Every morning I scan the cardiology literature so you don't have to. Here are three from the past week that are worth your time. 1. The first big Lp(a)-lowering drug trial came up empty. That doesn't mean Lp(a) doesn't matter. Lp(a) HORIZON tested pelacarsen in about 8,300 people who had already had a heart attack, stroke, or blocked leg artery. The drug lowered Lp(a) by about 78%, but heart attacks, strokes, and cardiovascular deaths were essentially the same as with placebo (12.3% vs 12.1%). Most participants were already very well treated, with LDL around 65 and controlled blood pressure, and they already had established disease. Experts think that may have left little room for extra benefit. Newer Lp(a) drugs are being tested in trials that report in 2028–2029, including one that enrolls people who haven't had an event yet. Takeaway: a high Lp(a) still raises your risk. Get it checked once, and if it's high, treat everything that can be treated (LDL/apoB, blood pressure, blood sugar) earlier and harder. Read more 2. A calcium score of zero may not tell the whole story Oxford researchers trained an AI to measure inflammation around the heart's arteries on a standard (non-contrast) coronary calcium scan. In more than 19,000 patients followed for about 7 years, this inflammation reading predicted cardiovascular death about as well as the same measurement on a full CT angiogram. Among people with a calcium score of 0 who later died of heart disease, about 45% had high inflammation readings. Takeaway: a zero score is reassuring, but it isn't a guarantee. Inflammation is a separate risk signal, and tools to read it from scans you may already have are coming. Read it 3. An AI can read an ordinary ECG and flag hidden sudden cardiac death risk Today, doctors mostly judge sudden cardiac death risk by the heart's pumping strength (ejection fraction). That approach misses most people who actually die suddenly. UC Berkeley researchers trained a deep-learning model on about 440,000 standard ECGs from Sweden and then tested it in patients from the US and Taipei. People the AI flagged as high risk had a 7% yearly rate of sudden cardiac death. The high-risk group picked out by ejection fraction had a 4.6% rate. Takeaway: the simple, inexpensive ECG may hold far more information than we've been using. This is still research, but AI reads of routine tests could soon help find people who need a closer look long before an emergency. Read more