Vericiguat in Heart Failure With Reduced Ejection Fraction With High Natriuretic Peptides

A Case of Too Little, Too Late?

Jonathan W. Cunningham, MD; Pardeep S. Jhund, MBCHB, MSC, PHD


JACC Heart Fail. 2020;8(11):940-942. 

In the VICTORIA (Vericiguat Global Study in Subjects with Heart Failure with Reduced Ejection Fraction) trial, the soluble guanylate cyclase stimulator vericiguat reduced the risk of cardiovascular (CV) death or heart failure (HF) hospitalization.[1] Vericiguat represents an important step forward in the treatment of HF with reduced ejection fraction (HFrEF). The size of the relative effect (hazard ratio [HR]: 0.90; 95% confidence interval [CI]: 0.82 to 0.98; p = 0.02) was modest compared with other therapies for HFrEF and was driven by a reduction in HF hospitalizations (HR: 0.90; 95% CI: 0.81 to 1.00) but not CV death (HR: 0.93; 95% CI: 0.81 to 1.06). Although these results would ordinarily make clinicians less enthusiastic about a new therapy, what made the results of the VICTORIA trial important was the patient population. The investigators purposely enrolled a population with recent worsening HF, higher natriuretic peptides, and consequently, a much higher risk of morbidity and mortality than populations in recent trials. The minimum N-terminal pro–B-type natriuretic peptide (NT-proBNP) levels that allowed entry into the VICTORIA trial (1,000 pg/ml in sinus rhythm and 1,600 pg/ml in atrial fibrillation) were approximately double those in the DAPA-HF (Dapagliflozin and Prevention of Adverse Outcomes in Heart Failure) trial (400 pg/ml in sinus rhythm with a recent HF event, 600 pg/ml in sinus rhythm without a recent HF event, and 900 pg/ml in atrial fibrillation) and the PARADIGM-HF (Prospective Comparison of ARNI with ACEI to Determine Impact on Global Mortality and Morbidity in Heart Failure) trial (400 pg/ml in with recent hospitalization, 600 pg/ml without hospitalization). NT-proBNP is a well-validated and powerful marker of risk this in this population.[2] These criteria resulted in a control group event rate 2 to 2.5 times greater than those in the PARADIGM-HF and DAPA-HF trials.

A key question in interpreting the VICTORIA trial is whether the more modest relative efficacy, compared with other recent trials, was due to a less effective drug, the play of chance, or a more challenging patient population. There is no reason to believe vericiguat was ineffective. HF hospitalizations, including total hospitalizations, were clearly reduced. The play of chance and power were a potential answer. The trial was powered to detect a HR of 0.80 on the outcome of CV death with 782 deaths, so with 855 deaths, they might have failed to detect a more modest effect on CV death as suggested by the final point estimate and 95% CIs. Perhaps this population was just too sick to benefit from this intervention. In a pre-specified subgroup analysis of the VICTORIA trial, there was a significant interaction by NT-proBNP level with a suggestion of no benefit in those in the highest quartile of NT-proBNP levels (>5,314 pg/ml). Could the results of the VICTORIA trial, and, in particular, the interaction with NT-proBNP be explained by the enrollment of sicker patients with high NT-proBNP?

In this issue of JACC: Heart Failure, Ezekowitz et al.[3] and the VICTORIA trial investigators examined this potential interaction in greater detail. Participants with lower baseline NT-proBNP benefited more from vericiguat than those with higher NT-proBNP. Dichotomization of continuous data can be problematic, but in this analysis, the treatment interaction remained highly statistically significant when NT-proBNP was measured on a continuous scale (interaction p = 0.001 for the primary endpoint and interaction p = 0.012 for CV death). This raised the possibility of a threshold of NT-proBNP above which a patient was too sick to benefit. In a post hoc analysis that excluded patients with NT-proBNP >8,000 pg/ml, there was a statistically significant reduction in CV death. The investigators also intriguingly reported a statistically significant interaction with HF hospitalizations, with potentially no reduction in HF hospitalizations in those with NT-proBNP >8,000 pg/ml (p for interaction = 0.003).

These important data raise the possibility that high-risk patients, as measured by NT-proBNP levels, had too advanced HF to benefit from vericiguat in VICTORIA. Perhaps the effect size for vericiguat would have been similar to sacubitril/valsartan or dapagliflozin had all 3 drugs been studied in patients with the same underlying risk. A higher dose of vericiguat might have provided a different result, as is being currently tested in other trials of vericiguat.[4] However, subgroup interactions in clinical trials should be viewed with skepticism, because spurious interactions might be identified by multiple hypothesis testing. Two key criteria on which to evaluate subgroup interactions are statistical strength and biologic plausibility.

The strong statistical significance of the NT-proBNP treatment interaction for the primary composite endpoint in VICTORIA (p for interaction = 0.001) argued for a true effect. In the primary manuscript, 11 pre-specified subgroups were tested, including quartiles of baseline NT-proBNP.[1] Although the interaction p value for the quartile-based analysis was not reported, the continuous interaction would easily withstand a Bonferroni correction for 11 hypothesis tests.

Although there is no clear mechanism for this interaction, at least 2 might be hypothesized. The first is related to the design of the study. The time-to-event primary endpoint, which included a measure of mortality, might bias HRs toward the null when event rates were high and there was a reduction in the length of follow-up. Benefits in mortality take time to become evident. A similar issue was reported in the EMPEROR-Reduced (Empagliflozin Outcome Trial in Patients with Chronic Heart Failure and a Reduced Ejection Fraction) trial, in which empagliflozin reduced HF hospitalizations but not CV death in a similarly enriched high-risk population of patients with HFrEF.[5] The interaction with HF hospitalizations in the VICTORIA trial is more difficult to explain. Patients with more advanced HF had more HF hospitalizations; therefore, we might expect an effect on HF hospitalizations to remain evident at the higher end of the NT-proBNP range. A second possible mechanism is that patients with higher NT-proBNP levels had higher cyclic guanosine monophosphate (cGMP) levels, and therefore, may benefit less from augmentation of cGMP signaling. Natriuretic peptides (in addition to nitric oxide) stimulate the hydrolysis of guanosine triphosphate to cGMP by guanylate cyclases. In the community, plasma NT-proBNP and cGMP levels are correlated.[6] An investigation to determine whether patients with higher baseline NT-proBNP also had smaller increases in urine cyclic GMP with vericiguat could evaluate this hypothesis and might help to explain why there was also an interaction with HF hospitalizations.

What do these results mean for clinicians and patients who must decide which medication to start first in moderate or high-risk HFrEF? The inclusion of some high-risk patients in VICTORIA might have obscured efficacy similar to that of sacubitril/valsartan or an SGLT2 inhibitor. Perhaps in patients with NT-proBNP >8,000 pg/ml and event rates over 75 per 100 patient-years, it was difficult to prevent the inevitable decline. There was undoubtedly a reduction in the risk of CV death and HF hospitalization in all but those with the highest NT-proBNP levels. In recent registry-based analyses of NT-proBNP, most patients had NT-proBNP levels <8,000 pg/ml threshold and would therefore stand to benefit from this drug.[1]

The analysis by Ezekowitz et al.[3] is important and adds additional nuance to the interpretation of the VICTORIA trial. The observation that the efficacy of vericiguat was attenuated in patients with higher NT-proBNP is statistically robust and different from previous HFrEF clinical trials. This result suggests that we need new therapies for very sick patients with HFrEF; perhaps quality of life is a better target in this group. However, this should not detract from the main results of the VICTORIA trial: in most patients seen in the real world (in whom NT-proBNP levels are <8,000 pg/ml), vericiguat offers a new approach to the treatment of HFrEF.