Looking for zebras
“When you hear hoofbeats, think of horses, not zebras” is a classic dictum of medical practice, useful because it avoids many futile diagnostic expeditions. However, it does mean that some zebras will be missed; how many is hard to say. A study that makes a novel use of the UK Biobank (a database that consolidates genomes and clinical histories of about 500,000 individuals) aimed to see whether patients diagnosed with common diseases MS, IBD (UC+CD), or atopic dermatitis potentially had an underlying “rare disease” genetic origin – a zebra. Leveraging this massive dataset, they identified genes and variants that could cause the symptoms and performed a confirmatory analysis using data from subjects in therapeutic clinical trials. The upshot is that 1-3% of patients with those common diagnoses are genetic “zebras” and further, that standard therapies are very unlikely to work for them. Should a GI doc sequence every patient with IBD? Eventually this will be the case (and for any patient with any disease), but we are probably not quite there yet. Beyond the scientific contribution, it’s an interesting collaboration between biopharma and academia, and I wonder how Abbvie’s leadership is thinking about the upside to their R&D capabilities from such a project. Common Diseases in Clinical Cohorts — Not Always What They Seem
A clinical case conference with human and AI discussants
It is an old tradition of many academic medical centers: a case is presented to an expert clinician who then interprets it and proposes a differential and a leading diagnosis. Cases discussed at MGH are published in the NEJM, and they are often very challenging to solve, either because the pathology is very rare, or there is a confusing constellation of symptoms as in this installment involving an alcoholic patient with abdominal pain, fever, and pulmonary symptoms. A physician’s approach (who must be an awesome clinician) was compared with an AI model trained on MGH case records. After a long erudite discussion, the clinician reached the correct diagnosis of an intestinal perforation from a toothpick. The AI model came very close with “a perforated or penetrating duodenal process,” though its discussion of the case was not quite as complete as that of the human, and it failed to recommend an endoscopy which would have been the reasonable next step. Still, it is likely that in the universe of physicians, few would have outperformed the AI model on this case. As AI models learn and improve, it will be even fewer next year… Case 28-2025: A 36-Year-Old Man with Abdominal Pain, Fever, and Hypoxemia
Saved by the mattress (but it was not a fall)
A man with a history of heart disease presented to the ED short of breath and because “his commercial mattress — which included a ballistocardiographic monitoring feature — had alerted him that his average heart rate the night before during sleep (42 beats per minute) was lower than usual (78 beats per minute).” He received a pacemaker, resolving his symptoms. How to properly leverage and synthesize the information flowing in from a myriad of health monitoring devices will be a growing question as technologies continue to develop. Complete Heart Block Detected by Consumer-Facing Digital Health Technologies
Tackling polypharmacy
Some observational analyses have surfaced an association between the use of multiple antihypertensive agents and death in elderly frail patients, but as with any observational studies, even with risk adjustment, the potential for a confounder effect is strong. Now comes a French trial in which 1048 nursing home patients on antihypertensive therapy (with SBP < 130 mmHg) were randomized to usual care vs. step-down therapy. Obviously, the authors were hoping to find a mortality advantage in the step-down arm, and they seem to bemoan the fact that both arms had indistinguishable mortality and morbidity. I think they have it wrong.
Looking more closely at the population, the average age was 90, the average Mini Mental Status Exam score was 13.4, and the average number of unique medications was 6.7; we are talking about very old, significantly cognitively-impaired people who are being deluged with pills. Anything that allows backing off on the therapeutic burden is worthwhile in my view, or as the 13th and ultimate Law of The House of God says, “The delivery of good medical care is to do as much nothing as possible.” Reduction of Antihypertensive Treatment in Nursing Home Residents
PoC for prime editing
It’s been a few years since prime editing came on the scene, and we are now seeing it in action in actual patients. The test case chosen by Prime Medicine was chronic granulomatous disease (GCD), a rare genetic condition of neutrophil dysfunction that is typically X-linked (but in this case they targeted a rarer autosomal recessive form). Much like in the CAR-T approach, two patients had their hematopoietic stem cells modified in vitro with the precise fix to their defect (a small deletion), then reintroducing the edited stem cells to the patients. Both showed normalization of neutrophil activity and were able to discontinue their chronic therapeutic regimen. This trial was a clear confirmation of the potential of prime editing to address disease, even though this was done ex vivo, a much easier case than treatment in vivo. However, the end of the article is a bit sobering: “Because the current clinical study has been terminated by the sponsor, treatment of additional persons with p47-CGD will be contingent on the sponsor identifying alternative avenues to make the treatment accessible to patients.” And who can blame a company like Prime Medicine, fighting to extend its cash runway, for refocusing on conditions that have more than double digit patients? Still, this does highlight the plight of ultrarare diseases to attract the investment they need. Prime Editing for p47phox-Deficient Chronic Granulomatous Disease
Gene editing for the masses?
At the other extreme of gene therapy, CRISPR Therapeutics’ ANGPLT3 targeting agent works through loss-of-function (instead of implementing a gain-of-function fix), is used in vivo (instead of ex vivo), and potentially addresses CV risk in a population in the millions (instead of in the double digits). What makes ANGPLT3 an attractive target is that population studies have shown that naturally-occurring loss-of-function variants reduce CV risk, seemingly mediated by a better lipid profile. In a phase 1 ascending dosage study with an LNP-encapsulated CRISPR (encoded with mRNA) with a guide RNA targeting ANGPLT3, participants at the higher doses had lowered ANGPLT3 activity and markedly ameliorated lipid profiles. Among the 15 patients, one had a significant transient increase in liver enzymes but a rapid return to normal. Does this approach have long-term promise? Not clear given how much concern has arisen since the death of a patient being treated for aTTR with Intellia’s CRISPR agent. There is also the question of the cost of gene therapy, relative to what can be done with statins, or PCSK9 agents where biosimilars should be emerging in the next few years. Will there be a follow-on trial? The company stated in November that there would be, but I think the uncertainty remains significant. Phase 1 Trial of CRISPR-Cas9 Gene Editing Targeting ANGPTL3
J&J in MM
In the last decade, J&J has assumed a preeminent position in the multiple myeloma (MM) treatment landscape, first with Darzalex (daratumumab targeting CD38) and its direct descendant Darzalex Faspro, then with Carvykti (ciltacabtagene autoleucel), a CAR-T therapy targeting BCMA. As with all autologous CAR-Ts, the treatment process for Carvykti is cumbersome and difficult to scale. An alternative is a bispecific T-cell engager (BiTE) such as teclistamab (Tecvayli, J&J) targeting BCMA+CD3, yet BiTEs are typically not quite as effective as the equivalent CAR-Ts. Normally, both therapies are reserved for late stages of the disease as a last-ditch effort, but in this case, J&J chose to test the use of Darzalex + Tecvayli in a relapsed patient population with relatively early disease progression (time since diagnosis 3.8 years). Results were impressive. Three years out, 35% of patients on standard of care had died, whereas only 18% had died in the Darzalex + Tecvayli arm. This is likely to become the standard of care, and it is potentially a huge blockbuster for J&J given that patients are going to be on treatment for many years. Teclistamab plus Daratumumab in Relapsed or Refractory Multiple Myeloma
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.