Podcast- Dr. Kieran Docherty, MD- Effect of Dapagliflozin, Compared With Placebo, According to Baseline Risk in DAPA-...

2 years 40 Views

From 2005 to 2011, Kieran Docherty studied medicine at the University of Glasgow. He is a Cardiology Specialist Registrar in the West of Scotland Deanery and a Clinical Research Fellow at the University of Glasgow's Institute of Cardiovascular and Medical Sciences. In this podcast Dr. Docherty discuss Effect of Dapagliflozin, Compared With Placebo, According to Baseline Risk in DAPA-HF.

Link to Abstract-


The authors wanted to see how dapagliflozin affected patients in the DAPA-HF study across the risk range.

The sodium-glucose cotransporter 2 inhibitor dapagliflozin reduced the risk of worsening HF events and cardiovascular death in individuals with HF and lower ejection fraction in the DAPA-HF (Dapagliflozin And Prevention of Adverse-outcomes in Heart Failure) trial.


Patients were classified into risk quintiles using the MAGGIC (Meta-analysis Global Group in Chronic Heart Failure) and PARADIGM-HF (Prospective Comparison of ARNI with ACEI to Determine Impact on Global Mortality and Morbidity in Heart Failure) PREDICT-HF (Risk of Events and Death in the Contemporary Treatment of Heart Failure) risk models. The authors looked at rates of the key composite outcome of a worsening HF event or cardiovascular death, as well as its components and all-cause mortality, by risk quintile, to see if risk influenced dapagliflozin's effect.


In DAPA-HF, the MAGGIC score was available for 4,740 patients out of 4,744 (median score 22 [IQR: 18–25]). A one-point rise in A1 was linked to an 8.2 percent (95 percent CI: 6.9%–9.4%) greater relative risk of the primary outcome (P 0.001). The primary endpoint advantage of dapagliflozin vs placebo was equal across the MAGGIC risk score spectrum (interaction P = 0.71). With dapagliflozin added to standard medication, the total relative risk reduction (26 percent) resulted in 7 fewer patients in the highest MAGGIC risk quintile experiencing a main outcome per 100 person-years of treatment, compared to 2 in the lowest quintile. The results were comparable with PREDICT-HF, however this model provided superior risk discrimination.


In DAPA-HF, the advantages of dapagliflozin were consistent over a wide range of baseline risk.


Individual clinical factors such as age, creatinine, and left ventricular ejection fraction (LVEF) are poor predictors of risk in individuals with heart failure (HF) (1-5). When assessing patients clinically, physicians informally integrate variables, although the most often used measure, the New York Heart Association (NYHA) classification, focuses on functional limitation associated to symptoms. Because of the lack of uniformity and subjectivity, as well as the minimal number of categories accessible (i.e., most patients are put in either class II or class III, despite classes I to IV are theoretically available), NYHA class has a limited function in predicting risk (1-5). At one end of the NYHA functional class continuum, well-treated patients may have minimal symptoms and little functional impairment, but they are nonetheless at significant risk, as seen by previous trials recruiting patients mostly in NYHA functional class II (6). Patients in NYHA functional class IV, on the other hand, are rarely enrolled in studies since this is often a temporary state from which patients improve or deteriorate to the point of hospital admission or death. Scores that incorporate all important characteristics provide a more accurate risk assessment. The MAGGIC (Meta-analysis Global Group in Chronic Heart Failure) score is one of the most extensively utilized since it just requires conventional clinical characteristics seen in most health-care settings (6-10). However, it was created before natriuretic peptides were widely used, and the only proven outcome it predicts is all-cause death (7). The PARADIGM-HF, a new prediction tool, was just released (Prospective comparison of ARNI with ACEI to Determine Impact on Global Mortality and morbidity in Heart Failure) PREDICT-HF (Risk of Events and Death in the Contemporary Treatment of Heart Failure), which contains natriuretic peptides, was constructed using the PARADIGM-HF study and verified in other large data sets (11). Using the MAGGIC and PREDICT-HF risk scores, we assessed the baseline risk of participants in the Dapagliflozin And Prevention of Adverse-outcomes in Heart Failure experiment. We also looked at the effect of dapagliflozin against placebo in terms of baseline risk.


DAPA-HF was a randomized, double-blind, controlled trial in patients with HF with decreased ejection fraction (HFrEF) that compared dapagliflozin 10 mg once day to placebo when added to usual therapy (12-14). The procedure was approved by ethics boards at each participating hospital, and all patients supplied written informed permission.

Patients in the research

Men and women in NYHA functional classes II–IV with an LVEF of less than 40%, a high NT-proBNP level, and who were receiving optimal pharmaceutical and device therapy were eligible for the study. Symptoms of hypotension or a systolic blood pressure (SBP) of less than 95 mm Hg, an eGFR of less than 30 mL/min/1.73 m2, type 1 diabetes mellitus, or another condition likely to prevent patient participation in the trial or severely limit life expectancy were among the key exclusion criteria (see the design paper for a complete list of exclusion criteria) (12).

Results of the research

The primary outcome was a composite of a worsening HF episode (unplanned hospitalization or an urgent visit culminating in HF IV therapy) or cardiovascular death, whichever came first. We looked at the primary composite outcome, its components, and the predetermined secondary endpoint of all-cause death in this study.

Adverse events leading to trial treatment discontinuation and adverse events of interest were included in pre-specified safety evaluations (ie, volume depletion, renal events, major hypoglycemia, fractures, diabetic ketoacidosis, amputation).

The MAGGIC risk score is a measure of how dangerous a person is.

In a nutshell, the MAGGIC score was created by examining 31 potential variables in 39,372 patients involved in 30 clinical trials and cohort studies using a multivariable risk model (7,15). Age (per 10 years), male sex, BMI (per 1 kg/m2 increase up to 30 kg/m2), current smoking, diabetes, SBP (per 10 mm Hg increase), NYHA functional class, LVEF (per 5% increase up to 40%), COPD, HF duration >18 months, creatinine (per 10 mol/L up to 350 mol/L), ACE inhibitor/ARB use, and -blocker use were identified as thirteen independent predictors of (significant interactions between LVEF and age and LVEF and SBP were also identified). The result was a simple integer score. The maximum score is 57, and a risk calculator (http://www.heartfailurerisk.org/) is available online.

PREDICT-HF is a risk assessment tool.

The PREDICT-HF risk calculator was created using a multivariable risk model based on an analysis of 63 candidate variables in 8,011 patients enrolled in the PARADIGM-HF (Prospective Comparison of ARNI with ACEI to Determine Impact on Global Mortality and Morbidity in Heart Failure) trial, and validated in 7,016 patients in the ATMOSPHERE (Aliskiren Trial to Minimize Outcomes in Patients with Heart Failure) trial and 2,794 (11). Thirty independent predictors of death from any cause were discovered (Supplemental Table 1). The total of the multiplication of each variable in the model by the -coefficient was used to create a PREDICT-HF pseudoscore for this investigation. A constant from PARADIGM-HF, the derivation cohort, was then used to determine baseline survival. A PREDICT-HF score was multiplied by a factor of ten for comparison with the MAGGIC risk score. The highest score recorded was 83.8. There is an online risk calculator (http://www.predict-hf.com).

Analytical statistics

Each of the MAGGIC and PREDICT-HF scores were grouped into quintiles. In a Cox regression model, the effect of dapagliflozin versus placebo on each prespecified outcome throughout the risk spectrum was investigated. In a Cox regression model containing an interaction term between score quintile and treatment, the effect of risk and treatment on the occurrence of each outcome was investigated. We utilized a semiparametric proportional rates model to examine the treatment impact and quantify the treatment difference for the composite endpoint of total (including recurrent) HF hospitalizations and cardiovascular death. Predetermined adverse events were also investigated based on the risk score category. By including a risk score element in the models outlined above, the association between a 1-point rise in MAGGIC and PREDICT-HF scores and the probability of outcomes was investigated. Exponentiating the -coefficient for the score variable from these models yielded the increase in risk associated with a 1-point increase (ie, the hazard or rate ratio per 1-point increase in score).

Patients with HF for 18 months or longer receive 2 points on the MAGGIC score. In DAPA-HF, however, the duration of HF was classified as 0 to 3 months, 3 to 6 months, 6 to 12 months, 1 to 2 years, 2 to 5 years, and >5 years. In the main analysis, we added 2 points to a patient's score if they had HF for less than a year, and we did the same in a sensitivity analysis (Supplemental Table 2) if they had HF for more than two years.

DAPA-HF provided serum potassium, hemoglobin, bilirubin, aspartate aminotransferase, urea, and NT-proBNP values. The medians from the PARADIGM-HF cohort were utilized for the other laboratory data (albumin, uric acid, percent monocytes, absolute neutrophils, chloride, low-density lipoprotein, and triglycerides).

Harrell's C-statistic was used to examine model discrimination for all-cause death. STATA version 16.1 was used for all of the analyses. A statistically significant P value of 0.05 was used.


A total of 4,744 patients, ranging in age from 22 to 94, were randomized. The average age of the participants was 66.3 years, with 76.6 percent of men, and 67.5 percent of patients in NYHA functional class II, 31.6 percent in functional class III, and 0.9 percent in functional class IV.