Why NAD+ Supplements May Miss the Problem in Diabetic Hearts

NAD+ supplements have an appealingly simple sales pitch: restore a cellular coenzyme, restore cellular energy. Diabetic heart disease may be less accommodating. A 2024 study identifies ACMSD, an enzyme that diverts tryptophan away from de novo NAD+ production, as a possible reason cardiac NAD+ falls—and why adding a precursor only partly repairs the damage in mice.

The finding matters because heart failure with preserved ejection fraction (HFpEF) often involves impaired relaxation and filling rather than weak pumping. Yet the study is a mechanism study, not a treatment trial. It includes diabetic mice and human heart-tissue validation, but no randomized trial has tested an ACMSD-targeted therapy or an NAD+ precursor in HFpEF.

The science identifies a plausible failure point in NAD+ metabolism. It does not establish that an NAD+ supplement treats HFpEF.

ACMSD can redirect the raw material for NAD+

NAD+ supports mitochondrial electron transport and enzymes involved in cellular stress responses, including sirtuins and PARPs. Cells can recycle NAD+ components through a salvage pathway, but they can also make NAD+ from tryptophan through the kynurenine pathway. In that route, quinolinic acid is converted by QPRT into nicotinic acid mononucleotide, a precursor of NAD+.

Zeng et al. in 2024 reported that diabetic conditions increased ACMSD expression in cardiac endothelial cells. ACMSD competes for an upstream substrate and diverts metabolic flux away from quinolinic-acid production. In diabetic mice, the authors linked higher ACMSD with lower myocardial NAD+, fewer capillaries, and more fibrosis. Those changes fit a pathway toward diastolic dysfunction, but the mouse experiments cannot prove that the same chain causes HFpEF in people.

The human component gives the result more weight than a mouse-only observation. Targeted mass spectrometry found higher ACMSD expression in human diabetic myocardial tissue, and higher ACMSD enzymatic activity was independently associated with diastolic-dysfunction risk. That is a clinically relevant association, not proof that ACMSD causes dysfunction in human hearts. The study also has not yet been independently replicated across larger tissue cohorts or across the many phenotypes grouped under HFpEF.

Why precursor loading produced only a partial rescue

The most useful experiment may be the one that disappoints supplement enthusiasts. The researchers treated diabetic mice with nicotinamide (NAM), a salvage-pathway NAD+ precursor that bypasses de novo synthesis. NAM partly restored myocardial capillary density and partly reduced fibrosis, but it did not fully normalize either outcome.

This result supports the idea that the pathway matters: changing NAD+-related metabolism improved parts of the phenotype. It also puts a ceiling on the conclusion. Partial rescue could reflect structural damage that had already accumulated, inadequate NAD+ delivery to heart tissue, effects of ACMSD beyond NAD+ production, or several processes at once. The experiment does not distinguish among those explanations.

It does, however, challenge the claim that more precursor reliably solves a cardiac NAD+ deficit. In a diabetic model with elevated ACMSD, salvage-pathway loading did not restore the whole phenotype. An ACMSD inhibitor would test the more direct therapeutic hypothesis, and preclinical inhibitors have been described, but none has reached clinical evaluation for this problem.

The human NAD+ trial answers a different question

Human evidence is thin and mismatched to the proposed use. Yu et al. conducted a single-center, placebo-controlled randomized trial in 180 people with ischemic cardiomyopathy and reduced ejection fraction, defined as left-ventricular ejection fraction of 45% or less. Participants received intravenous NAD+ at 10 mg per day for seven consecutive days or placebo.

The available results showed a nonsignificant trend toward lower NT-proBNP with NAD+. The trial did not assess diastolic-function parameters, and the available report does not permit confirmation of an effect on ejection fraction. NT-proBNP is a biomarker, not a direct measure of symptoms, hospitalization, or survival. More importantly, ischemic HFrEF and diabetic HFpEF differ in cardiac structure, physiology, and likely drivers of NAD+ depletion. A seven-day intravenous protocol also cannot validate a long-term oral precursor strategy.

That trial therefore does not refute the ACMSD hypothesis. It simply does not test it.

HFpEF is modifiable, but that does not validate this mechanism

Several interventions show that HFpEF-spectrum biology can change. In a randomized study of 46 middle-aged adults with left-ventricular hypertrophy and elevated cardiac biomarkers, Hieda et al. assigned 30 participants to one year of high-intensity exercise training and 16 to attention control. The training program reversed abnormal left-ventricular myocardial stiffness in this Stage B, high-risk population. These participants did not have established HFpEF, and the study did not test NAD+ biology, but it shows that stiffness is not necessarily fixed.

Drug evidence also depends heavily on phenotype. A 2025 meta-analysis of six randomized trials involving 7,282 people with obesity and HF with mildly reduced or preserved ejection fraction found that incretin-based therapies reduced cardiovascular death or worsening heart failure (hazard ratio 0.74; 95% confidence interval 0.63 to 0.88). That analysis supports treatment in obesity-associated disease; it does not show that incretins correct myocardial NAD+ depletion or help every HFpEF phenotype.

A separate 2025 meta-analysis found that GLP-1 receptor agonists and tirzepatide reduced heart-failure risk primarily in obese, rather than non-obese, people with cardiovascular or renal disease. The phenotype-specific pattern is a warning against treating HFpEF as one disease with one metabolic defect.

There is also active testing of non-NAD+ mechanisms. The GLADIATOR-HFpEF trial was a phase 2, double-blind, randomized, placebo-controlled crossover study of geranylgeranylacetone versus placebo, with each treatment period lasting three months and a six-week washout. Its dual primary endpoints were E/e′, a measure related to diastolic function, and reactive hyperemia index, a measure of microvascular function. The reported results improved both measures, according to the 2026 trial report. This is encouraging for myocardial compliance and endothelial function, but it is not evidence for ACMSD inhibition or NAD+ supplements.

Several routes can deplete NAD+

ACMSD is unlikely to be the only relevant defect. Preclinical work has also implicated PARP1 activation after oxidative DNA damage, which consumes NAD+; increased NNMT activity in impaired autophagy, which may accelerate clearance; and disrupted upstream signaling that regulates NAD+ biosynthesis. In another preclinical model, nicotinamide only partly rescued the effect of GPR91 loss. NR also failed in animals lacking HMGCS2, an enzyme required for cardiac ketogenesis. These findings make a one-size-fits-all precursor strategy biologically implausible, but they need direct confirmation in human HFpEF myocardium.

What to do with this now

For people considering NAD+ supplements, the practical conclusion is restraint. No randomized clinical trial has tested NAM, NR, NMN, or direct NAD+ in HFpEF, and no human study in this evidence base shows that oral precursors raise myocardial NAD+ concentrations. The diabetic-mouse experiment found partial, not complete, structural rescue with NAM.

The more compelling research question is whether diabetic HFpEF includes a measurable ACMSD-high subgroup, whether ACMSD inhibition restores de novo NAD+ synthesis in human cardiac tissue, and whether that change improves patient outcomes. Until those studies arrive, ACMSD is a credible mechanistic lead—not a reason to treat a supplement label like a cardiology prescription.