Novartis reveals setbacks, not the numbers

Lowering a blood marker that genetics link to heart disease should, in theory, prevent heart attacks. In September, Novartis learned that theory and a large clinical trial can point in different directions.

On 4 September, Novartis announced that pelacarsen successfully lowered its target blood marker but did not demonstrate a statistically significant reduction in a composite cardiovascular outcome covering heart attacks, strokes, cardiovascular deaths and urgent coronary procedures. The trial, called Lp(a)HORIZON, enrolled 8,323 people (ClinicalTrials.gov). Days earlier, the company had paused eight clinical trials of a cell therapy called rapcabtagene autoleucel after three patients with autoimmune diseases died from the same inflammatory complication. In both cases, the data needed to interpret what happened remain undisclosed.

A genetically promising target, a clinically empty result

Lipoprotein(a), or Lp(a), is a cholesterol-carrying particle whose levels are mostly set by your DNA. Unlike ordinary "bad" cholesterol, it barely responds to diet, exercise or standard drugs. Genetic studies tracking large populations have found a dose-dependent link between high Lp(a) and coronary disease, making it one of the most genetically validated cardiovascular risk factors without an approved treatment.

Pelacarsen is designed to reduce Lp(a) production in the liver. In a Phase II trial of 286 people, it cut Lp(a) levels by up to 80% depending on dose. The reasoning was straightforward: if the particle drives disease, lowering it should reduce events.

Lp(a)HORIZON tested that reasoning at scale. It enrolled people who had already suffered a heart attack, stroke or peripheral artery disease and whose Lp(a) exceeded a high threshold. The trial was designed with a high chance of detecting the reduction its planners expected, around 20%, but a smaller benefit could still escape statistical significance.

The topline release said the primary endpoint was not met, then stopped. No estimate of the drug's effect, no measure of how uncertain that estimate was, no event counts, no breakdown for patients with the highest Lp(a) levels. Ionis, which developed the molecule, described the Lp(a) reductions as "substantially lower" than in earlier studies but gave no figure. As analysts observed, "did not meet" could mean anything from a near-miss to a flat line.

Why might lowering a genetically linked factor fail to help? Several explanations are plausible. Genetic risk reflects lifelong exposure beginning before disease develops; HORIZON treated people whose arteries were already damaged. A genetic analysis has also suggested the Lp(a) reduction needed for a meaningful coronary benefit may need to be very large. Rival drugs from Amgen (olpasiran) and Lilly (lepodisiran) achieve deeper suppression and are running their own outcome trials. This result weakens confidence in pelacarsen under this regimen and population, but whether the broader Lp(a) treatment idea survives depends on the size of the miss and what those later trials show.

Three deaths without a denominator

The cell-therapy pause raises a different problem. Rapcabtagene autoleucel, or rap-cel, is a CAR-T therapy: a patient's own immune cells are collected, genetically reprogrammed to target a specific protein on B cells, then infused back. In autoimmune disease, the goal is to wipe out the immune cells driving the attack and let the system rebuild.

Three patients died from IEC-HS, a recognised CAR-T complication in which activated immune cells trigger a severe, self-reinforcing inflammatory reaction. The critical missing number is how many patients received the therapy in total. Without it, no one outside Novartis can calculate the fatal-event rate or compare it meaningfully with other CAR-T products. Novartis has disclosed neither the total treated nor which diseases, doses or manufacturing lots were involved.

The acceptable level of risk for a cell therapy depends on disease severity, available alternatives and expected benefit, and differs substantially from the calculation in advanced cancer, where patients face few options.

What to watch

For pelacarsen, the full dataset, promised at an unspecified medical congress, will reveal whether the miss was narrow or decisive. For rap-cel, an analysis accounting for total patient exposure will determine whether the programme can restart. In both cases, the numbers needed to interpret the results exist. They just have not been shared.