Clinical Research

Abstract

Objectives

This study aimed to assess the prognostic importance of hyponatremia and the effects of dapagliflozin on serum sodium in the DAPA-HF (Dapagliflozin And Prevention of Adverse outcomes in Heart Failure) trial.

Background

Hyponatremia is common and prognostically important in hospitalized patients with heart failure with reduced ejection fraction, but its prevalence and importance in ambulatory patients are uncertain.

Methods

We calculated the incidence of the primary outcome (cardiovascular death or worsening heart failure) and secondary outcomes according to sodium category (≤135 and >135 mmol/L). Additionally, we assessed: 1) whether baseline serum sodium modified the treatment effect of dapagliflozin; and 2) the effect of dapagliflozin on serum sodium.

Results

Of 4,740 participants with a baseline measurement, 398 (8.4%) had sodium ≤135 mmol/L. Participants with hyponatremia were more likely to have diabetes, be treated with diuretics, and have lower systolic blood pressure, left ventricular ejection fraction, and estimated glomerular filtration rate. Hyponatremia was associated with worse outcomes even after adjustment for predictive variables (adjusted HRs for the primary outcome 1.50 [95% CI: 1.23-1.84] and all-cause death 1.59 [95% CI: 1.26-2.01]). The benefits of dapagliflozin were similar in patients with and without hyponatremia (HR for primary endpoint: 0.83 [95% CI: 0.57-1.19] and 0.73 [95% CI: 0.63-0.84], respectively, P for interaction = 0.54; HR for all-cause death: 0.85 [95% CI: 0.56-1.29] and 0.83 [95% CI: 0.70-0.98], respectively, P for interaction = 0.96). Between baseline and day 14, more patients on dapagliflozin developed hyponatremia (11.3% vs 9.4%; P = 0.04); thereafter, this pattern reversed and at 12 months fewer patients on dapagliflozin had hyponatremia (4.6% vs 6.7%; P = 0.003).

Conclusions

Baseline serum sodium concentration was prognostically important, but did not modify the benefits of dapagliflozin on morbidity and mortality in heart failure with reduced ejection fraction. (Study to Evaluate the Effect of Dapagliflozin on the Incidence of Worsening Heart Failure or Cardiovascular Death in Patients With Chronic Heart Failure [DAPA-HF]: NCT03036124)

Discussion

In a contemporary, well-treated ambulatory cohort of patients with HFrEF, most of whom had mild symptoms, the prevalence of hyponatremia was low (8.4%) and there were few cases of severe hyponatremia (0.06%). However, hyponatremia remained an independent predictor of outcomes despite adjustment for other prognostic variables, including NT-proBNP. The benefit of dapagliflozin was consistent across the range of sodium concentrations measured at baseline. Dapagliflozin had a small biphasic effect on serum sodium concentration. Initially, compared with placebo, dapagliflozin led to a small, although statistically significant, decrease in sodium. However, after 2 weeks, the opposite pattern was observed.

Although hyponatremia is recognized as the most common electrolyte disorder among hospitalized patients with HF, there are few reports of the prevalence of hyponatremia in ambulatory patients with HFrEF and none in patients comprehensively managed with contemporary guideline-recommended medical therapy.^9-11 Even accounting for different definitions, the prevalence of hyponatremia in our outpatient cohort (8.4%) was less than half that reported in hospitalized patients (generally 20% to 25%).^1-4

Although most cases of hyponatremia in the DAPA-HF trial were mild, low sodium still predicted worse outcomes. This excess risk persisted despite adjustment for other recognized prognostic variables, many of which showed an imbalance between patients with and without hyponatremia. Indeed, we know of no prior study where such extensive adjustment was made, including for natriuretic peptide level, in ambulatory patients. ^9-11 Moreover, most studies to date have only reported the association between hyponatremia and all-cause mortality, whereas we have also shown that low sodium was independently predictive of worsening HF events (principally HF hospitalization) and symptoms.^16,17

The prognostic importance of a single sodium measurement was remarkable given the rapid and frequent resolution of hyponatremia on rechecking blood chemistry. In the placebo group, almost half of cases of hyponatremia had resolved at the 2-week measurement after randomization and about two-thirds of cases had resolved by 8 months. This substantial recategorization occurred because the initial measurement was only slightly below normal in many patients. However, almost as many people in the placebo group developed new hyponatremia at each timepoint during follow-up as showed resolution of hyponatremia. Dapagliflozin had a surprising, previously unrecognized, biphasic effect on new hyponatremia. The incidence of hyponatremia was increased during the first 14 days after randomization but was decreased thereafter in patients treated with dapagliflozin compared to placebo. The explanation for this pattern is uncertain. The initial osmotic and natriuretic diuresis induced by SGLT2 inhibitors causes an increase in vasopressin secretion and a reduction in free-water clearance, experimentally and clinically, which might account for the early transient reduction in serum sodium concentration.^18-21 The subsequent effects on serum sodium concentration are harder to predict given the direct effects of SGLT2 inhibitors and the compensatory responses to these. The diuresis induced by SGLT2 inhibitors is believed to lead to a reduction in intravascular volume and blood pressure, and the increased delivery of sodium to the distal nephron results in a decline in eGFR by inducing tubuloglomerular feedback.^22-25 However, it has been hypothesized that SGLT2 inhibitors reduce blood volume less than conventional diuretics.²⁶ Although the initial decrease in sodium mirrors the early decline in eGFR after starting dapagliflozin, subsequently, serum sodium concentration increased more in the dapagliflozin group than the placebo group, to the extent that the mean concentration was eventually significantly higher in the dapagliflozin group. Although the initial decrease in eGFR also partially recovers, eGFR does not recover back to the same level as in the placebo group (as is also observed in other trials and real-world data over the same period) and eGFR does not crossover as for sodium.^27,28 So, it seems unlikely that the effect of SGLT2 inhibitors of eGFR alone explain the early effect on sodium, although it might explain the longer-term effect if there is a relative increase in free-water clearance with these agents (as seems likely) and sodium excretion is maintained (and sodium retention does not occur), which may be the case if eGFR is maintained. The complexity of these effects is reflected in the seeming paradox of the early decline in serum sodium concentration occurring contemporaneously with an increase in hematocrit, questioning whether the latter can be wholly explained by volume contraction. Although detailed analyses of change in hemoglobin have been reported in other trials, the effect of other SGLT2 inhibitors on serum sodium has not been reported.²⁹ Irrespective of the possible mechanisms, the important overarching finding was that after 14 days, patients treated with dapagliflozin were less likely to develop new hyponatremia and more likely to show resolution of existing hyponatremia than individuals treated with placebo, which may be a favorable effect of SGLT2 inhibition in HF. The benefits of dapagliflozin on the primary and secondary cardiovascular outcomes were consistent in patients with and without hyponatremia (and across the range of serum sodium concentration at baseline), despite the initial transient small decline in serum sodium concentration. Indeed, the absolute risk reduction with dapagliflozin was 1.5- to 2.0-fold greater in patients with hyponatremia than in those without. Similarly, dapagliflozin was also well-tolerated in patients with hyponatremia, and the safety of dapagliflozin was similar in patients with and without hyponatremia.

Source: JACC