The perils of standing still

A patient told me last week that his former cardiologist had told him, “Your numbers look fine. There’s nothing more to do.” He is 48. His LDL is 132 mg/dL. He’s never had a coronary calcium score, never had an Lp(a) measured, never had an ApoB. The conversation that ended his last visit could have happened in 2008. A lot has changed in those eighteen years. The interesting question is why so much of cardiology care has not.

I was a cardiology fellow at Mass General Hospital in the mid-2000s. The lipid culture I trained in was simpler than the one I practice in now. Some of what we believed then has held up. But a lot has not. That decade had its own ideas about preventing heart attacks, and most of them have either been quietly retired or transformed beyond recognition.

The vulnerable plaque era

The dominant story in coronary disease in the early-2000s was about the vulnerable plaque. The idea was elegant. Most heart attacks come not from the worst stenosis on an angiogram but from a particular kind of unstable lesion: a thin fibrous cap covering a large lipid core, inflammation in and around the plaque, the whole thing primed to rupture and clot. If we could find those plaques, the thinking went, we could stent them before they broke.

A lot of research effort poured into this. The largest of the imaging studies, PROSPECT (Stone et al, NEJM, 2011), used intravascular ultrasound to look for the predicted high-risk features and followed patients to see whether those features predicted events. The findings were sobering. Yes, you could find those plaques. No, they did not predict events as cleanly as anyone had hoped. The features overlapped massively with stable disease, and the events that did occur arose from lesions that had not been flagged as high-risk on the index scan.

What replaced it was a softer idea: the vulnerable patient. Same conceptual problem (who is going to have an event?), but framed as systemic risk rather than per-lesion anatomy. The practical implementation was a risk calculator and a statin. In other words, the field gave up on the original question, which patients are about to have an event and what can we do specifically for them?, and defaulted to the answer that statins worked broadly for risk-stratified populations.

That wasn’t itself a wrong answer. But it was a partial one. The original question never went away; the tools to ask it just were not good enough yet.

Calcium scores, orphaned

The coronary artery calcium score was already a recognizable test in the mid-2000s. A low-radiation CT, no contrast, that counts the calcified plaque in your coronary arteries. The data showing that the calcium score predicted events independently of standard risk factors had already accumulated.

It was treated as a curiosity. Useful in the abstract. A “tie-breaker” you could reach for when the calculator was equivocal. Insurance did not pay for it. Guidelines did not embrace it. The conversation between primary care and cardiology rarely included it. If a patient had read about it in a magazine and asked, you might shrug and say it was reasonable. If the patient did not ask, you usually did not bring it up.

This was a real loss. The information was there. The tool was there. The field was not yet structured to use it.

Back when statins were the whole toolkit

For a patient who was difficult, really high LDL, intolerant of one statin, on the maximum dose of another and still above goal, the menu was thin. Niacin was tried and largely abandoned (AIM-HIGH, HPS2-THRIVE). Fibrates had a narrow niche. Bile-acid sequestrants existed and were tolerated by approximately no one. Ezetimibe was around, but for years before IMPROVE-IT (Cannon, NEJM, 2015) it was widely dismissed as a “real-world placebo”, efficacious for the LDL number, unproven for outcomes.

If the patient could not take a statin, the truth was that we did not have much for them. If they were on a maximum dose and still above target, we mostly shrugged.

The field moved on

The transformation of preventive cardiology between roughly 2014 and 2024 is hard to overstate. The pieces did not arrive all at once. They arrived as a slow accumulation, and they are still arriving.

The LDL hypothesis settled. IMPROVE-IT showed in 2015 that adding ezetimibe to a statin reduced cardiovascular events in patients with established disease — a result that, more than anything else, established that the mechanism of LDL lowering did not matter as much as the fact of LDL lowering. That single result reframed the next decade of drug development. FOURIER (Sabatine, NEJM, 2017) and ODYSSEY OUTCOMES (Schwartz, NEJM, 2018) validated PCSK9 inhibitors using the same logic. Bempedoic acid joined the menu after CLEAR Outcomes (Nissen, NEJM, 2023). Inclisiran moved through trials and into clinical use. The treatment ladder grew from “statin or nothing” to “statin → ezetimibe → PCSK9 inhibitor or inclisiran → bempedoic acid,” with each rung supported by outcomes data and each fitting a different patient pattern.

Coronary calcium graduated. The 2018 ACC/AHA cholesterol guideline gave the calcium score an explicit role — not as a tie-breaker but as a decision modifier for borderline- and intermediate-risk patients. The 2026 dyslipidemia guideline further expanded the role. A coronary calcium score is no longer a curiosity. It is a routine tool that changes treatment thresholds.

The most recent twist is the AI-incidental calcium finding. Patients undergoing CT scans for unrelated reasons such as lung cancer screening, pulmonary embolism workups, or abdominal imaging have their coronary calcium visible on those scans. Software now reads it automatically. The 2026 guidelines recognize this incidental calcium score as actionable. For a meaningful slice of patients, a piece of preventive cardiology information now arrives unprompted, attached to a scan ordered for something else.

CCTA found the patients calcium missed. Coronary CT angiography with AI-quantified plaque (Cleerly, HeartFlow, and others) closes a gap calcium scoring leaves open. People with a calcium score of zero are not all the same; a subset has substantial non-calcified, low-attenuation, “soft” plaque that calcium scoring is structurally blind to. CCTA finds it. The picture that emerges, a young or middle-aged patient with inflammatory soft plaque that has not yet calcified, has a name in the new framework: a vulnerable patient phenotype with measurable substrate. The original 2000s question, who is going to have an event? now has a tool that can begin to answer it. The radiation is lower than the nuclear stress test we used to default to. The test is faster. The interpretation is better structured (see “Cleerly, CCTA, and calcium scores: which heart test, when?”). And with the PREVENT trial (Park, Lancet, 2024) perhaps the vulnerable plaque idea might even have been resurrected.

The cardiometabolic framework arrived. The 2023 AHA Presidential Advisory on cardiovascular-kidney-metabolic syndrome (the first-author, incidentally, was my Brigham and Women’s internal medicine residency classmate: Ndumele et al., Circulation 2023) gave language to a clinical pattern that cardiologists had been talking around for years: cardiac risk in patients with obesity, insulin resistance, and kidney disease is not the sum of three independent problems but a single intertwined process. The PREVENT calculator was built to estimate risk in that framework. It is replacing the Pooled Cohort Equations across the guideline landscape (see “The new cholesterol guidelines”).

The patient changed. Almost as importantly, the conversation a cardiologist now has with their patient is different. People arrive having ordered their own labs from Function Health or a DTC panel (see “A new pattern for cardiology visits”). They have read a Substack or Facebook page about Lp(a). They want their ApoB measured. They have a calcium score from a scan ordered three years ago for a lung nodule. They have spoken to an AI assistant before they showed up. They are not coming for a verdict; they are coming for an interpretation. The vocabulary they bring is prevention and longevity, not risk factor reduction.

The trap

Here is what is challenging: none of the above is hidden. The guidelines are public. The tools are widely available. The medications are prescribed every day at academic centers.

And yet a meaningful slice of cardiology visits in 2026 still end the way that 2008 visit ended: your numbers look fine, there is nothing else to do.

Standing still is not lying. It is something more subtle. It is having built a mental toolkit twenty years ago and never quite replacing it. The patient asking about an Lp(a) gets told it does not matter because the cardiologist’s mental model formed in an era when there was no Lp(a)-lowering drug on the horizon. The patient asking about a CAC score is told the calculator is enough because that is what it was developed to do. The patient who cannot tolerate rosuvastatin is sent away because the cardiologist last reviewed the non-statin menu before bempedoic acid was approved.

The result is not malpractice. It is a slow, quiet under-treatment of patients who would benefit from the modern toolkit. They sense it. The most striking change of the past few years is how often patients arrive at my visits already knowing they have been receiving care from an earlier era, and looking for someone who is paying attention to the field as it evolves.