CABG beats PCI in Medicare data, aficamten works in non-obstructive HCM, and pressure versus flow off-pump

1. CABG versus PCI for stable multivessel disease in the current era

Journal · Published: Annals of Thoracic Surgery, 26 August 2026 — genuinely new, 4 days old DeFazio, Wei, Chauhan, Hayanga, Mascio, Daggubati, Badhwar, Mehaffey

The framing is explicitly a challenge to guidelines: “Recent changes to society-specific guidelines for the management of stable coronary artery disease have challenged decades of data comparing CABG and PCI.”

US Centers for Medicare and Medicaid Services database, all beneficiaries aged ≥65 undergoing multivessel CABG (n=105,729) vs multivessel PCI (n=13,192) for stable CAD, 2018–2023. A predicted-risk model including frailty was fitted to simulate the STS mortality model (AUC 0.807), with doubly robust risk adjustment and time-to-event analyses in matched groups.

• Propensity matching produced well-balanced groups, 9,206 per arm; median age 74, 71.2% men, predicted mortality risk 1.2%
• Over six years, CABG was associated with superior longitudinal survival — HR 0.44 (95% CI 0.40–0.49)
• Conclusions also report lower hospital mortality and greater freedom from myocardial infarction and coronary reintervention

The deposited abstract truncates mid-sentence at the p-value for the survival hazard ratio, and the full text was not reachable, so I can’t give it. The confidence interval is well clear of unity.

Interpretation. Read the effect size sceptically before reading it enthusiastically. HR 0.44 is enormous — far larger than the survival differences in SYNTAX, FREEDOM or EXCEL — and effects that size in observational data usually signal residual confounding rather than a treatment effect that randomised trials somehow missed. The 8:1 imbalance in group sizes before matching is the tell: only 13,192 PCI patients met criteria against 105,729 CABG, so the matched PCI cohort is drawn from a pool selected for something. Propensity matching on measured covariates cannot fix the thing that made a heart team send a patient for PCI rather than surgery — anatomy, comorbidity, frailty as actually perceived rather than as coded.

That said, the paper’s target is real. The direction of travel in recent guidelines has been toward PCI in stable multivessel disease, and this is the surgical community’s registry-scale answer, from Badhwar’s group. It will be cited hard in the guideline argument, and it deserves to be read alongside its limitations rather than as a settled result. Worth knowing about before someone quotes HR 0.44 at you in a heart team meeting.

Read the paper · PMID 42413683

2. ACACIA-HCM — aficamten in non-obstructive hypertrophic cardiomyopathy

Journal · Published: Presented at ESC Congress 2026 and published simultaneously in NEJM, 28 August 2026 — genuinely new, 2 days old

Access note: numbers below are from conference reporting and the sponsor’s release; the NEJM paper itself was not reachable.

Double-blind phase III, 182 international sites, 516–517 adults with symptomatic non-obstructive HCM randomised 1:1 to aficamten or placebo for up to 72 weeks. Aficamten 5 mg daily, escalatable to 20 mg on echocardiographic findings. Dual primary endpoints at week 36.

• KCCQ Clinical Summary Score: +11.4 vs +8.4 (least-squares means), P=0.021
• Peak VO₂: +0.64 vs −0.03 mL/kg/min, P=0.003
• Secondary: NYHA class improvement, NT-proBNP reduction; KCCQ gains appeared around 8 weeks and were maintained through 72 weeks
• Safety — the part that matters: LVEF <50% in 10% vs 1%; LVEF <40% led to discontinuation in 3%; two serious heart failure events associated with LVEF reduction

Interpretation. The first successful cardiac myosin inhibitor trial in non-obstructive HCM, against the backdrop of mavacamten’s failed ODYSSEY-HCM in the same population — so this is a genuine first, and mechanistically interesting given there is no outflow gradient to relieve.

But look at the effect sizes against the harms. The KCCQ difference is 3.0 points between arms, below the ~5-point threshold usually taken as clinically meaningful, and peak VO₂ improves by 0.67 mL/kg/min — real, but modest. Against that, one in ten patients dropped their LVEF below 50%. An investigator’s own framing is that this frequency is unsurprising given a maximally-tolerated-dose strategy, which is candid, and also the point: the therapeutic window here looks narrow.

For perioperative practice, the relevant consequence is that a growing population of HCM patients will arrive on a negative inotrope with a documented incidence of systolic dysfunction. Worth knowing the drug, the echo surveillance it requires, and that a reduced EF in such a patient may be pharmacological rather than progressive disease.

Conference coverage

3. Pressure, flow and tissue oxygenation during off-pump coronary bypass

Journal · Published: Anesthesiology, 27 August 2026 — genuinely new, 3 days old Wang, Zhang, Guo, Liu, Han, Meng

Analysis of intraoperative 1-Hz signals during off-pump CABG. Synchronised mean arterial pressure, cardiac index, cerebral tissue oxygenation (SctO₂) and forearm tissue oxygenation (SftO₂) modelled with generalised additive models; two-dimensional smoothing over the MAP–cardiac index plane, with prespecified covariates and bootstrapped confidence intervals.

1,864 patients for left SctO₂, 1,863 right, 1,860 for SftO₂ — each with more than 20 million eligible observations.

• Cerebral oxygenation depended on both pressure and flow. The MAP–SctO₂ relationship showed a substantial reduction in slope across MAP 70–100 mmHg, mimicking cerebral pressure autoregulation.
• SctO₂ rose with cardiac index to about 4.2 l·min⁻¹·m⁻², then plateaued or slightly decreased.
• Forearm oxygenation tracked flow, not pressure — rising most steeply at cardiac index ≤2.8, flattening between 2.8 and 4.2, and continuing to rise above that.
• Patterns were more prominent in younger patients, women, and those with baseline BP <140/90.

The authors close with an unusually disciplined caution: these slope changes “may reflect physiologic regulation but should not be interpreted as evidence for treatment targets or practice-changing interventions.”

Interpretation. This is the best kind of physiology paper — enormous granular data, careful modelling, and authors who refuse to over-claim. The finding that cerebral oxygenation is sensitive to both pressure and flow while peripheral oxygenation tracks flow alone is a direct argument against managing the brain by blood pressure alone, which is what most intraoperative practice does.

It also lands neatly against the intraoperative hypotension meta-analysis (20 August), which found no outcome difference between higher and lower MAP targets and noted that achieved separation between arms was usually under 10 mmHg. Put the two together: those trials may have been titrating the wrong variable, in a range where the autoregulatory plateau makes pressure a poor proxy for perfusion anyway. That is a hypothesis, not a conclusion — and the authors would be the first to say so.

Read the paper · PMID 42295102