AHA 2017 Scientific Sessions (American Heart Association)

November 11-15, 2017; Anaheim, CA; Days #1-2 Highlights – Draft

Executive Highlights

Greetings from Anaheim! Disneyland tourists are joined this week by cardiologists from all over the world, visiting the city of Anaheim for the American Heart Association’s (AHA) annual Scientific Sessions. This meeting is off to a phenomenal start, and we’re already noticing more on diabetes than ever before, with research at the intersection of diabetes, CV disease, and renal disease. This report details our top five highlights from the weekend, but there’s much more to come, as you’ll see in our AHA 2017 conference preview (including new analyses from CANVAS, EXSCEL, and EMPA-REG OUTCOME). Stay tuned for more on-the-ground coverage in the days ahead!

Top Five Highlights

1. Dr. Mikhail Kosiborod reviewed CVD-REAL results supporting that (i) the CV benefits SGLT-2 inhibitors have demonstrated in RCTs may extend to primary prevention in a broad population of patients with type 2 diabetes, and that (ii) CV benefit is likely an SGLT-2 inhibitor class effect. In providing the discussant, Dr. Philippe Gabriel Steg considered whether SGLT-2 inhibitors should become first-line therapy in patients with diabetes without baseline CV disease.

2. Dr. Faiez Zannad decried the common exclusion of patients with low eGFR from diabetes trials, even though chronic kidney disease (CKD) is so often comorbid with type 2 diabetes. He highlighted CKD as the strongest predictor of CV events, before noting that ~50% of studies across therapeutic areas exclude patients with kidney disease due to potentially diminished treatment effects, safety concerns, complex pathophysiology, or difficulty in recruitment and retention. Dr. Zannad alluded to a bright spot in upcoming studies of SGLT-2 inhibitors for diabetic kidney disease and chronic kidney disease.

3. A post-hoc analysis of ACCORD was presented on a poster, and identified five key predictors of hospitalization or death due to heart failure in this study population with type 2 diabetes: (i) range of urinary creatinine (UCr), (ii) history of heart failure, (iii) minimum urinary albumin, (iv) minimum urinary albumin-to-creatinine ratio (UACR), and (v) loop diuretic use. This work aimed to understand which biomarkers can best predict heart failure in patients with type 2 diabetes, which will be a key insight as cardioprotection takes center stage in diabetes care, and as the field pushes slowly but surely toward personalized treatment plans.

4. Researchers from the University of Toronto presented a poster showing synergistic CV benefits to co-administration of SGLT-2 inhibitor dapagliflozin (AZ’s Farxiga) and GLP-1 agonist exenatide (AZ’s Bydureon) in a mouse model.

5. Louisville’s Dr. Harold Bays presented a post-hoc analysis of the ODYSSEY program for Sanofi/Regeneron’s PCSK9 inhibitor Praluent (alirocumab). The poster focused on the 1% of patients treated with alirocumab who did not achieve LDL reductions >15% from baseline (33 participants out of 3,120 across 10 studies), and concluded that the hyporesponsiveness was more likely due to adherence issues rather than anti-drug antibodies.

Top Five Highlights

1. Drs. Kosiborod and Steg on CVD-REAL: SGLT-2 Inhibitors for Primary CV Prevention?

Dr. Mikhail Kosiborod reviewed CVD-REAL results supporting that (i) the CV benefits SGLT-2 inhibitors have demonstrated in RCTs may extend to primary prevention in a broad population of patients with type 2 diabetes, and that (ii) CV benefit is likely an SGLT-2 inhibitor class effect. The AZ-sponsored CVD-REAL study (n=~300,000) is an observational, real-world comparison of CV outcomes between patients newly taking SGLT-2 inhibitors vs. other glucose-lowering drugs. As we learned when Dr. Kosiborod presented initial CVD-REAL results at ACC 2017, SGLT-2 inhibitors showed significant risk reduction for heart failure hospitalization and all-cause mortality. On the primary outcome of hospitalization for heart failure, CVD-REAL found an average 39% relative risk reduction across the US, Norway, Denmark, Sweden, Germany, and the UK with SGLT-2 inhibitors (HR=0.61, 95% CI: 0.51-0.73, p<0.001). Excluding Germany (due to a lack of complete data), similar results were seen for all-cause death (51% relative risk reduction, HR=0.49, 95% CI: 0.41-0.57, p<0.001) and for the composite of heart failure hospitalization/all-cause death (46% relative risk reduction, HR=0.54, 95% CI: 0.48-0.60, p<0.001). With 87% of included patients having no baseline CV disease (i.e. representing a primary prevention cohort), and with the data showing no effect modification in any subgroup, CVD-REAL suggests that cardioprotection with SGLT-2 inhibitors may extend to those with type 2 diabetes but no established CV disease. Notably, the FDA approved a CV indication for Lilly/BI’s Jardiance (empagliflozin), and this applies to people with type 2 diabetes facing high CV risk. It’ll be interesting to see if another SGLT-2 inhibitor receives a broader CV indication – J&J submitted CANVAS data for consideration on the Invokana (canagliflozin) label, requesting an indication for high CV risk patients, while AZ’s ongoing DECLARE trial for Farxiga (dapagliflozin) has enrolled a larger primary prevention cohort (~50% vs. ~33% in CANVAS and ~1% in EMPA-REG OUTCOME). We appreciate the limitations to conducting an RCT in a lower-risk population, since these are event driven trials, and lengthy/expensive ones at that. To this end, we believe the real-world evidence from CVD-REAL is highly-suggestive and compelling, and in general we’d love to see cardioprotective therapies like SGLT-2 inhibitors (as well as GLP-1 agonists) initiated earlier in the course of treatment to prevent CV disease before the first event.

  • Despite geographically-skewed use of different SGLT-2 inhibitors – Farxiga made up ~90% of exposure time in Europe (where it was first-to-market), while Invokana made up ~75% in the US (where it was first-to-market) – these CV benefits were significant or near-significant in every individual country, weighing in favor of a class effect. The DECLARE CVOT of AZ’s Farxiga will shed further light on this, as the third RCT/outcomes study to report for an SGLT-2 inhibitor (Merck/Pfizer’s VERTIS CV for ertugliflozin is also ongoing, expected to complete in October 2019). Dr. Kosiborod noted that CVD-REAL results could be impacted by an unmeasured confounding.
  • In providing the discussant, Dr. Philippe Gabriel Steg considered whether SGLT-2 inhibitors should become first-line therapy in patients with diabetes without baseline CV disease. Dr. Steg argued that without question, SGLT-2 inhibitors should join ACE inhibitors, statins, beta blockers, and antiplatelet agents as the fifth key secondary prevention drug in patients with a prior CV event(s) and diabetes. While optimistic about the primary prevention benefits observed in CVD-REAL, Dr. Steg cautioned the audience against drawing conclusions of causality from observational studies, though he underscored how CVD-REAL complements and is consistent with RCTs published to-date. Ultimately, with little geographic or drug-based variation in effect, Dr. Steg maintained that CVD-REAL confirms results from EMPA-REG OUTCOME and CANVAS, extending them in a population 87% devoid of CV disease and pointing – definitely pointing – to their use in primary prevention. SGLT-2 inhibitor use is a hot topic within the larger treatment guidelines conversation, and we’ve heard the esteemed Dr. Ralph DeFronzo endorse SGLT-2 inhibitors, along with GLP-1 agonists and pioglitazone, as first-line therapy in all patients with type 2 diabetes. IDF 2017 is dedicating a debate to the question of first-line SGLT-2 inhibitors, and we look forward to reporting on that conversation in December.

2. Dr. Zannad Advocates for More Emphasis on the Kidneys in Large Outcomes Trials

Dr. Faiez Zannad decried the common exclusion of patients with low eGFR from diabetes trials, even though chronic kidney disease (CKD) is so often comorbid with type 2 diabetes. He highlighted CKD as the strongest predictor of CV events, before noting that ~50% of studies across therapeutic areas exclude patients with kidney disease due to potentially diminished treatment effects, safety concerns, complex pathophysiology, or difficulty in recruitment and retention (these patients already face substantial treatment burdens, with complicated medication regimens, etc.). In considering diabetes, specifically, Dr. Zannad focused on EMPA-REG OUTCOME (Lilly/BI’s CVOT for SGLT-2 inhibitor Jardiance), which excluded patients with eGFR <30 ml/min/1.73 m2 and enrolled 26% of patients with eGFR <60 ml/min/1.73 m2 and only ~8% with eGFR <45 ml/min/1.73 m2. Very notably, CV outcomes with empagliflozin vs. placebo did not differ significantly based on eGFR. Moreover, microvascular analysis found a 39% risk reduction for the composite renal endpoint (incident or worsening diabetic nephropathy) with empagliflozin vs. placebo (HR=0.61, 95% CI: 0.55-0.69, p<0.001), which suggests that this agent may actually exert a beneficial, protective effect on the kidneys. To bolster this notion of renal protection, CANVAS for J&J’s SGLT-2 inhibitor Invokana (canagliflozin) found a similar microvascular benefit. For now, Jardiance (and all SGLT-2 products, for that matter) remains contraindicated for people with severe renal impairment, end-stage renal disease, and dialysis, and it is not recommended to be initiated when eGFR is <45 ml/min/1.73 m2. As Dr. David Fitchett emphasized at ESC, however, this contraindication is based mainly on a presumed attenuation of efficacy, and new outcomes data may overturn this. Dr. Zannad also implied that CKD could be the next frontier in diabetes therapy, and we see this frontier already opening up: J&J’s CREDENCE trial of Invokana (canagliflozin) in diabetic kidney disease should complete in June 2019, while AZ’s CKD outcomes trial for Farxiga (dapagliflozin) should complete in November 2020. Lilly/BI have announced plans for a similar CKD outcomes trial of Jardiance, though no timing has been disclosed. Dr. Zannad advocated for more inclusion of patients with kidney disease in CV trials – more specifically, he suggested that patients with advanced kidney disease be enrolled in all trials, but allowed to be excluded from primary efficacy analyses. This would ameliorate concern from sponsors over blurring of efficacy signals, while also allowing data to be gathered on these underserved patients. To our understanding, this might necessitate larger enrollment numbers overall to sufficiently power the non-kidney-disease analysis, but we agree that far more work needs to be done to figure out optimal care for individuals with reduced kidney function.

3. Urinary Biomarkers Identified as Predictors of Heart Failure in Type 2 Diabetes Population (Post-Hoc Analysis of ACCORD)

A post-hoc analysis of ACCORD was presented on a poster, and identified five key predictors of hospitalization or death due to heart failure in this study population with type 2 diabetes: (i) range of urinary creatinine (UCr), (ii) history of heart failure, (iii) minimum urinary albumin, (iv) minimum urinary albumin-to-creatinine ratio (UACR), and (v) loop diuretic use. An alternative Cox regression found that, for women, (i) range of UCr, (ii) minimum systolic blood pressure, (iii) shortness of breath, (iv) loop diuretic use, and (v) nitrate use were the most important predictors (C-statistic=0.87, where a perfect fit=1.00). For men, it was (i) range of UCr, (ii) HDL and LDL cholesterol, (iii) loop diuretic use, (iv) minimum systolic blood pressure, (v) shortness of breath, and (vi) CV disease history that emerged as the most important predictors (C-statistic=0.88). In total, 378 variables on demographics and clinical and laboratory data, and their change from baseline to follow-up, were analyzed as possible predictors of heart failure using a machine learning methodology called random survival forests (RSF). The top 20 predictors from 500 trees were included in a stepwise multivariate analysis using a Cox proportional hazards model, based on heart failure hospitalization or heart-failure related mortality events occurring in 146/3,952 women and 298/6,299 men. Interestingly, the study authors noted that the predictive accuracy of the RSF and Cox regression models was “markedly” increased when variables of range and change from baseline were included, from a C-statistic of 0.64 to 0.87 for the RSF. This work aims to understand which biomarkers can best predict hospitalization and death due to heart failure in patients with type 2 diabetes, which will be a key insight as cardioprotection takes center stage in diabetes care, and as the field pushes slowly but surely toward personalized treatment plans. We’d also note that all of the most important measures to surface in this study are commonly collected in patients with diabetes and CV disease, and we see potential applications for this work in risk stratification/targeting the highest-risk patients. Further, we’d love to see these biomarkers evaluated in a real-world dataset, to see if/how they hold up outside of a clinical trial setting.

4. SGLT-2 Inhibitor (Dapa) + GLP-1 Agonist (Exenatide) Combination Shows Superior Cardioprotection vs. Either Agent Alone in a Mouse Model

Researchers from the University of Toronto presented a poster showing synergistic CV benefits to co-administration of SGLT-2 inhibitor dapagliflozin (AZ’s Farxiga) and GLP-1 agonist exenatide (AZ’s Bydureon) in a mouse model. Non-diabetic male mice were randomized to dapagliflozin alone, exenatide alone, both dapagliflozin/exenatide, or to placebo for seven days, and then received an ischemia-reperfusion injury. Measurements of functional recovery and infarct size were then collected two-days post injury and 28 days post-injury. While all three active treatments produced superior cardiac recovery vs. placebo after two days, the combination only demonstrated superiority vs. both monotherapies after 28 days: Left ventricular ejection fraction (LVEF), which measures the proportion of total blood volume pumped out of the left ventricle with each heartbeat (higher is thus better, within range), improved to 41% with dapagliflozin and with exenatide, but improved to 56% with the SGLT-2/GLP-1 combination approach, and to only 30% with placebo (p<0.05 for all comparisons). Fractional shortening improved to 13% with dapagliflozin, 14% with exenatide, 19% with both, and 9% with placebo (p<0.01 for all comparisons). These cardioprotective effects were associated with reduced infarct size. The poster concluded that the combination regimen showed superior CV benefit vs. dapagliflozin alone (p<0.001), vs. exenatide alone (p<0.05), and vs. placebo (p<0.05). First author Dr. Dorrin Zarrin Khat explained the rationale for this study – agents from both these drug classes have independently demonstrated risk reduction for CV outcomes in large, human clinical trials – and shared that her research team next plans to replicate this procedure in a mouse model of diabetes. To be sure, we are extremely intrigued by the potential for added cardioprotection with SGLT-2/GLP-1 combination therapy, especially because agents from each class are thought to exert CV benefit through distinct mechanisms of action. This animal research may be early-stage, but anything that lays the groundwork for more in-human study of this particular combination – we find that exciting (after all, SGLT-2 inhibitors and GLP-1 agonists are among the most advanced treatments for type 2 diabetes available today, offering A1c-lowering efficacy, weight loss, and possible CV as well as renal protection). AZ did investigate Farxiga/Bydureon co-administration in the DURATION-8 trial, and data showing superior glycemic and weight loss efficacy with the combination is now on both product labels in the US and EU. We understand that conducting a CVOT for the combination approach would be a tall order, considering the massive investment of time and resources that goes into one of these outcomes studies… but boy would we love to see this. Lilly also has both an SGLT-2 inhibitor (Jardiance) and a GLP-1 agonist (Trulicity) in its diabetes portfolio, and Jardiance (empagliflozin) has already shown compelling CV risk reduction in EMPA-REG OUTCOME. In the meantime, studies like this one contribute to our mechanistic understanding of cardioprotection from diabetes drugs. And as the great Dr. Daniel Drucker articulated recently, “if you want to make drugs better, with next-generation versions with even more potent glucose-lowering and even more cardioprotection, then you do need to know how they work to know where to focus attention for innovation.”

5. Pooled Analysis of 10 ODYSSEY Studies Supports Robust Lipid-Lowering Effects of Sanofi/Regeneron’s PCSK9 Inhibitor Praluent; “Non-Responders” Attributed to Adherence Issues

Louisville’s Dr. Harold Bays presented a post-hoc analysis of the ODYSSEY program for Sanofi/Regeneron’s PCSK9 inhibitor Praluent (alirocumab). The poster focused on the 1% of patients treated with alirocumab who did not achieve LDL reductions >15% from baseline (33 participants out of 3,120 across 10 studies), and concluded that the hyporesponsiveness was more likely due to adherence issues rather than anti-drug antibodies. Though daily caregivers reported at least 80% adherence from all 33 of these individuals, PK assessments of 26 of them found that alirocumab was undetectable in the blood for 13 among this small group, meaning these people were not injecting the medication as prescribed (forgetting doses, skipping doses, or not taking the drug altogether). Five others discontinued treatment early, and PK values were missing for two individuals. For the remaining six participants in this cohort, researchers found low adherence to concomitant lipid-lowering therapies, namely statins. Dr. Bays emphasized that the most pronounced effects on hyperlipidemia are achieved with PCSK9 inhibitors on top of maximally-tolerated statin therapy (indeed, this is a key piece of Praluent’s FDA-approved indication), and he pointed to the discontinuation of statins as another form of low adherence. This pooled analysis supports Praluent’s robust lipid-lowering efficacy, given that extremely few individuals experienced <15% LDL drop from baseline in the first place, and those who did were most likely not taking full doses of the drug throughout the treatment period. As we hear often at diabetes meetings, the most powerful medicine on the planet won’t do patients any good if they don’t take it, and it’s up to healthcare teams to support and enhance patient engagement. Moreover, Pfizer’s PCSK9 candidate bococizumab was discontinued from phase 3 due to the development of anti-drug antibodies and a subsequent attenuation of LDL-lowering effect. Dr. Bays’ study addresses and refutes that this issue extends to alirocumab, providing additional reassurance overall that Praluent is a very effective treatment for dyslipidemia. According to Sanofi’s recent 3Q17 update, data from ODYSSEY Outcomes (CVOT expected to complete in December 2017) will inform next commercial steps for Praluent. We await these results on the edge of our seats, and we imagine this large trial could even more definitively put to bed any slight, lingering concerns over anti-drug antibodies.

 

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