By admin 
The next time someone tells you that artificial intelligence, or any other new technology, is going to cure every disease, point them to the effort to fight heart attacks and strokes, which together remain the largest killer in the developed world despite decades of pharmaceutical intervention. The promise of modern medicine often feels like an approaching horizon—always visible, yet perpetually out of reach. For years, the narrative in biotechnology has been dominated by the idea that we are entering an era of "precision medicine," where genetic insights and machine learning algorithms will eliminate the guesswork of drug development. However, the harsh reality of clinical science often tells a different story, one where biological complexity defies even our most sophisticated models.
Because again and again, pharmaceutical researchers have learned the same hard lesson: technologies that were supposed to speed new medicines to patients have fallen short when confronted with the messy, unpredictable nature of human cardiovascular systems. Two major studies presented this past Saturday at the annual meeting of the American Heart Association (AHA) serve as a sobering reminder of this gap. For nearly a decade, a new understanding of genetics was supposed to increase the odds of success by identifying ironclad targets for intervention. Instead, the results from these high-stakes trials did anything but confirm that confidence, leaving researchers and investors questioning the "genetics-first" approach to drug discovery.
The first major blow came from Novartis and its highly anticipated drug, pelacarsen. Pelacarsen was designed to tackle one of the most stubborn risk factors in cardiovascular health: Lipoprotein(a), or Lp(a). Unlike the more commonly known LDL cholesterol (the "bad" cholesterol), Lp(a) levels are almost entirely determined by genetics and are largely unaffected by diet, exercise, or traditional statin therapy. For years, epidemiological data and Mendelian randomization studies suggested a clear, causal link: if you have high Lp(a), you have a significantly higher risk of heart attack, stroke, and aortic stenosis. The logic seemed foolproof. If we could use antisense technology—essentially a genetic "silencer"—to lower Lp(a) levels, we should, by extension, reduce the rate of catastrophic heart events.
The results of the pelacarsen trial, however, were a jarring departure from that logic. Despite the drug successfully and dramatically lowering the levels of Lp(a) in the blood of trial participants, the study showed no statistically significant effect on the actual rate of heart attacks and strokes. This creates a profound "biomarker paradox." We have become very good at moving the needles on blood tests, but we are finding that moving those needles doesn’t always translate into saving lives. It suggests that while Lp(a) is a marker of risk, the biological pathway might be more complex than a simple "remove the substance, remove the disease" equation. Perhaps by the time a patient reaches the age where they are enrolled in a Phase 3 trial, the damage caused by high Lp(a) is already irreversible, or perhaps other compensatory mechanisms in the body negate the benefits of lowering it.
The second disappointment came from Novo Nordisk, a company that has recently enjoyed unprecedented success with its obesity and diabetes franchise. Their study focused on a different frontier of heart disease: inflammation. For years, the "inflammation hypothesis" has suggested that even when cholesterol is controlled, a smoldering fire of chronic inflammation in the arteries continues to drive plaque rupture and heart attacks. Novo Nordisk’s drug, ziltivekimab, was designed to target a specific pro-inflammatory protein called Interleukin-6 (IL-6).
The trial, known as ZEUS, targeted patients with high levels of inflammation and chronic kidney disease. The hope was that by quenching IL-6, the drug would provide a new pillar of treatment alongside statins and blood pressure medications. However, the data presented at the AHA conference showed that ziltivekimab also failed to show a clear clinical benefit in reducing cardiovascular events. This failure is particularly stinging because it follows the mixed legacy of the CANTOS trial from years prior, which targeted a different part of the inflammation pathway. While the science of inflammation is sound, the pharmacological execution remains elusive. The immune system is a delicate web; pulling on one thread (like IL-6) may trigger a cascade of counter-reactions that we do not yet fully understand.
These failures highlight a broader crisis of confidence in the "tech-enabled" drug discovery boom. In recent years, companies like Isomorphic Labs and Anthropic have touted AI’s ability to revolutionize how we identify drug targets and predict how molecules will interact with the human body. The promise is that AI will move us past the "hit or miss" era of traditional chemistry. But heart disease represents a unique challenge for AI. Unlike certain cancers, which can often be traced to a single genetic mutation or a specific signaling pathway, cardiovascular disease is a systemic, multi-decade process involving blood pressure, lipid metabolism, inflammatory responses, and the mechanical stresses of the circulatory system.
The "valley of death" in drug development—the transition from successful laboratory and early-phase testing to large-scale Phase 3 clinical trials—is wider in cardiology than perhaps any other field. To prove a heart drug works, companies must enroll tens of thousands of patients and follow them for years. This makes the cost of failure astronomical. When a drug like pelacarsen or ziltivekimab fails, it isn’t just a loss of billions of dollars; it is a setback for a whole field of scientific inquiry.
Experts in the field are now calling for a "back to basics" approach, even as they embrace new tools. "We have to be humble about what we don’t know," says one senior cardiologist who attended the AHA sessions. "We have these incredible tools that allow us to sequence the genome and model proteins in three dimensions, but we are still operating within a biological system that has evolved over millions of years to be redundant and resilient. If you block one pathway, the body often finds another way to create the same outcome."
The implications for the pharmaceutical industry are significant. For years, the move has been toward "de-risking" drug development by using genetic validation. The theory was that if a drug target had genetic support, it was twice as likely to succeed in clinical trials. The pelacarsen and ziltivekimab results suggest that genetic validation is a necessary, but not sufficient, condition for success. It tells us that a protein is involved in a disease, but it doesn’t guarantee that modifying that protein in an adult human will alter the course of that disease.
This leaves other companies in a precarious position. Amgen, for instance, is developing its own Lp(a)-lowering drug, olpasiran. While their technology is slightly different (using small interfering RNA, or siRNA, instead of antisense), the underlying biological target is the same. The Novartis failure casts a long shadow over the entire Lp(a) class. If lowering the substance doesn’t work for one drug, the burden of proof for the next one becomes significantly higher.
Furthermore, these results underscore the difficulty of competing with the "standard of care." Today’s heart patients are already treated with highly effective, low-cost generic statins, beta-blockers, and ACE inhibitors. For a new, expensive drug to show a benefit on top of these existing therapies, it must be exceptionally powerful. We are reaching a point of diminishing returns where the "residual risk" that remains after standard treatment is incredibly difficult to move.
Despite the gloomy atmosphere at the AHA conference, there is a silver lining in the rigor of the scientific process. These trials, though failures in terms of bringing a product to market, are successes in terms of expanding human knowledge. They prevent the healthcare system from spending billions on treatments that do not work and force researchers to refine their theories. The path to a cure for heart disease is not a straight line powered by AI; it is a winding, often frustrating journey through the complexities of human biology.
As we look toward 2027 and beyond, the lesson from 2026 is clear: data is not wisdom, and a lower biomarker is not a cured patient. The revolution in drug discovery will likely happen, but it will be slower, more expensive, and more fraught with failure than the tech optimists would have us believe. The fight against heart disease continues, not with a "silver bullet" provided by an algorithm, but through the slow, methodical, and often heartbreaking work of large-scale clinical evidence. For the millions of patients at risk of heart attack and stroke, the wait for the next breakthrough continues, reminding us that in medicine, the hardest lessons are often the most important ones.
