Metabolic and Electrical Heart Failure Signals

Evidence

Heart failure is not a single biological state, and two July studies tested measurements aimed at different parts of that complexity. One followed a blood-based metabolic score in people with heart failure with preserved ejection fraction (HFpEF). The other examined an electrical timing measure in people with reduced pumping function who received cardiac resynchronization therapy (CRT). Read together, they show why a risk signal must be interpreted in relation to the population, time point, and outcome it was designed to describe.

The metabolic study analyzed paired baseline and 12-month plasma samples from 46 TOPCAT trial participants with HFpEF. Nuclear magnetic resonance testing generated a Metabolic Vulnerability Index (MVX), a multimarker score derived from inflammation and malnutrition markers. Participants had a mean age of 72.5 years; 50% were female and 89% were White. The mean MVX change was 1.52 points, with a 95% confidence interval from -0.97 to 4.01 and a P value of 0.225. In this small cohort, the population mean did not change significantly over one year. That finding supports year-to-year stability in this selected sample; it does not show that every person's score was unchanged or that MVX predicted a clinical event in this analysis.

The electrical study retrospectively evaluated 415 CRT recipients at a Swedish tertiary center. Median age was 72.8 years, 77.3% were male, median baseline left ventricular ejection fraction was 27.5%, and 43.1% had ischemic heart failure. Investigators measured left ventricular activation time (LVAT) on a standard electrocardiogram from QRS onset to the maximum deflection in lead V6. The combined endpoint was heart-failure hospitalization or death from any cause, with median follow-up of 2.8 years and follow-up extending to 7.6 years.

Among 389 people with left bundle branch block (LBBB) or intraventricular conduction delay, each 10-millisecond longer pre-implant LVAT was associated with a lower endpoint rate in univariable analysis (hazard ratio 0.92, 95% confidence interval 0.86–0.99; P=0.026). The association weakened after adjustment for clinical factors (hazard ratio 0.93, 0.85–1.01; P=0.086). A comparison at the 80-millisecond median likewise was significant before adjustment but not after it. In the LBBB subgroup, Kaplan–Meier curves differed (P=0.046), while the adjusted association was not significant. Post-implant LVAT was not associated with the endpoint, and LVAT did not outperform conventional QRS duration in the adjusted models.

These results do not conflict. MVX was studied as a serial metabolic measure in HFpEF without an outcome test in this paper; LVAT was studied as a baseline electrical feature linked to outcomes after a specific device therapy. One asks whether a score moves over time. The other asks whether a pre-treatment substrate separates later event rates.

Analysis — Risk markers have different clocks

The cross-study pattern is that a useful heart-failure signal may be stable in one setting and predictive in another because the measurements describe different processes and are tied to different questions. This is analysis, not an established combined model. The stable average MVX could represent a relatively slow-moving background of metabolic vulnerability over 12 months, although the 46-person study cannot establish that interpretation. Baseline LVAT, by contrast, may reflect an electrical substrate that CRT is intended to correct. That offers a plausible explanation for the initially counterintuitive observation that longer activation time tracked with fewer events after CRT: a more visible delay might identify a substrate more amenable to resynchronization. However, the association lost conventional statistical significance after adjustment, so that mechanism remains a hypothesis rather than proof.

An emerging, unproven direction is to test whether slow-changing systemic measures and treatment-specific electrical measures contribute complementary information. These studies do not validate combining MVX and LVAT, and their populations are too different for direct pooling. Their contribution is narrower: they illustrate that heart-failure risk research needs to specify whether a marker captures background vulnerability, a modifiable treatment substrate, or both—and to validate each role against outcomes in the intended phenotype.

Limitations

The MVX evidence available for this synthesis was abstract-level. Its sample was only 46 people, was drawn from a clinical-trial population, and was predominantly White. The analysis tested mean change, not equivalence, individual trajectories, or an association with hospitalization or mortality. A non-significant P value therefore cannot prove biological constancy or prognostic value.

The LVAT study was retrospective and single-center, leaving selection, referral, attrition, and residual-confounding risks. Its cohort was predominantly male, subgroup sizes were uneven, and the main LVAT association did not remain significant after multivariable adjustment. Post-implant measurement was complicated by pacing-related electrical changes, and detailed lead-position information was unavailable. Most importantly, the studies involved different heart-failure phenotypes and endpoints. They cannot establish that either measurement improves care, that the measurements should be combined, or that the population-level findings predict any individual's outcome. Prospective, diverse cohorts with prespecified validation would be needed to test those possibilities.