even from a low base, and the hazard ratio was consistent between CANVAS and CANVAS-R, making it unlikely to be a statistical fluke. A post-hoc safety analysis of CANVAS was presented at EASD 2017, and even after controlling for other known risk factors (peripheral vascular disease, neuropathy, etc.), canagliflozin treatment increased the likelihood of a lower limb amputation nearly two-fold. Dr. Kosiborod talked about adjudication and collection of amputation events, explaining that while people “mistrust” EMPA-REG OUTCOME’s amputation data because it didn’t label amputations an event of special interest, amputations only became of special interest in CANVAS after an early signal and action by the independent data monitoring committee. The main question now, in Dr. Kosiborod’s view, is whether amputations are an SGLT-2 inhibitor class effect – there’s no compelling evidence that supports this, but other trials are now collecting that data more thoroughly. Turning to SUSTAIN 6 and the retinopathy signal seen with semaglutide (Novo Nordisk’s GLP-1 Ozempic), Dr. Eckel described early worsening phenomenon, in which rapid drops in A1c are associated with a risk of progressive retinopathy in those with background retinopathy – and background retinopathy is often missed. He highlighted that early worsening  is only a theoretical explanation right now, though our sense is that most thought leaders seem to agree with it. Dr. Kosiborod referenced additional analyses of SUSTAIN 6, showing that people with more dramatic A1c-lowering were at higher risk for retinopathy regardless of treatment arm. Further, the absolute increase in retinopathy risk was much greater for those who already had retinopathy at baseline, even though semaglutide appeared to increase risk relative to placebo regardless of baseline retinopathy status. We don’t anticipate that retinopathy-related concerns will impede Ozempic uptake in the real world, given how powerful and potent this drug seems to be on A1c and weight loss, but this will unfold in the coming months as the product was just launched in the US in February. Meanwhile, amputation-related concerns have been a significant headwind for the Invokana business, ever since FDA issued a black box warning in May 2017. For our part, we believe amputation risk should be manageable in the real world, but Invokana sales have trended sharply down in recent quarters.

• We were happy that Dr. Eckel immediately criticized the new ACP guidelines recommending an A1c target of 7%-8% for the majority of type 2s, which is not consistent with any endocrinology organization’s guidelines. We heard some hallway chatter about these guidelines, and attendees did seem a bit wary of the recommended A1c goals, so we’re glad Dr. Eckel discussed the potential negative impact that these ACP guidance statements could have on diabetes care. We think it’s so important to spread the message that these are not widely-applicable guidelines for type 2 diabetes.

• The dinner conversation spanned a range of other topics as well. Dr. Eckel shared that he’s not a fan of SUs, as the insulin secretion they promote likely causes earlier beta cell burnout, and the lack of glucose-dependent function can cause hypoglycemia. We agree wholeheartedly with this view, and we wish the majority of second-line diabetes prescriptions in the US weren’t still going to a sulfonylurea. Dr. Kosiborod noted that eGFR restrictions for SGLT-2 inhibitors (contraindicated below 30 mL/min/1.73 m²) don’t arise from a safety concern as much as a worry that glycemic efficacy will be lost at very low renal function. That said, there’s no evidence to-date that the blood pressure, weight loss, or even CV benefits are attenuated for people with lower eGFR. Dr. Eckel added that the story is starting to change, with SGLT-2 manufacturers beginning to show that patients with eGFR <40 mL/min/1.73 m² still benefit. Indeed, we think upcoming trials of SGLT-2s in kidney disease will elucidate the full range of benefits for this advanced, oral therapy class: This includes CREDENCE for Invokana, Dapa-CKD for Farxiga, and a planned study for Jardiance in CKD. We also heard distinctly positive commentary on fixed-ratio basal/GLP-1 combinations. Dr. Eckel cited Dr. John Buse in talking about these combo drugs replacing basal insulin and even basal-bolus – we’re thrilled by this concept.

• Shifting perspective, Dr. Kosiborod offered practical advice on how to prescribe diabetes medications as a cardiologist without stepping on a PCP’s or endocrinologist’s toes. He shared a story in which a cardiology colleague prescribed an SGLT-2 inhibitor and shortly after received an angry call from the patient’s PCP for trying to manage diabetes. To avoid such scenarios, Dr. Kosiborod recommended explicitly putting in the patient note that the prescription is not for the management of blood glucose, but for minimizing CV risk: “I put nothing about A1c and glucose control in the note because that’s not why I’m doing it.” In his view (and ours), it’s absolutely under the purview of cardiologists to prescribe medications to lower CV risk even if they were originally diabetes medications. He urged cardiologists in the room to familiarize themselves with SGLT-2 inhibitors and GLP-1 agonists: “If you think you’re using these drugs to control blood glucose, it’s going to be a no-go from most of us. If you’re using it to control CV risk, we know what to talk about with the patient.” Georgetown’s Ms. Melissa Magwire shared her own experience in which a PCP asked a cardiologist to assess whether a patient would be a good candidate for an SGLT-2 inhibitor or GLP-1 agonist so that the PCP could start them on it. While this setup certainly requires more organization and structure, Ms. Magwire emphasized that communication is paramount in provider-provider relationships, not just patient-provider relationships. Finally, Dr. Kosiborod underscored the HCP’s responsibility to explain to patients what they should expect on these medications: For SGLT-2 inhibitors, this is one extra bladder void per day and a slight chance of yeast infections, and men should be asked if they’re circumcised (with an explanation why); for GLP-1 agonists, patients should be warned about nausea and other potential GI side-effects.

• Ms. Magwire advocated for coordinated, multidisciplinary teams to improve chronic disease management and patient outcomes. The hospital where she practices, Shawnee Mission Health in DC, established a glycemic steering committee in 2008, later adding half a dozen other interdisciplinary teams and specialist positions aimed at coordinating care. To underscore the need for improved communication, Ms. Magwire introduced a single patient who was receiving care from more than a dozen providers, including an endocrinologist, RN, PA, CDE, cardiologist, PCP, the PCP nurse practitioner, a vascular surgeon, a nephrologist, a pharmacist, social services, cardiac rehabilitation, a retinal specialist, a podiatrist, and a dietician. In situations with even far fewer people in the mix, communication is often sub-optimal, and that missing link negatively impacts outcomes, fragments care, and leads to questions and uncertainty surrounding treatment plans (for both provider and patient). Improved multidisciplinary communication should involve (i) organization of care, (ii) medication management, and (iii) promotion of patient-centered care and shared decision-making, which in and of itself improves satisfaction and outcomes. Above all else, however, Ms. Magwire emphasized the need to keep patient expectations in mind: Patients are content to receive care from multiple team members as long as there’s a perception of coordination, which is good news because no one person can offer best practice care for all aspects of diabetes.

• In the same session, Dr. Laurence Sperling noted that there are ~7,000 endocrinologists in the US (and only ~half really involved in diabetes care) compared to ~30,000 cardiologists. Thus, it makes basic sense for cardiologists to help minimize the CV risk conferred by diabetes. How can all HCPs help affect change in the diabetes arena? Dr. Sperling advocated for frequent examination of diabetes patients for nephropathy, smoking history, retinopathy, and foot complications. He also placed major emphasis on shared decision-making between the patient/provider. Dr. Sperling envisions “an extension of the existing healthcare paradigm,” and he put forth the idea of a Collaborative Care Command Center to encourage home and remote care provisions, enabled by the growing reach of different facets of digital health. While intensive lifestyle intervention seems to be the most effective way to reduce CV risk, Dr. Sperling underscored that we also have at least 12 medication classes proven to treat diabetes – and two that we know can reduce CV risk – so we need to work to maximize the benefit of all the treatment tools (pharmacological, technological, and personal).

Criticality of LDL-C Lowering (Sponsored by Amgen)

Criticality of LDL-C Lowering

James Underberg, MD (New York University, New York, NY)

In an Amgen-sponsored session, NYU’s Dr. James Underberg made a strong case for the urgency of LDL-lowering in patients with diabetes, while also highlighting the factors that make meeting LDL targets difficult. He presented lipids as one of the more modifiable CV risk factors, compared to type 2 diabetes (hyperglycemia), smoking, obesity, and blood pressure. In characterizing groups with the highest CV risk, Dr. Underberg placed specific emphasis on people with atherosclerotic CV disease and comorbid diabetes, naming a target LDL level of <70 mg/dl for these patients in line with the 2017 National Lipid Association Recommendations for Patient-Centered Management of Dyslipidemia. This echoes commentary we’ve heard from other thought leaders as well – diabetes and dyslipidemia are both independent risk factors for CV morbidity/mortality, and when the two exist together, CV risk reduction strategies become especially urgent. Unfortunately, Dr. Underberg showed how hyperlipidemia is undertreated worldwide: Although statins are relatively inexpensive (~$10/month in the US), ~70% of people globally are not reaching LDL goal <70 mg/dl. Dr. Underberg cited limited exposure to statins in some populations and difficulty maintaining adherence, but also genetic dispositions (including but not limited to familial hypercholesterolemia). He also mentioned lack of responsiveness to statins that makes lipid control difficult even on high-dose statins – this is where PCSK9 inhibitors come in. We appreciated Dr. Underberg’s emphasis on patients with diabetes as a particularly high-risk group. We imagine that more can be done to mitigate CV risk in patients with diabetes – and that some risk factors, like hyperlipidemia, might even sit in the shadow of A1c. Dr. Underberg spoke to a major theme of ACC: To move the needle on CV morbidity/mortality, patients with cardiometabolic dysfunction need more multidisciplinary medical care from coordinated teams of specialists.

Posters

Design and Baseline Characteristics of the eValuation of ERTuglifozin effIcacy and Safety Cardiovascular
Outcomes Trial (VERTIS-CV)

C Cannon, D McGuire, R Pratley, S Dagogo-Jack, J Mancuso, S Huych, B Charbonnel, W Shih, S Gallo, U Masiukiewicz, G Golm, C Francesco, B Lauring, S Terra

Dr. Christopher Cannon presented a poster on the design and baseline characteristics of VERTIS-CV, the CVOT for ertugliflozin (Merck/Pfizer’s SGLT-2 inhibitor Steglatro). A paper detailing this information was also just published in JACC. VERTIS-CV (n=8,327) isn’t expected to complete until October 2019, but this will be a very important read out for Merck, Pfizer, and the long-term success of Steglatro on the market (considering that two SGLT-2 inhibitors have already demonstrated CV benefit – Lilly/BI’s Jardiance has this benefit reflected on its label, while J&J’s Invokana may also receive a CV indication by year-end). Dr. Cannon showed that a particularly high proportion of patients with baseline heart failure (22%) were enrolled in VERTIS-CV, compared to DECLARE (10%), CANVAS (14%), or EMPA-REG (10%). In line with inclusion criteria, 99% of patients had baseline CV disease, similar to EMPA-REG and in contrast to 41% in DECLARE and 66% in CANVAS. Coronary artery disease was known in 76% of VERTIS-CV participants at baseline, while cerebrovascular disease was known in 23% and peripheral arterial disease (PAD) in 19%. Stage 3 CKD was present in 22% of VERTIS-CV patients at study start, while micro- and macroalbuminuria were present in 30% and 9% of participants, respectively. Overall, we agree with Dr. Cannon’s characterization of this study population as representing high baseline CV risk. Participants were randomized 1:1:1 to placebo, 5 mg ertugliflozin, or 15 mg ertugliflozin once-daily. Like the other SGLT-2 inhibitor CVOTs, the primary endpoint is three-point MACE comprised of CV death, non-fatal MI, and non-fatal stroke. Key secondary endpoints include CV death or hospitalization for heart failure, CV death alone, and a renal composite of renal death, dialysis/transplant, or doubling of serum creatinine from baseline (to be sure, the renal benefits of SGLT-2s have been a wonderful surprise finding from CVOTs completed to-date, and we’ll be looking for this from ertugliflozin in VERTIS-CV as well). While most CVOTs over-represent white men, VERTIS-CV has the highest enrollment of white patients by eight percentage points (at 88%), and we expect criticism from some on this point.

Eligibility Varies Across the 3 Sodium-Glucose Cotransporter-2 Inhibitor Cardiovascular Outcomes Trials
Among Adults With Type 2 Diabetes: Implications From Analysis of the Diabetes Collaborative Registry

E Wittbrodt, D Chamberlain, S Arnold, F Tang, M Kosiborod

In a poster presentation, Dr. Eric Wittbrodt highlighted that ~half of patients with type 2 diabetes do not meet the eligibility criteria for any SGLT-2 inhibitor CVOT. Based on exclusion criteria related to A1c, chronic kidney disease (CKD) status, and CV history, only 21% of type 2s in the Diabetes Collaborate Registry (DCR) would be eligible for EMPA-REG OUTCOME, 30% for CANVAS, and 46% for DECLARE. Only 16% of patients were eligible for all three trials, Dr. Wittbrodt explained, while 46% of patients were eligible for none. Most notable in our opinion is that nearly half of the real-world population with type 2 diabetes is not eligible for any of these CVOTs, suggesting that results from these RCTs may not generalize to many patients seen in usual clinical practice. Moreover, there are typically eligibility criteria above and beyond these three categories of A1c, CKD, and CV history, so even fewer patients would likely qualify for these CVOTs in reality. While we don’t think this fact should diminish excitement about the CV efficacy of SGLT-2 inhibitors in any way (after all, EMPA-REG and CANVAS showed compelling CV safety, and there’s a chance for any patient to reap the CV benefits), it does show the unmet need for outcomes-based research in a very “real-world” patient population. According to Dr. Wittbrodt, the DCR is the first interdisciplinary diabetes database to monitor and boost care for diabetes and metabolic disease patients, covering ~two million patients and ~10,000 providers total; we’re intrigued to see what else might come out of this dataset. Patients in this study (n=407,410) were ³18 years of age with type 2 diabetes diagnosed between January 1, 2012 and March 31, 2017. Type 1 patients, people with type 2 diabetes managed with diet alone (without pharmacotherapy), and patients with prediabetes were not included. Dr. Wittbrodt mentioned a few limitations, including a potential gap in accurate US population illustration due to the fact that 29% of the data provided was from HCPs. This sample included patients from 689 sites and 9,266 providers. At baseline, patients had a mean age of 66 years, 49% were female, mean A1c was 7.9%, 53% had baseline atherosclerotic CV disease, and the 10-year risk for atherosclerotic CV disease in those without it at baseline was 25%. Among type 2 patients, there were high rates of hypertension and dyslipidemia (>80%), though only ~8% of the people in this category actually took SGLT-2 inhibitors (which matches data from another Diabetes Care publication pointing to only ~7% of second-line diabetes prescriptions in the US going to an SGLT-2 product). We agree with researchers when they say that these findings offer a chance for renewed attention to provision of SGLT-2 inhibitors for more sustainable patient health outcomes in the future.

• We also think it’s important to note significant differences in study population across trials: A1c criteria ~halved the eligible population for CANVAS and EMPA-REG, but had a much less sizable impact on DECLARE, though DECLARE did have more restrictive CKD criteria. Overall, these findings suggest that DECLARE will be a more “real-world” CVOT, similar to AZ’s other diabetes CVOT, EXSCEL for GLP-1 agonist Bydureon, which has been called out for its “pragmatic design.”

Predicting Cardiovascular Risk Using Common Utilization Management Criteria for Proprotein Convertase Subtilisin/Kexin Type 9 Inhibitors in Commercially Insured Patients with Atherosclerotic Cardiovascular Disease

M Bonafede, S DiMario, J Patel, B Johnson, D Harrison

Payers create utilization management (UM) criteria to ensure that the patients most in-need have access to advanced (expensive) therapies. However, this analysis showed that current criteria for PCSK9 products are not actually associated with increased risk of CV events. Researchers used a drug formulary database to identify four common UM criteria for PCSK9 inhibitors: (i) statin use, (ii) high-intensity statins, (iii) duration of statin use, and (iv) duration of ezetimibe use. Patients were considered to be participants (n=5,276) if they had at least one LDL reading in the past year ≥70 mg/dl; all participants were commercially-insured and had atherosclerotic CV disease (51% female, mean age 56). A multivariate analysis evaluated whether UM criteria were associated with more or less risk of CV events in the year following an LDL reading of at least 70 mg/dl. After controlling for demographic characteristics and baseline comorbidities, no association with CV events was found for (i) current vs. prior statin use (95% CI: 0.77-1.48), (ii) number of high-intensity statins used, compared to one statin (95% CI: 0.63-1.10 for zero statins, 0.91-3.64 for two+ statins), (iii) duration of statin use, compared to <60 days (95% CI: 0.98-3.15 for 60-90 days, 0.61-1.72 for 90-180 days, and 0.37-1.01 for >180 days), or (iv) duration of ezetimibe use, compared to 1-180 days (95% CI: 0.29-1.01 for no ezetimibe, 0.47-2.22 for >180 days). Since all of these 95% confidence intervals cross unity, the implication is that none of these criteria were significantly associated with CV risk, which is arguably how payers should determine candidates first in line for PCSK9 inhibitor coverage. The poster noted that some selection bias may have influenced results for ezetimibe and multiple statin usage, as patients with high CV risk are more likely to meet those criteria. The authors emphasized that <half of prior authorizations for PCSK9 inhibitors are approved based on these criteria – our sense is that this number has improved as providers have learned to work the system (according to a 2016 survey, only 36% of PCSK9 prior authorizations were approved), but payers are nevertheless using these potentially irrelevant UM criteria to make coverage decisions. We certainly agree with the authors that ample data and provider expertise exist and should be used to create a more accurate and data-driven algorithm for assessing CV risk and determining who could benefit most from treatment with a PCSK9 inhibitor.

Cost Effectiveness of an Outcome Based Reimbursement Model of PCSK9 Inhibitors

O Oren, M Oren

A poster concluded that an outcomes-based reimbursement model for Amgen’s Repatha (evolocumab), in which the company would offer refunds to payers and/or patients who “failed” therapy, would not meaningfully change the cost-effectiveness of PCSK9 inhibitors. Researchers calculated four hypothetical incremental cost-effectiveness ratios (ICERs) using FOURIER data, assuming full reimbursement to payers/patients if evolocumab (i) did not reduce LDL cholesterol to one of three thresholds or (ii) in the case of a CV event. In this cost-utility analysis, the ICER was synonymous with the QALY, or cost per quality-adjusted life year. Naturally, the literature-derived ICER for evolocumab was much higher than that of atorvastatin, at $274,000/QALY vs. $20,300/QALY. In the case of compensation based on a CV event, the ICER for evolocumab fell to $246,000/QALY. With thresholds of LDL >70 mg/dl, >40 mg/dl, and >25 mg/dl evolocumab’s ICER value dropped to $239,000/QALY, $184,000/QALY, and $115,000/QALY, respectively. Thus, even with an outcomes-based contract of sorts, it’s difficult to make Repatha as appealing to payers as atorvastatin. Indeed, we’ve heard quite a few presenters remark that PCSK9 inhibitors are simply not an effective population-level solution for hyperlipidemia at their current price. We’re noticing more mentions of risk stratification as cardiologists try to determine which of their patients would benefit most from PCSK9 inhibitor therapy. How can risk stratification be most effectively incorporated into treatment algorithms? While we’re glad manufacturers are working on expanding access broadly from their end, we acknowledge that (at least for now) targeting of PCSK9s to the patients most in-need is probably the only way to move the needle on reimbursement. That being said, Sanofi/Regeneron have announced plans to lower net price for payers willing to lower access barriers, in line with a new cost-effectiveness analysis from the Institute for Clinical and Economic Review (also ICER) including ODYSSEY Outcomes results (above) – we’ll keep a close eye on how this impacts access in the coming months and years.

Greater Likelihood of Regression of Coronary Atherosclerosis With the PCSK9 Inhibitor, Evolocumab, in Patients With Higher Lp(a) Levels

D Scherer, R Puri, C Ballantyne, L Cho, D Brennan, H Kassahun, R Somaratne, S Wasserman, S Nissen, S Nicholls

Dr. Daniel Scherer (South Australian Health and Medical Research Institute, Adelaide, Australia) presented a poster showing that higher Lipoprotein(a), or Lp(a), levels could help identify patients likely to achieve a greater degree of atheroma volume regression with evolocumab (Amgen’s Repatha). Dr. Scherer’s analysis was a post-hoc of GLAGOV (n=970), an Amgen-sponsored study looking at change in percent and total atheroma volume with evolocumab vs. placebo as add-on to statin therapy. Dr. Scherer stratified patients according to baseline Lp(a), a cholesterol and fat transporter and risk factor for CV disease. His group found that 71% of those with above-median Lp(a) experienced percent atheroma volume (PAV) regression vs. only 59% of those below-median (p=0.01 for comparison). PAV reduction from baseline was 1.2% in the high Lp(a) group vs. 0.8% in the low Lp(a) group. Total atheroma volume (TAV) reduction was 7.7 mm³ vs. 5.3 mm³ in the above-median Lp(a) and below-median Lp(a) groups, respectively. Median Lp(a) was 11.8 mg/dl. No p-values were reported for a between-groups comparison of PAV or TAV regression. We were particularly compelled by Dr. Scherer’s analysis showing that baseline Lp(a) >11.8 mg/dl was associated with increased plaque regression with evolocumab (p=0.04 vs. placebo); however, this became marginally non-significant (p=0.09) when adjusting for baseline plaque burden. In GLAGOV participants with Lp(a) <11.8 mg/dl at baseline, a similar degree of regression was seen with evolocumab vs. placebo (p=0.35). We think these findings are most interesting in light of ODYSSEY Outcomes data showing that people with higher baseline LDL (≥100 mg/dl) experienced seemingly all of the CV benefit (see highlight no. 1 above). In GLAGOV, too, higher Lp(a) drove benefit on PAV. Certainly, more analyses of how baseline lipid status affects response to PCSK9 inhibitors are needed to help determine who the “best responders” might be. Dr. Scherer suggested that Lp(a) may indicate a more “modifiable” form of atherosclerosis, but it seems just as likely to us that patients with higher Lp(a) are high-risk overall, making the benefit on a CV biomarker more visible (this seems to be partially realized in the drop in statistical significance when adjusting for baseline plaque burden). Interestingly, patients with an above-median baseline Lp(a) level experienced lower risk for new-onset diabetes (16% vs. 26%, p=0.02) and hypertension (75% vs. 87%, p=0.001), and also had lower baseline CRP levels (1.3 vs. 1.8 mg/l, p=0.02) but slightly higher on-treatment LDL levels (33.9 vs. 32.6 mg/dl, p=0.02). Ultimately, we’re not sure that the marginal differences in some of these measurements are clinically-meaningful, but we’re looking forward to future analyses aimed at predicting PCSK9 inhibitor response.

Empagliflozin Induces a Myocardial Metabolic Shift from Glucose Consumption to Ketone Metabolism that Mitigates Adverse Cardiac Remodeling and Improves Myocardial Contractility

C Santos-Gellego, J Ibanez, R San Antonio, K Ishikawa, S Watanabe, M Botija, A Sanz Salvo, R Hajjar, V Fuster, J Badimon

A study from Mount Sinai used a pig model to show that empagliflozin shifts myocardial metabolism from glucose oxidation to ketone metabolism, improving myocardial energetics and lowering LV remodeling. After MI was induced via two-hour balloon occlusion of the proximal LAD, 16 pigs were randomized to either 10 mg empagliflozin daily or to placebo; they also received cardiac MRI and 3D-echo at one day and two months post-MI. Empagliflozin (Lilly/BI’s Jardiance) was associated with lower cardiac remodeling at two months, as well as reduced LV mass (p=0.02), volume (p<0.01), and sphericity (p=0.01). Empagliflozin also gave improved LV systolic function, as measured by LVEF and strain. Neurohormonal activation, as measured by BNP and catecholamine, was reduced in the empagliflozin group compared to the placebo group (p<0.01 and p=0.03, respectively). Moreover, sampling of myocardial metabolites from the LAD and coronary sinus was conducted, demonstrating that control pigs had higher myocardial glucose consumption (p=0.01), while empagliflozin raised plasma ketone levels (p<0.01) and myocardial ketone consumption (p<0.001), resulting in higher lactate levels (p<0.05). If that weren’t enough, empagliflozin treatment also improved ATP/ADP and NADH/NAD+ ratios in the myocardium (p<0.05 and p=0.02, respectively). We find the variety of measurements included in this study particularly compelling, and we think it’s quite likely that the ketone metabolism hypothesis plays some role in mediating the CV benefit associated with SGLT-2 inhibitors. That said, we also agree with Dr. Benjamin Scirica that SGLT-2 inhibitors (and GLP-1 agonists) are likely exerting their effects via multiple axes that compound into significant collective benefit.

Empagliflozin and Canagliflozin Attenuate Inflammatory Cytokines Interferon-γ, Tumor Necrosis Factor-α, Interleukin-6: Possible Mechanism of Decreasing Cardiovascular Risk in Diabetes Mellitus

S Tan, L Tan

An in-human study found that empagliflozin but not canagliflozin (J&J’s Invokana) significantly reduced inflammatory cytokines. Interestingly, canagliflozin was more effective in lowering A1c, but empagliflozin had the edge on cytokine reduction. Thirty-two men with type 2 diabetes were randomized 1:1 to empagliflozin 10 mg or canagliflozin 100 mg for six months, then switched to the other therapy for another six months. At the first six-month mark, canagliflozin was associated with a 0.9% drop in A1c, while empagliflozin decreased A1c by 0.7%. After another six months, the group that went from canagliflozin to empagliflozin experienced a 0.1% increase in A1c, while the empagliflozin to canagliflozin group experienced an additional 0.1% A1c decline. Turning to the inflammatory biomarker data: At six months, canagliflozin dropped IFN-γ (inflammatory cytokines interferon gamma) from 23 pg/ml to 18 pg/ml (p=0.38), while empagliflozin reduced IFY-γ to 10 mg/ml from the same baseline (p=0.001). The numerical difference here between canagliflozin’s and empagliflozin’s effect is more than marginal, and of course, only empagliflozin showed a statistically significant change. On TNF-α (tumor necrosis factor alpha), canagliflozin reduced levels from 40 pg/ml to 33 pg/ml (p=0.29), but empagliflozin dropped the same to 25 pg/ml (p=0.002), reaching statistical significance for this improvement in inflammation. Finally, for IL-6 (interleukin 6), canagliflozin gave a reduction from 20 pg/ml to 16 pg/ml (p=0.27), compared to 10 pg/ml with empagliflozin (p=0.022). Thus, treatment with empagliflozin moved the needle on three inflammatory biomarkers into “normal” ranges. Moreover, similar magnitudes of reduction were seen in the group that received canagliflozin first once they were switched to empagliflozin. In the group to switch from empagliflozin to canagliflozin, inflammatory biomarkers had climbed back up by one year (to just below the six-month canagliflozin data), indicating to us that inflammation is a very modifiable pathology. To be sure, this is a small trial and did not assess CV outcomes, but we find the data incredibly interesting nonetheless. We still think much more science is needed to figure out the role inflammation plays in cardiometabolic disorders, but these data are certainly compelling.

Characterization of the Biological Mechanisms of Empagliflozin through Large-Scale Proteomics

A Murthy, S Weiss, N Sattar, R Ostroff, S Williams, P Ganz, E Ferrannini

This study assessed the impact of empagliflozin on biomarker plasma proteins (proteomics), identifying 43 altered plasma protein levels (using a p<0.05 cutoff) after four weeks of treatment with the SGLT-2 inhibitor, pointing to the widespread effect of this agent on physiology. In 144 paired plasma samples from 72 participants (one sample at baseline, one after four weeks of treatment with empagliflozin 25 mg) across a spectrum of glucose tolerance, 3,713 different protein levels were evaluated with a modified aptamer assay. Empagliflozin was found to impact 43 plasma proteins – seven related to cardiomyocyte contraction or relaxation, five to iron handling, and one to sphingosine/ceramide metabolism (which the authors clarified is a known pathway of CV disease). Fourteen more plasma proteins implicated are involved in metabolic and lipid pathway regulation, and the most significant changes were (i) an increase in IGFBP1 and desmocollin and (ii) a decrease in ferritin. There’s no doubt that “omics” hold great promise in understanding and treating cardiometabolic disease, and even though we don’t yet know the specific implications of each protein identified in this study, we’re very intrigued by these results. If nothing else, it’s clear that SGLT-2 inhibitors are having a widespread impact on metabolism.

Understanding Public Awareness Regarding the Association Between Type 2 Diabetes and
Cardiovascular Disease: Findings From the “For Your SweetHeart” Survey

L Perreault, M Boardman, J Pak

BI’s Dr. Jonathan Pak presented new results from the Lilly/BI-sponsored online survey called “For Your SweetHeart,” and he emphasized low patient awareness of the correlation between type 2 diabetes and CV disease. Among people with type 2 diabetes (n=501), 52% were unaware of the heightened risk for CV disease and macrovascular events associated with their diagnosis, though they were more aware of risk for microvascular complications including retinopathy (57%), nephropathy (57%), and neuropathy (64%). Only 41% of the 501 patients were aware of their increased risk for MI, while only 43% were aware of their increased risk for stroke. Only 67% were aware that CV disease is the leading cause of death among people with diabetes. This adds color and detail to very preliminary survey results released by Lilly/BI in November 2016. While immensely disheartening, the survey – available from October 24, 2016 until November 1, 2016 – also found that 88% of patients with diabetes would modify their diet and 81% would talk to their healthcare provider if they knew about their increased CV disease risk. There is hope that with much better patient education, people with this chronic disease will be open to and engaged in CV risk reduction strategies, although we acknowledge that what individuals endorse in a survey doesn’t always align with behavior. These results parallel findings from Novo Nordisk/IDF’s “Taking Diabetes to Heart” survey, which also showed a greater understanding of microvascular rather than macrovascular complication risk. Selection bias may have played a role in these online surveys, with more engaged patients participating, so there’s a chance that awareness of CV risk was actually overestimated (and the reality is even worse). The “For Your SweetHeart” survey (n=1,505) included responses from 501 patients with type 2 diabetes, 364 who knew someone with type 2 diabetes, and 1,004 controls (neither). Adults ³18 years were encouraged to participate in the survey. Gender was largely balanced, and the 35-54 age range saw the most overall participants.

Exhibit Hall

Amgen

Amgen hosted one of the largest and most well-staffed booths in the ACC Exhibit Hall, promoting Repatha (evolocumab) and the PCSK9 inhibitor’s new indication for prevention of heart attack, stroke, and coronary revascularization in people with established CV disease. The sprawling space was dedicated almost entirely to Repatha and featured informational panels on both the drug and the importance of LDL-lowering more generally. The booth itself was very open, including a charging area for attendees to power up their electronics, where a large screen played a Repatha commercial on loop. A robust medical affairs corner offered information on CV disease states. The main area included take-home resources on best practices for gaining access to Repatha, including a co-pay card for monthly payments as low as $5 – navigating the payer landscape was a theme in common with Sanofi/Regeneron’s booth, though it was more heavily featured in the Amgen booth, as far as we could tell. Overall, Repatha might have been the most highly-advertised drug in Orlando (alongside a nice effort from Sanofi/Regeneron for Praluent).

AstraZeneca

AZ’s large, central booth was brightly-illuminated. White couches were surrounded mainly by advertisements for CV drug Brillinta, but a diabetes display also promoted SGLT-2 inhibitor Farxiga (dapagliflozin) and the new Bydureon (exenatide) BCise autoinjector, which was also available for demo. AZ has been a bit quiet about BCise launch since its approval last October, though the company originally planned to make the product in US pharmacies in 1Q18. We’re hopeful that the autoinjector (more patient friendly than the single-dose reconstitution kits or dual chamber Bydureon pen) can help further grow the Bydureon franchise. We also got to hear from AZ’s Dr. Jim McDermott (VP of US Medical Affairs for Diabetes) about DECLARE, the CVOT for Farxiga on track to report this year. Dr. McDermott expressed particular optimism about the impressive size of the trial (n=17,276, including a remarkable primary prevention cohort at ~60%), which could lead to very robust findings on Farxiga’s CV effects. We imagine the large sample size will help drive event rate, despite the lower-risk population overall relative to EMPA-REG (0% primary prevention) or CANVAS (34% primary prevention). Dr. McDermott also highlighted that DECLARE has a co-primary endpoint of CV death or hospitalization for heart failure, distinguishing it from EMPA-REG and CANVAS (where heart failure was only a secondary, exploratory endpoint). Could this support a heart failure indication prior to the read out of Dapa-HF, a dedicated study investigating dapagliflozin in chronic heart failure?

J&J (Janssen)

Although Janssen’s booth was dedicated almost entirely to blood thinner Xarelto, a small but mighty back corner highlighted the Invokana (canagliflozin) franchise. This reflects the strong and growing interest of cardiologists in SGLT-2 inhibitors. While J&J’s sNDA for a CV indication for Invokana, submitted October 2017, is still under review at FDA (a decision is expected late 3Q18 or early 4Q18), our sense is that many thought leaders have already accepted a CV benefit associated with canagliflozin. On-site reps didn’t have anything to share on where the sNDA might stand at FDA or Janssen, but we imagine they’re confident: FDA has already shown its openness to putting CV indications on diabetes drug labels with SGLT-2 inhibitor Jardiance (Lilly/BI’s empagliflozin) and GLP-1 agonist Victoza (Novo Nordisk’s liraglutide), and Invokana gave nearly identical results to Jardiance on three-point MACE. The one caveat is the signal for lower limb amputations in CANVAS (which was absent from EMPA-REG, at least the way the data was collected), although in our view, this shouldn’t interfere with FDA recognizing the tremendous CV benefit of the drug and granting the label change. A boxed warning for amputations is already on the Invokana label.

Lilly/BI

Lilly/BI’s mid-sized booth was easily recognized by the characteristic deep teal/yellow color scheme we’ve come to associate with SGLT-2 Jardiance (empagliflozin). We also noticed ads scattered about the convention center. The booth promoted Jardiance as the only SGLT-2 inhibitor with a CV indication on its label, and EMPA-REG OUTCOME data on display was targeted to cardiologists (highlighting all the reasons that CV specialists should care about – and prescribe – a diabetes therapy). Lilly/BI broke into the cardiology conference circuit following the December 2016 FDA approval of Jardiance’s CV indication; since then, we’ve noticed increasing enthusiasm for SGLT-2 inhibitors within the cardiology community.

Merck

We were disappointed to see that Merck’s booth was an unmanned charging station, though we understand that launch of SGLT-2 inhibitor Steglatro (ertugliflozin), Steglujan (ertugliflozin/sitagliptin), and Segluromet (ertugliflozin/metformin) is progressing and that Merck/Pfizer aren’t ready to promote their new SGLT-2 franchise to cardiologists since VERTIS CV is ongoing (expected to complete October 2019). Still, we see this as somewhat of a missed opportunity, since the cardiology field has shown distinct interest in the SGLT-2 class, and since the field as a whole is making an effort to get cardiologists more involved in diabetes care.

Merck/Pfizer’s pricing strategy could be the key to Steglatro’s success. At our local CVS in San Francisco, standalone Steglatro is listed at $319 for a 30-day supply (at both the 5 mg and 15 mg doses), which corresponds to a list price of ~$10.63/day. Merck/Pfizer previously announced a list price of $8.94/day, but we understand that price can differ across pharmacies. The average price for other standalone SGLT-2 inhibitors (Lilly/BI’s Jardiance, AZ’s Farxiga, and J&J’s Invokana) is $17/day – significantly higher than $8.94/day or even $10.63/day, which could make Steglatro an attractive option for payers and patients. A 30-day supply of fixed-dose SGLT-2/DPP-4 Steglujan costs $620 at our local CVS, which translates to ~$20.67/day. Merck/Pfizer previously announced a list price of $17.45/day, but nevertheless, Steglujan is cheaper than its competitors – Lilly/BI’s Glyxambi (empagliflozin/linagliptin) costs ~$22/day while AZ’s Qtern (dapagliflozin/saxagliptin) costs $24/day.

Novo Nordisk

A small but notable booth from Novo Nordisk was framed by two informational panels, promoting second-gen GLP-1 agonist Ozempic (semaglutide) – recently launched in the US – and GLP-1 agonist Victoza’s (liraglutide) CV indication in equal turn. Reps predominantly spoke about Victoza and the 13% risk reduction for three-point MACE seen in LEADER (p=0.01 for superiority vs. placebo), which is understandable, as only Victoza has a CV indication (Novo Nordisk will initiate the SOUL CVOT in mid-2018 toward a CV indication for Ozempic). No on-the-ground reps were able to comment on the relative interest they were seeing from cardiology attendees in Victoza vs. Ozempic, or in GLP-1 agonists as a class. Our overall sense from the meeting was that many cardiologists are still reluctant to incorporate GLP-1 agonists into their clinical practice, largely due to a lack of familiarity with injectable therapies on the part of the provider – indeed, Dr. Lawrence Leiter asserted on Saturday that providers are much more reluctant than patients to initiate injectable therapy. We do think this is changing – with PCSK9s as well as GLP-1s demonstrating highly-significant and clinically-meaningful cardioprotection – but we’d like to see this change faster.

Sanofi/Regeneron

With a major focus on Praluent (alirocumab), Sanofi/Regeneron’s booth embodied the emphasis on PCSK9 inhibitors that we saw throughout the meeting. Lots of content on Praluent was also fitting, considering that ODYSSEY Outcomes CVOT results were a key conference highlight (read our coverage above). Beyond this, however, we were excited to see that diabetes reps were on-site to promote next-gen basal insulin Toujeo (glargine U300) and Soliqua (insulin glargine/lixisenatide fixed-ratio combination), in line with the greater presence of diabetes companies and therapies overall at this year’s ACC. A large wall of the exhibit was dedicated to MyPraluent, a free program for patients that offers seven days/week live nurse support, injection reminders via text message, and other practical, educational resources. We’re glad to see this digital health effort from Sanofi/Regeneron to support adherence and patient success on Praluent, which we think could boost real-world outcomes – to this end, we wonder how many MyPraluent users there are. Observational data on display showed that, at the end of 12 months, ~25% more patients refilled their prescription when enrolled in MyPraluent (42% of patients vs. 34% without MyPraluent). We were also intrigued to learn that Praluent leads the PCSK9 inhibitor class in lives covered (meaning more lives covered vs. Amgen’s Repatha). That said, Praluent lags slightly behind Repatha in volume/sales, despite this greater commercial opportunity. In 2017, Praluent revenue totaled ~$194 million compared to ~$319 million for Repatha. As of 4Q17, Amgen held 61% of the market by value vs. Sanofi/Regeneron’s 39%.

-- by Ann Carracher, Megan Clyne, Payal Marathe, and Kelly Close