For years, Ozempic cost about $1,000 a month in the US. This year, Novo Nordisk dropped its own cash price to $349. New customers got an introductory rate of $199 for the first two months. Same drug. Same manufacturer. A price cut of more than half. About 1 in 8 adults in the US have taken a GLP-1 drug like this. More than half of them say affording it has been a real struggle. So the obvious question is the one worth sitting with. If the price could drop this much this fast, what was the original $1,000 actually made of. Not the manufacturing cost. Not the R&D that got the drug approved in the first place, that part is real and it's expensive. The distance between list price and what a drug can apparently be sold for once real competition shows up. Lilly's competing drug. Compounding pharmacies offering cheaper alternatives. Pressure from every direction at once. That gap is where a huge amount of the US drug pricing conversation actually lives, and most people never see it directly. They just see a number on a pharmacy receipt and assume that's what the drug costs. It isn't. It's what the market was willing to charge before it had to compete for the actual number. If you work anywhere near drug pricing, reimbursement, or market access, this is a live case study happening in real time. The price didn't move because the science changed. It moved because competition finally asked what the number was actually built on.

“Early detection saves lives” is true and also does a lot of hiding in cancer screening marketing. Multi-cancer early detection tests, the blood draws promising to catch dozens of cancer types at once, are one of the most hyped tools in oncology right now. The science behind them is real. These tests look for circulating tumor DNA, fragments that cancers shed into the bloodstream, and detecting that signal before symptoms appear is a genuinely hard technical achievement. What gets lost is the difference between detecting cancer earlier and actually improving outcomes. Finding a slow-growing cancer five years before it would have caused symptoms sounds like a win. Sometimes it is. Sometimes that cancer would never have caused harm in the person’s lifetime, and now they’re getting biopsies, scans, and treatment for something that didn’t need treating. This is called overdiagnosis, and it’s not a hypothetical, it shows up consistently in the screening literature for cancers like thyroid and low-grade prostate. The trials that would actually answer whether these new blood tests reduce cancer deaths are still running. Early sensitivity numbers get published and headlined. The mortality data, the number that actually matters, takes years longer to arrive because you have to follow people long enough to see whether they lived or died, not just whether the test found something. None of this is an argument against screening. It’s an argument for asking what a test was actually shown to do before assuming earlier always means better.

Every "longevity supplement" study you see cited on social media has the same flaw nobody mentions: the effect size. NAD+ boosters, resveratrol, rapamycin protocols. The papers are real. The p-values are real. The effect sizes are often so small they wouldn't survive a single confounding variable in a larger cohort. A study finding a "significant" 4% improvement in a biomarker is not the same as a study finding a treatment that changes your life. Marketing collapses that distinction on purpose. If a supplement company leads with "clinically proven," ask one question: proven to do what, and by how much. That single question filters out 90% of the longevity industry.

I'm teaming up with a group of medical science, biotech, and biomedical engineering people to launch a video podcast/show. The focus: real conversations about cutting-edge science across an unusually wide range of topics, from medicine and longevity to materials science and engineering. Think guests who can explain not just what's changing in their field, but why it actually matters. One episode might dig into the biology of aging, the next might break down the materials science behind armored vehicles. We want it to feel less like a press release and more like sitting in on a conversation between people who are genuinely curious about how things work. More details coming soon. If you're doing interesting work in a technical or scientific field (or know someone who is), I'd love to hear from you.

For 30 years, SMA drugs have only done half the job. This week, the FDA approved the drug built for the other half. Isembyld (apitegromab), from Scholar Rock, just became the first-ever FDA-approved therapy that targets muscle directly in spinal muscular atrophy, rather than the motor neurons that control it. Every SMA drug before this one worked the same way. Spinraza and Evrysdi, the two SMN2-targeting therapies that became standard of care, slow or stop the loss of motor neurons. That's real, disease-altering progress. But motor neurons aren't muscle. A patient can stop losing neurons and still be losing strength, because the muscle itself keeps atrophying underneath. Isembyld goes after that second half of the problem. It's a monoclonal antibody that blocks the activation of myostatin, a protein that acts as a brake on muscle growth. Take the brake off, and muscle has more room to grow and function, independent of what's happening to the neuron. The data is what makes this more than a mechanistic curiosity. In the Phase 3 SAPPHIRE trial, patients already on an SMN2-targeted drug who added Isembyld improved on the Hammersmith Functional Motor Scale by a statistically significant, clinically meaningful margin over one year. Patients on the SMN2 drug alone lost motor function over that same year. Roughly a third of treated patients hit a meaningful functional improvement threshold, versus about one in eight on placebo. Here's the fine print. This is an add-on therapy, approved for patients already receiving an SMN2-targeted drug, not a replacement for one. The strongest data comes from children ages 2 to 12; benefit in older patients is based on a smaller exploratory group. Myostatin inhibition has a rough track record elsewhere in medicine, with multiple prior attempts in other muscle-wasting diseases failing to translate. SMA may simply be the disease where this mechanism finally had the right biological conditions to work. What this really signals is a shift in how SMA gets treated: not one drug doing everything, but combinations that each address a distinct piece of the disease. Motor neuron protection plus muscle-directed therapy is now a real regimen, not a hypothesis. Expect the next wave of neuromuscular drug development to be built around stacking mechanisms like this one, rather than searching for a single drug that does it all. I write posts like this one for biotech founders and executives who need investors and partners to understand exactly why their science matters before the first call. If that's you, let's talk.

75 percent of previously untreated lung cancer patients responded to a pill approved this week. That number is the headline. The mutation it hit is the real story. On September 9th, the FDA expanded its approval of sevabertinib (Hyrnuo, from Bayer) to first-line, previously untreated patients with metastatic non-squamous non-small cell lung cancer carrying a HER2 activating mutation. It had already been approved for patients who'd failed prior treatment. This clears it for the front line. HER2 is a name most people only recognize from breast cancer. Herceptin made it famous there two decades ago, targeting a receptor that, when overactive, tells cells to keep growing. It turns out the same mutated receptor drives a meaningful subset of lung tumors too, specifically in the tyrosine kinase domain, the part of the protein that switches growth signaling on. Until recently, lung cancer patients carrying that mutation had no drug built for it. They got the same chemo regimen as everyone else, regardless of what was actually driving their tumor. The data is what makes this more than an incremental label update. In the trial supporting approval, 69 previously untreated patients took sevabertinib. The confirmed response rate was 75 percent. Of those who responded, 73 percent were still responding at six months, and 38 percent were still responding at a year. For a first-line lung cancer therapy, that is a strong signal, not a modest one. Two things to think about. This is an accelerated approval, meaning it's based on tumor shrinkage, not yet on survival data, which is still being collected. And the drug only reaches the patients it can help if someone tests for the mutation in the first place, using an FDA-authorized test, in a cancer where testing has historically been built around a different set of genes. The safety profile is real too: liver toxicity, interstitial lung disease, declining heart function, and eye toxicity are all being monitored in patients on the drug. The pattern underneath this is familiar by now. Herceptin didn't just treat one cancer. It proved a specific receptor mattered more than the organ it happened to be sitting in. Twenty years later, the same mutation is showing up in an organ nobody thought to test for, and a drug built for that mutation is already crossing the boundary chemo never could. If you're building or backing a company at this same intersection, where the biology crosses tissue lines faster than clinical practice catches up, this is the level of translation I write for founders and executives before investors and partners ever get on a call with them. Happy to talk.

For over a century, cardiology has operated on one unshakeable assumption: once heart muscle dies, it's gone forever. That assumption just broke. Researchers from the University of Sydney, the Baird Institute, and Royal Prince Alfred Hospital have shown, for the first time in humans, that heart muscle cells can re-enter cell division after a heart attack. Not just survive. Divide. Multiply. The heart tries to rebuild itself. This had only ever been documented in mice. A heart attack can wipe out up to a third of the heart's muscle cells in minutes. The tissue that survives doesn't regenerate; it scars, which is why so many heart attack survivors go on to develop heart failure. That's been the working model for decades: the heart heals like a knee, with a scar, not like a liver, with new tissue. The reason no one could confirm this in humans until now comes down to access. You cannot ethically biopsy a beating human heart to watch cells divide in real time. This team solved that by collecting "pre-mortem" tissue samples from consenting patients already undergoing bypass surgery, comparing damaged and healthy regions from the same living heart. In the damaged regions, they found clear signs of cardiomyocyte mitosis, cells actively dividing, and cytokinesis, cells completing the split into two. The heart's repair machinery was switching on. In the healthy regions, it largely wasn't. Here's the fine print. This regrowth is real, but it is nowhere near enough to replace what a heart attack destroys. Nobody is regenerating their way out of a heart attack on the strength of this alone. The researchers were careful to frame it as a repair signal worth amplifying, not a repair that's already working. What makes this matter is the target it hands drug developers. If the heart already has a switch that turns on division after injury, the next question isn't whether regeneration is biologically possible in humans. It's how to turn that switch up. The team has already identified candidate proteins tied to this process in mice that may translate to humans. Heart disease remains the leading cause of death worldwide. A finding that reopens the idea of a heart repairing itself, rather than just being managed after damage, is the kind of headline that's easy to scroll past and hard to overstate.

Pancreatic cancer survival just moved for the first time in decades, and the number is 13.2 months versus 6.7. That's the result behind Rasonque (daraxonrasib), the drug the FDA approved last month for metastatic pancreatic cancer. It's not a new chemo regimen or a better combination of old drugs. It's the first targeted therapy ever approved for this disease, given after standard treatment stops working, not instead of it. The mechanism is why it took this long. RAS mutations drive over 90 percent of pancreatic cancers, and RAS was considered undruggable for nearly 40 years. The protein's surface is smooth, with nowhere for a drug to grab onto in its inactive state. Daraxonrasib solves this by targeting RAS only when it flips into its active, cancer-driving shape, a chemistry problem the field spent four decades failing to crack. The trial data is what makes this a real result and not a hopeful one. In the randomized RASolute 302 trial, patients on daraxonrasib had a median overall survival of 13.2 months versus 6.7 months on chemotherapy, cutting the risk of death by 60 percent. Now the part getting less attention than the approval did. Expanded access programs closed the moment the drug was approved, on the assumption commercial insurance would catch up immediately. It hasn't. Patients who were finally told they'd get this drug are now stuck in the gap between approved and covered, in a disease where a few weeks of delay is not a small thing. Revolution Medicines is already testing the same RAS-targeting approach in lung and colorectal cancer, where the identical mutation shows up. Pancreatic cancer went first only because it had the fewest other options left to lose.

7) The real story isn't just this drug. It's the pipeline: Aging biomarkers built into a disease trial from day one, on a molecule that AI both targeted and designed. If that becomes standard, the next geroprotective drug won't be discovered. It'll be built to be one.

6) Reality check: 42 patients, Phase 2a, built for safety and lung function. The aging analysis was pre-planned, not a post-hoc fishing trip. 6 clocks agreeing matters because they weren't designed to agree. But proteomic aging clocks are still a young field.

5) The detail that should make scientists sit up: the dose with the best lung outcome wasn't the dose with the best aging signal. That's a sign the anti-aging effect isn't just downstream of healthier lungs. The drug may be hitting the aging process itself, semi-independently.

4) This isn't the usual "supplement lowered my biological age" story. Rentosertib's target, a gene called TNIK, was picked by an AI platform because it touches multiple hallmarks of aging, not just fibrosis. Then a separate AI system designed the molecule from scratch to hit it. New target. New molecule. Not an old drug repurposed.

3) In a new Nature Biotechnology paper, researchers ran 6 different biological aging clocks (built by different academic groups, using different methods) on blood from rentosertib's Phase 2 trial. All 6 agreed: patients on the drug looked biologically younger than placebo. In as little as 4 weeks.

2) Rentosertib treats idiopathic pulmonary fibrosis, a disease where lung tissue stiffens until breathing becomes a losing fight. That's what it was built for. Not why it's in the news.

1) A drug built to treat scarred lungs just did something it wasn't designed to do. It made people's blood look younger.

Three cancer breakthroughs in three weeks. A melanoma vaccine, a pancreatic cancer pill, a breast cancer drug. Every headline named the organ. Every headline missed the point. None of these drugs was really approved for an organ. They were approved for a mutation. The melanoma vaccine is built from the specific mutations in one patient's tumor. It doesn't know or care that the tumor started in skin. The pancreatic drug, daraxonrasib, hits RAS in its active state, a switch that's stuck on in over 90 percent of pancreatic cancers and in a large share of lung and colorectal cancers too. The breast cancer drug, camizestrant, was approved on the strength of an ESR1 mutation showing up in a blood test before scans changed. The trigger was a sequence, not a scan. For most of medical history, cancer was classified by geography. Where did it start? Breast, lung, colon. Treatment followed the map. That made sense when the only tools were surgery, radiation, and chemo, all of which are aimed at a location or at fast-dividing cells in general. Targeted drugs don't work that way. They work on a molecular lesion, and the lesion doesn't respect organ boundaries. A RAS mutation in the pancreas and a RAS mutation in the lung are, at the level the drug operates, close to the same problem. The regulatory system is slowly catching up. The FDA gave its first tumor-agnostic approval in 2017, for pembrolizumab in any cancer with mismatch repair deficiency. Larotrectinib followed for NTRK fusions. Those were treated as exceptions. Here's the fine print, and it's important. Tissue still matters, sometimes a lot. The same BRAF mutation that responds beautifully to a BRAF inhibitor in melanoma barely budges in colorectal cancer, because the gut tissue has a feedback loop that reroutes signaling. The mutation tells you what the tumor is running on. The organ tells you what else is in the room. Any founder pitching "one target, every cancer" is going to get asked about that. But the direction is clear. The question oncologists ask is shifting from "where is it" to "what is it running on," and the sequencing, the liquid biopsies, and the drugs are all lining up behind that question. Pancreatic, breast, and melanoma just happened to go first. The next headlines will name other organs. Read past them. Connect with me if you want a research and marketing professional to accurately translate the nuances of science to broad audiences.

The new breast cancer drug the FDA approved on Friday isn't the story. What it was approved for is. Camizestrant is an oral estrogen receptor degrader for HR-positive, HER2-negative metastatic breast cancer. Useful, but not revolutionary on its own. There are already drugs in this class. The revolutionary part is the trigger. This is the first cancer therapy the FDA has ever approved based on a resistance mutation showing up in a blood test before scans show the cancer is growing. Here's why that matters. Roughly 40 percent of patients on first-line hormone therapy eventually develop an ESR1 mutation. It's how the tumor learns to ignore the drug. Historically, you'd find out when the disease progressed on imaging, then switch. That's treating the cancer after it has already won a round. The SERENA-6 trial did something different. Patients had their blood monitored for circulating tumor DNA during treatment. The moment an ESR1 mutation appeared, and before anything changed on a scan, they either switched to camizestrant or stayed on their aromatase inhibitor. The switchers went about 16.8 months before progression versus 9.2 months for those who stayed. That's a 55 percent risk reduction. So the drug is the vehicle. The real approval is for a strategy: intercept resistance while it's still invisible. This is an accelerated approval, which means the FDA is betting on progression-free survival as a stand-in for the outcome that matters. Overall survival isn't in yet, and confirmatory data are required. There's a fair open question from oncologists about whether acting early truly beats switching at progression, or whether it just moves the clock. The trial wasn't built to answer that fully. But the precedent is set. Once regulators accept a liquid biopsy as grounds to change treatment, every drug targeting a known resistance mutation becomes a candidate for the same playbook. Lung, prostate, and colorectal all have these mutations. Breast cancer just happened to go first.

Your body produces about 25 million new cells every second, enough that the version of you reading this sentence has already replaced roughly a million cells since you started reading it.