Peeved by AI
A perspective on Utah’s much reported experiment allowing AI to renew prescription medications without direct human oversight (“Doctronic”). The take is overwhelmingly negative based on quality, safety, and legal issues, and a large part of the article reads as legal advice on how to sue to stop this and other future experiments. The argument would be a lot more convincing if it acknowledged two realities: First, the idealized standard of “careful drug monitoring and physician oversight” is somewhat illusory when in reality, plenty of prescribers are too busy or too low-performing (yes it happens!) to deliver to that standard. Second, the main point of doing this and other AI efforts is to make care cheaper for a population increasingly beleaguered by the economic weight of a system for which they pay either directly through co-pays, co-insurance, deductibles, or cash charges, or indirectly through health insurance premiums, taxes, or government borrowing. Utah’s Prescription-Renewal Pilot Program — Autonomous AI Managing Patient Care
An mRNA vaccine for influenza
Results from a phase 3 randomized control trial with 40,703 participants randomized to mRNA-1010 (Moderna) vs. standard dose trivalent influenza vaccines showed that the risk of getting the flu (confirmed by nasal swab PCR) was about 26% lower for individuals that received the mRNA vaccine compared to individuals that received its traditional counterpart (for the 2024-2025 northern hemisphere flu season). However, individuals who received the mRNA vaccines had more injection site pain and systemic symptoms (fatigue, myalgia, etc.). Serious adverse events were rare and about the same in both groups, including a case of cardiomyopathy in the mRNA arm and a case of pericarditis in the standard arm (though both occurred months after vaccination, so a causal link is in doubt). After much equivocation, the FDA is reviewing mRNA-1010 with a decision expected in August, and even though there was a favorable 9-0 Adcom vote, your guess for the outcome is as good as mine… In the final analysis, would I sign up for a day of feeling crummy in exchange for a relative risk reduction of getting the flu of 26%? Not sure. However, it is possible that in many flu seasons, the advantage will be much larger – that’s because with the long lead time needed for manufacturing of classical vaccines, selected antigens are frequently mismatched with the dominant circulating strains once the flu season begins; for mRNA vaccines, antigen selection can take place months later and lead to a better match. Efficacy and Safety of an mRNA Seasonal Influenza Vaccine in Adults
Attacking RAS in pancreatic cancer
Frequently mutated in cancer, but long considered undruggable, KRAS has received widespread attention in the last few years with a couple of drug approvals targeting the specific G12C mutation that is common in lung cancer (Lumakras from Amgen, and Krazati from Mirati). Although KRAS mutations are also common in pancreatic cancer, G12C mutations are rare and therefore alternative approaches are needed. This quarter, the Journal reports on two phase 1-2 studies with different agents each with their own approach. The first, setidegrasib (Astellas), is specifically designed to degrade KRAS with the G12D mutation, which is present in ~40% of pancreatic cancers. In 21 patients on second and third line therapy, treatment at the recommended phase 2 dose resulted in 5 partial responses, and 7 with stable disease status. In addition, rebiopsies of patients under treatment confirmed actual degradation of KRAS G12D in the tumors. The second agent, daraxonrasib (Revolution Medicines), approaches the problem by inhibiting all activated RAS, mutant or not. There, the results were nothing short of spectacular: in 83 patients in second and later lines of therapy, >90% achieved at least stable disease status or better. Yet at the same time, because daraxonrasib is not selective, patients developed side effects, chief among them a florid rash. Still, after decades of disappointing progress in pancreatic cancer, we are now looking at the potential for meaningful extension of survival in this horrible disease. Setidegrasib in Advanced Non–Small-Cell Lung Cancer and Pancreatic Cancer; Daraxonrasib in Previously Treated Advanced RAS-Mutated Pancreatic Cancer
Progress toward a functional cure for Hepatitis B
I have previously discussed efforts to attain a treatment-free functional cure to chronic hepatis B infection through a therapeutic course based on RNAi suppression. The success rate was low (~7%) but it was a poof of principle that this is actually feasible. Now, two phase 3 RCTs report on 24 weeks of treatment with bepirovirsen (GSK licensed from Ionis), an antisense oligonucleotide (ASO). In this case about 20% of patients with active disease attained a functional cure (vs. 0% for the placebo arm). It is definitely significant progress, and worth a shot for some patients, especially if predictive biomarkers can be identified (the only one mentioned is HbsAg with a level below 1,000 IU at baseline increasing success to 26%). Two comments: First, we may be following the path of Hepatitis C, which — after many years of trial and error starting from low cure rates — eventually led to the triumph we all know. Second, I have generally observed that RNAi approaches tend to be more successful than ASO – this seems like an exception worth noting. Phase 3 Results of Bepirovirsen Treatment for Chronic Hepatitis B Virus Infection
Sickle cell and thalassemia: when it rains it pours
A plethora of studies on gene therapies in sickle cell and beta-thalassemia with exa-cel (Casvegy, Vertex), reni-cel (Editas), and risto-cel (Beam) all had excellent results per reduction of vaso-occlusive crises or transfusion need. They all rely on ex-vivo modification of hematopoietic stem cells disrupting the pathway responsible for suppression of fetal hemoglobin after birth – unlike Bluebird’s lentiglobin which introduced a replacement hemoglobin gene – showing again that it’s a lot easier to use gene therapy to break things than to fix them. Meanwhile though, seeing an increasingly crowded landscape and a tough path to commercial success, Editas has thrown-in the towel on reni-cel to focus on other programs. CRISPR-Cas12a Gene Editing of HBG1 and HBG2 Promoters to Treat Sickle Cell Disease; CRISPR-Cas12a Gene Editing of HBG1 and HBG2 Promoters to Treat β-Thalassemia; Base Editing of HBG1 and HBG2 Promoters for Sickle Cell Disease; Exa-cel in Children with Transfusion-Dependent β-Thalassemia or Sickle Cell Disease
Atrial confusion
The management of permanent atrial fibrillation is a delicate equilibrium between the risk of embolic stroke and the risk of hemorrhage from anticoagulation. Since some (most?) stroke-causing clots seem to emerge from the left atrial appendage, devices have been developed that are delivered intravascularly to occlude the appendage and presumably reduce the risk of stroke without anticoagulation. Are these now marketed devices a good alternative to anticoagulation for patients who are deemed to be at increased risk for hemorrhage? Well, in the April and June issues of the Journal, we find two studies with contradictory findings. First, CLOSURE-AF, a German-funded study with 912 patients randomized to either closure or anticoagulation/best care showed worse overall performance for the closure group, and no benefit related to bleeding events. Second, CHAMPION-AF, a Boston Scientific-funded study with 3,000 patients randomized to either closure or anticoagulation showed non-inferiority for closure overall, and significant decrease in bleeding long-term, though that advantage went away when taking into account peri-procedure bleed. What is peculiar is that the German study enrolled patients with substantially higher risk of bleeding for which one would expect a greater relative benefit from closure, but things turned out the other way around.
So, what is one to think? These articles came with editorials, one of which is lapidary on closure – “Another Overused Method in Cardiology” – and the other more neutral but with the following delightful euphemism: “…shared decision making should be a central component in the management of atrial fibrillation. Owing to concerns that monetary incentives among operators may corrupt comprehensive discussions, organizations considering broadening indications may be wise to incorporate an independent, third-party interlocutor to ensure that adequate information is communicated to prospective patients.” I did put a made-up clinical scenario matching the trial inclusion criteria to ChatGPT Health, OpenEvidence, and Claude – they all quoted both studies and ended up recommending sticking to anticoagulation. Left Atrial Appendage Closure or Medical Therapy in Atrial Fibrillation; Left Atrial Appendage Closure — Another Overused Method in Cardiology?; Left Atrial Appendage Closure or Anticoagulation for Atrial Fibrillation; Left Atrial Appendage Closure — Should Recommendations Be Expanded?
Engineering cardiac muscle
A report from Germany on 20 patients with advanced heart failure who were implanted with bioengineered tissue to boost their cardiac function. The implants (Biological ventricular assist tissue – BioVAT) are derived from allogeneic iPSCs differentiated into cardiomyocytes and stromal cells which were then cultured into templated strips of organized myocardium with contractile activity. Then, 5-20 strips were tacked on the epicardium (the outside of the heart muscle) in strategic areas of poor contractility. It sort of worked – recipients felt better and the heart wall thickness and ejection fraction improved a bit, though objective performance measures such as the 6-minute walk test did not improve (but at least it did not get worse). Does that mean that we have a new modality for heart failure? I am unconvinced. However, that should not take away from this technological breakthrough. A sound tissue architecture is the one-time gift of fetal development: once you lose it through injury or other disruption, it’s very hard to get it back. But now, we know it is not impossible, thanks to the remarkable efforts of this team. Stem-Cell–Derived Biologic Ventricular Assist Tissue in Heart Failure
The New England Journal of Medicine is a premier weekly medical journal covering many topics of interest to the health sector. In this series we offer an opinionated perspective on selected highlights that might be of interest to our clients and others.