Understanding Last Month’s Nobel-Worthy Breakthroughs
Edit Delete
Edit Delete
This is Porter & Co. Biotechnology – where Porter Stansberry presents a revamped portfolio to capitalize on the next biotech bull market. The details of the recommendations – and the specific name of the companies and buy prices – are for paid-up subscribers only. To learn more about subscribing to Porter & Co. Biotechnology, click here.
We’ve forgotten how hard it was to build modern healthcare.
In 1900, pneumonia, tuberculosis, and diarrheal diseases were the three leading causes of death in America. Staphylococcus aureus in the bloodstream killed 82% of the people who got it and most of them were young and healthy. Even as late as 1935, maternal mortality was still 600 per 10,000 live births.
Then, penicillin emerged – as an accident. A mold spore drifted onto a culture plate in Alexander Fleming’s London laboratory in 1928. It killed the bacteria growing there. And, unbelievably, it was safe enough to be dosed into humans. Out of all the world’s innovations and technologies, only synthetic nitrogen fertilizer has enabled more human life than the discovery of antibiotics.
By 1950, maternal mortality had fallen to 100 women per 10,000 live births. Today, it’s between 10 and 20.
In 1900, 40% of the deaths from pneumonia were children under five. Today, only the very old and the very sick die from pneumonia. And, overall, less than 3% of all deaths in America are attributable to the leading causes of death in 1900.
But these innovations were not easy.
Modern drug discovery began at Rutgers University in the late 1930s. Professor Selman Waksman (pictured right, below) set out to discover new antibiotics with a systematic approach. His method was witchcraft crude by today’s standards. He dug up soil samples, cultured whatever was growing in it, and then painstakingly tested every culture against live bacteria, until something worked. It was like a blind man searching for a needle in 10,000 haystacks.
But it was foolproof. A student streaked a soil microbe in a line down the middle of a plate. Then he streaked a disease bacterium across that line at a right angle. If the microbes in the soil were producing something effective at killing bacteria, it left a clear band through the bacterial streak. About 1,000 of Waksman’s 10,000 soil cultures produced a clear band. But that didn’t mean he discovered 1,000 new antibiotics. This was only the first hurdle.
Nearly every antibiotic he discovered came out of actinomycetes: the thread-like soil bacteria that produce the smell of fresh-turned dirt. One actinomycete in 10 makes the same compound, streptothricin. Waksman isolated it, tested it, and discovered it was toxic.
And then there was another problem: many of the compounds he discovered killed the wrong type of bacteria. As you probably know, bacteria come in two forms: gram-positive and gram-negative. Gram-positive bacteria have a thick cell wall that’s “naked.” Gram-negative bacteria have a thin cell wall that’s protected by a membrane.
Penicillin works by jamming the enzyme that builds the cell wall, which causes the cell to burst as it grows. Against bacteria with a thick bare wall (gram-positive) the drug is devastating. But against a gram-negative bacterium, it never reaches the wall at all: the outer membrane protects it. Plague, cholera, and typhoid are gram-negative – penicillin did nothing to them. What Waksman needed was a compound that could kill what penicillin couldn’t. He needed to find a compound that could destroy gram-negative bacteria.
And then there was a final problem. Some of the gram-negative killing compounds could not be manufactured. Producing an antibiotic in quantity means growing the microbe in thousands of gallons of liquid inside a steel tank. Plenty of microbes that secrete freely across a solid plate will not secrete anything once they are submerged.
Out of 10,000 cultures, roughly 100 survived all of these challenges. Then they all had to be isolated. That meant separating the one active molecule out of the soup the microbe grew in, purifying it, and then making enough of it to inject into an animal. That took the sorting from 100 down to only 10.
Then the animals they were testing these new antibiotics on kept dying. In 1940, actinomycin, the first antibiotic the program found, was the most powerful anyone had ever measured. But just one milligram of it killed a full-grown chicken. Streptothricin passed its early animal tests and then destroyed kidneys weeks after the dose. Clavacin and fumigacin failed the same way.
Waksman discovered that, relatively speaking, killing bacteria wasn’t that hard. Killing bacteria at a dose the patient can survive was vastly harder. But finally, one of Waksman’s researchers found one.
Albert Schatz, a 23-year-old graduate student working alone in the basement at Rutgers, isolated streptomycin on the afternoon of October 19, 1943. It cured tuberculosis, a deadly disease that had killed millions. The discovery won Waksman the Nobel Prize in 1952.
Diminishing Returns On Drug Research
Over the following decade, researchers in his laboratory worked through more than 100,000 more cultures. But instead of improving with time, his lab discovered fewer new compounds, even as it spent more money on discovery. Economists call the problem Eroom’s Law, which is Moore’s Law spelled backward. Just as Moore’s Law held that computer chip speed doubled every two years, Eroom’s Law holds that the number of new drugs approved per billion dollars of research spending has been cut in half roughly every nine years since 1950. In the early 1950s, an inflation-adjusted billion dollars of research bought 40 to 50 approved medicines. By 2018, that same billion bought 0.7.
Today, the Tufts Center For The Study Of Drug Development puts the all-in cost of one approved drug at $2.6 billion. Where does all the money go? It goes toward the overwhelming percentage of drugs that fail.
That’s why the drug business has always been organized around diversifying the risks as much as possible. But that’s finally changing.
In much the same way that technology converted the oil-and-gas industry from a discovery challenge to an engineering challenge, with enormous horizontal wells, technology is changing the essential challenge of medicine from discovery to manufacturing.
For a century, the oil business ran on the same odds as when Waksman’s antibiotics were discovered. Until the late 1970s, the industry considered a 10% wildcat success rate attractive: one successful well for every nine dry holes. The American Association of Petroleum Geologists tracked it by decade. New-field wildcat success ran 11.3% from 1946 to 1955, 10.8% from 1956 to 1965, and 11.2% from 1966 to 1975.
Thirty years, and no improvement at all.
Then, the industry discovered three-dimensional seismic surveying. This technology bounces sound waves off rock layers and reconstructs the subsurface inside a computer. It first emerged in the 1980s and became standard practice by the 1990s. Amoco measured the difference. Between 1990 and 1996, wells drilled with 3D seismic hit 49% of the time, against 14% for wells drilled without it. Over those same years, the share of Amoco’s prospects covered by 3D rose from 1% to 65%, and its cost of finding a barrel of oil fell dramatically – from $8 to less than $1.
And what came next made it virtually impossible to drill a dry hole.
Texas real estate developer George Mitchell spent 17 years and roughly a quarter of a billion dollars drilling the Barnett Shale in North Texas without turning a profit. In 1997, one of his engineers, Nick Steinsberger, replaced the expensive gel in the well’s fracturing fluid with water slickened by a trace of chemicals. The following year that well, S.H. Griffin No. 3, out-produced every well the company had ever drilled.
Mitchell did not discover where the gas was. Everybody already knew where the shale was: it covered thousands of square miles. What Mitchell proved was, with the right chemistry, gas could be manufactured out of that rock.
That’s what I mean when I say that drilling stopped being exploration and became engineering. Today the challenge isn’t finding oil and gas. Development wells in proven shale succeed 85% to 95% of the time. Today the challenge is purely engineering: how efficiently can you drill a given area of shale?
In 2010, the horizontal section of the average new Permian Basin well ran 3,870 feet. By 2025, it ran 10,867 feet and 15% of new wells now stretch past three miles. Occidental Petroleum (OXY) cut its drilling time in the Wolfcamp shale (within the Permian in West Texas) from 43 days per well in 2014 to 26 days in 2015, and the best operators now drill these wells in 13 days. Output per Permian rig went from 130 barrels a day in 2014 to 800 barrels a day by 2019.
Nobody found more oil. They engineered more oil.
Drug discovery is making that same turn right now. Artificial intelligence (“AI”) will play a growing role in testing new drug targets. Much like 3D seismic, AI will dramatically reduce the cost of discovery. The key to success in the drug business will be engineering – figuring out more efficient, more effective, and cheaper ways of hitting the best targets.
Our portfolio will be built around who owns the best engineering.
Chronicling The Remarkable Breakthroughs
As you know, our strategy in Porter & Co. Biotechnology is always to follow the science.
In the month of August, there were two Nobel Prize-worthy technological breakthroughs that I’d like to share with you.
Edit Delete
Porter further expands on the portfolio that he developed last month – with two of his newest recommendations up more than 30%. To see the remainder of this issue, new recommendations, and full access to the Porter & Co. Biotechnology portfolio, click here.
The most important breakthrough was on August 19. Merck (MRK) and Moderna (MRNA) announced a seminal breakthrough in the treatment of solid cancer tumors, the first successful Phase III trial of an individualized mRNA cancer vaccine. This new class of medicine holds the promise of stopping solid tumors from metastasizing following surgical removal.
Six days earlier, the New England Journal of Medicine published the details of a successful application of gene editing, in which a single infusion permanently rewrote the genetic code of human liver cells. The result dropped patient LDL levels by 62%, permanently.
The point made by highlighting these discoveries – which you’ll see me driving home in Biotechnology – is that we are living through a period of profound medical innovation. We’re getting closer to delivering cures at the cellular level by simply rewriting DNA – stopping disease before it can even begin.
And today, no one in the world understands the engineering behind drug discovery and optimization better than Eli Lilly (NYSE: LLY).
Yes, I know that’s a major pharmaceutical company. But we follow the science. Here’s what I mean.
The successful gene-editing program that I described above, the one that permanently reduced LDL in patients, is owned by Verve Therapeutics. Harvard cardiologist and human geneticist Skar Kathiresan founded Verve because his brother died of a heart attack at 42. Kathiresan founded the company to stop treating heart disease as a chronic condition by installing the natural protective mutations that nature already gives some people. In other words, rewrite the code of life inside your cells to give you nature’s best protection from heart disease. Obviously, the potential value of this technology is enormous, and it could have many applications beyond cardiology.
Verve’s initial funding came from Alphabet (GOOG) in May 2019. It went public in June 2021, during the massive post-COVID tech stock boom. On its first day, it gained 68%.
The company has three main programs. The breakthrough program that just published its results is VERVE-102, which is targeting PCSK9 for LDL cholesterol. It has Food And Drug Administration (“FDA”) Fast Track designation and is very promising.
The company’s next program is called VERVE-201. It’s targeting another gene in the liver (ANGPTL3) that makes a protein whose job is to put the brakes on fat clearance. It inhibits two enzymes (lipoprotein lipase and endothelial lipase) that pull triglycerides and cholesterol out of the bloodstream. Silence that gene, the brakes come off, and blood fats clear faster. It is unusual in that it moves all three lipid measures (LDL, triglycerides, and HDL).
Again, the company knew to target this particular gene because people who naturally have one broken copy of ANGPTL3 have 17% lower triglycerides, 12% lower LDL, and 34% lower odds of coronary artery disease. People with two broken copies of this gene have a recessive condition called familial combined hypolipidemia – very low triglycerides, LDL, and HDL – and they are healthy. Losing this gene entirely costs you nothing but gives you much better cardio health.
And then there’s VERVE-301, which targets lipoprotein(a) – Lp(a) – which I wrote about last month. This is the largest potential target in all of cardiology.
So what does this have to do with Eli Lilly? Lilly is the best in the world at finding the best targets and building the best therapeutics. In June 2023, Lilly signed a global research collaboration with Verve on its pre-clinical Lp(a) gene-editing program, paying $60 million upfront. After the results of VERVE-102 (targeting PCSK9) were first published last spring, Lilly bought Verve Therapeutics for $1 billion. That was $10.50 per share, which was less than the initial public offering (“IPO”) price and about 86% below the stock’s all-time closing high of $73.99 (set September 10, 2021).
Ruth Gimeno, Lilly’s group vice president for diabetes and metabolic research and development, described the deal as a chance to “shift the treatment paradigm for cardiovascular disease from chronic care to one-and-done treatment.”
This is more than just another modality. This is the difference between a treatment and a cure, and it could profoundly reshape the entire drug industry. I’ve told you that I believe gene silencing will be the best-performing modality over the next 20 years. But what if there aren’t any genes to silence because they’ve all been edited away?
Here are three points about that…
- Gene silencing will be used for a long time because changing your DNA permanently carries a lot of unknowns, and it may be irreversible. An edited genome has no off switch and no antidote. There will be a lot of people who are more comfortable taking a therapy once or twice a year – knowing that if something changes, they can always change therapies.
- Dosing has value too. A physician titrates silencing: more drug suppresses more protein, less drug suppresses less, and the dose can be adjusted as the patient changes. An edit delivers whatever suppression the editing efficiency achieved, forever. Where the safe range is narrow, or where medicine wants to reduce a protein rather than eliminate it, the adjustable drug wins.
- And then there’s the financial question. Insurers know how to pay for a twice-yearly injection, because they pay for thousands of chronic drugs that way. A one-time cure demands a single enormous payment with no established mechanism for spreading that cost across the decades of benefit it delivers. That is precisely why gene therapy has been such a commercial disappointment so far. BioMarin Pharmaceutical (BMRN) pulled a hemophilia gene therapy off the market entirely, despite the drug working because nobody would pay for it.
And regulators discriminate too. An irreversible intervention requires longer follow-up, mandatory patient registries, and restrictions confining early use to the sickest patients. For a 45-year-old managing a chronic condition, the reversible drug clears approval faster and reaches the prescription pad sooner.
Cholesterol treatment already works exactly this way. Statins come as daily pills. Antibodies against PCSK9 are injected twice monthly. Alnylam Pharmaceuticals (Nasdaq: ALNY) and Novartis’ (NYSE: NVS) Inclisiran (Leqvio) silences PCSK9 twice a year. And Verve’s gene-editing technology can rewrite it away once. Four technologies, one disease, and the market will determine who gets these drugs by how sick the patient is, how old, how reliably they take medication, and who is paying.
In other words, I don’t think gene editing is as big a risk to gene silencing as gene silencing is a risk to other modalities. I believe, going forward, more disease is going to be treated at the cellular level by silencing genes or editing them.
There’s no question that Lilly will be a leader in this field. And yes, I know its shares trade for more than $1,000 and the company’s enterprise value is more than $1 trillion. But that’s cheap compared to what the company will earn over the next decade.
Specifically, today you’re paying 25.5x what analysts expect Lilly to earn next year. Consensus estimates put earnings at $36.39 per share this year and $46.62 next year, on revenue climbing from $88.3 billion to $101.8 billion. And I believe revenue growth will far surpass these estimates.
The reason isn’t gene editing. Or even the company’s weight loss shots. The reason is a pill that’s going to transform human health.
Eli Lilly Is Medicine’s Nvidia
Novo Nordisk’s (NVO) Ozempic and Wegovy and Lilly’s Mounjaro and Zepbound are the current best-selling weight-loss shots. These therapies are peptides: a short chain of amino acids. Chemists build them the way a jeweler strings a necklace together. They anchor the first amino acid to a plastic bead, couple the second amino acid onto it, chemically unmask the new end so a third can attach, wash the bead in solvent, and repeat.
Tirzepatide, the molecule inside Mounjaro and Zepbound, runs 39 amino acids long. Every link demands its own coupling, unmasking, and washing cycle. Then there’s a purification process, which pushes dilute liquid through columns the size of oil drums for days. It’s a chemical engineering marvel. It is also expensive.
But orforglipron, Eli Lilly’s weight-loss pill, isn’t a peptide. It’s a small molecule.
Chemists make this drug the way they make aspirin or the cholesterol drug Lipitor: a handful of reactions, solid crystals you can filter and weigh at each stage, and purification by crystallization instead of chromatography. Thus, small molecule drugs are vastly cheaper to make than peptide chains.
The industry measures what a process costs to run using process mass intensity (“PMI”). Weigh everything you feed into the manufacturing line, the reagents, solvents, water, all of it, and divide by the weight of finished drug that comes out the other end. PMI captures manufacturing cost better than any other single number because solvent purchase and solvent disposal, not raw ingredients, dominate the bill. The American Chemical Society’s Green Chemistry Institute collected real production data from 14 pharmaceutical companies across 40 peptide manufacturing processes. Peptides consume 13,063 kilograms of input for every kilogram of drug delivered. Small molecules consume roughly 200.
Solvent accounts for roughly 44% of that mass. It’s expensive and frequently toxic. One solvent, dimethylformamide, generates 39% of all the waste. European regulators classified dimethylformamide as a Substance of Very High Concern and restricted it in December 2021 over reproductive toxicity. Piperidine, the reagent that unmasks each link on a peptide chain, sits on the Drug Enforcement Administration’s controlled-chemical list. Peptide plants have enormous compliance costs.
Over time, chemical engineers can drive costs out of small-molecule production. That’s why generic pills cost pennies. But peptide manufacturing shows no such expense-curve improvement between early development and commercial scale. It doesn’t get easier to make at scale.
The weight-loss market won’t be won by finding a better target. Incretins (GLP-1s) are incredibly safe and effective. The weight-loss market will be won by producing and marketing enormous quantities of these drugs at vastly lower costs. Much like a shale field, this market is not about discovery. It is an engineering battle. And Lilly has won.
I predict that Eli Lilly’s pill won’t just garner a lot of the market. I expect it will garner virtually all the market, better than 90% share over the next decade, for one simple reason: it’s vastly cheaper to produce and to distribute.
But wait a minute, Porter, wasn’t Novo Nordisk first to market with a GLP-1 pill?
Yes, Novo got there first. But it did it by manufacturing its peptide into a pill. That isn’t easy to do. It’s also very expensive because, in pill form, it requires vastly more of the GLP-1 peptide, semaglutide. And that’s an important point.
Bioavailability measures the fraction of a swallowed dose that survives the stomach and intestine and arrives in the bloodstream intact. Stomach acid and digestive enzymes exist specifically to dismantle proteins into their component amino acids, and a peptide is a small protein. Thus, your stomach treats oral semaglutide as lunch.
The FDA measured oral semaglutide’s bioavailability two separate ways during its review and got 0.471% and 0.795%. Swallow 1,000 units of oral semaglutide, and only five to eight of them reach your bloodstream, while the gut destroys or expels the other 992. In other words, Novo’s pill is an extremely inefficient delivery vehicle.
Novo tried to engineer a solution. Every tablet of Rybelsus, Novo’s original semaglutide pill, carries 300 milligrams of an absorption enhancer called salcaprozate sodium, which raises the pH immediately around the tablet so the peptide slips across the stomach lining before enzymes reach it. But there’s a problem: absorption still collapses under ordinary conditions. In the food-effect study Novo filed with the FDA, 14 of 26 subjects who swallowed the tablet after a meal showed no measurable semaglutide in their blood at all. None. The label therefore instructs patients to dose on an empty stomach, with no more than 120 milliliters of plain water, and to eat, drink, or take nothing else for at least 30 minutes.
Good luck in the real world.
Injected Wegovy delivers 2.4 milligrams once weekly, which is about 10.3 milligrams of semaglutide per patient per month. The approved Wegovy pill delivers 25 milligrams every day. Call it 750 milligrams a month. Switch a patient from the Wegovy injection to the Wegovy pill, and that patient consumes 73x more semaglutide than they did on the shot, whether any of it reaches their bloodstream or not. Every additional milligram must come out of the same peptide plants that took Novo five years and $10 billion a year to build. Now think about trying to serve a billion people that pill every day. That’s not going to scale.
Researchers priced generic semaglutide across formulations using 33 active ingredients. A year of the injection can be produced for well under $150, and most of that isn’t the drug at all: it’s the disposable injector pen. A year of the 14-milligram pill costs about $360 to produce. And Novo’s weight-loss pill isn’t 14 milligrams. It’s 25 milligrams. Our estimate is that Novo’s pill costs the company 4x to 5x as much as its injection costs to make.
Now, perhaps some people would pay more for a pill – but not many. And especially not when there’s a pill that’s cheaper and vastly more effective in real-world dosing.
Lilly’s orforglipron, approved April 1 and sold as Foundayo, is a small molecule that delivers 30% to 40% of a swallowed dose into the bloodstream. It needs no absorption enhancer and imposes no food restriction. It absorbs 40x to 100x more efficiently than oral semaglutide, and it belongs to a chemical family that costs one-50th as much to produce.
The cost of making enough semaglutide to enable pill production shows up in Novo’s financial statements. Novo’s gross profit margin fell from 84.9% in Q2 of 2024 to 78.2% in Q2 of 2026. Its capital spending has climbed from 6 billion kroner in 2021 to 60 billion kroner in 2025 (or US$935 million rising to US$9.4 billion), a 10-fold increase in four years. Novo also paid $16.5 billion in cash to buy the contract manufacturer Catalent (formerly CTLT) outright, purely to get its hands on more pen-filling capacity.
You can see how Novo is dealing with the threat from Lilly’s superior chemistry by watching Novo’s mergers-and-acquisitions activity. In May 2025, Novo Nordisk agreed to pay Septerna (SEPN) more than $200 million up front as part of a $2.2 billion deal for access to a platform that discovers small molecules acting on the same receptors Novo’s drugs target. Novo funds all the research. In February, it committed up to $2.1 billion to Vivtex, an MIT spinout, for a robotic screening system that tests thousands of formulations a day looking for ones that let a peptide survive the gut.
The battle isn’t about discovery anymore. The battle is engineering. When Lilly begins to compete on price, how will Novo recoup the cost of these massive investments? My bet: it won’t.
Lilly told investors a year ago that it expected to launch Foundayo worldwide without supply constraints. No peptide manufacturer can make that promise. Novo spent 2022 through 2024 rationing semaglutide, restricting starter doses and warning patients of shortages, because it could not make the peptide fast enough. Lilly, meanwhile, manufactured its pill before the FDA had even approved it. Its 2025 annual report disclosed $1.5 billion of pre-launch inventory sitting on the balance sheet as of December 31. Billions of tablets, ready to be sold.
Foundayo launched on April 1 at $149 a month for the starting dose through Lilly’s own pharmacy, rising to $349 per month at the maintenance dose. Patients with commercial insurance pay as little as $25 a month. Eligible Medicare patients pay $50 a month beginning July 1. Novo is competing with a list price near $1,349 for its semaglutide pill. Who is going to buy it?
Lilly launched Foundayo in Britain on August 23. Private providers there sell it at £100 to £120 a month ($136 to $163). Mounjaro, Lilly’s own injection, lists at £133 a month for the starting dose and £330 for the highest. So Lilly priced its pill below the cheapest injection it sells.
Lilly has filed orforglipron for approval in more than 40 countries. It filed with China’s National Medical Products Administration at the end of 2025, for both obesity and type 2 diabetes. In March, it committed another $3 billion over 10 years, adding tablet production in Beijing and expanding injection capacity at the plant it has run in Suzhou since 1996. That takes Lilly’s total investment in China to nearly $6 billion.
In total, Lilly has committed more than $50 billion to manufacturing since 2020. It’s building a $6.5 billion chemical plant in Houston to keep ramping Foundayo. It’s building a $3.5 billion injectables factory in Lehigh Valley, Pennsylvania, for its forthcoming new peptide, retatrutide. The plant won’t open until 2031 – so clearly Lilly believes there’s a vast amount of future demand for metabolic health products. And I believe they’re right.
Apple (AAPL) has shipped 2.64 billion iPhones as of March 31, 2026. Cumulative iPhone revenue since 2007 runs well past $2 trillion. The iPhone is, by far, the best-selling consumer product of all time. But I believe Lilly’s portfolio of metabolic health products will surpass Apple’s iPhone revenue. Do other analysts see this kind of potential? Morgan Stanley puts the combined diabetes and obesity market at $190 billion a year by 2035.
And with its next drug (retatrutide), Lilly will show the world what’s truly possible with this class of drugs.
Lilly’s current GLP-1 product, tirzepatide, activates two gut-hormone receptors. GLP-1 tells the pancreas to release insulin after a meal, slows how quickly the stomach empties, and signals the brain that eating can stop.
GIP is the second receptor, and it is the reason Lilly’s shot outperforms Novo’s. GIP stands for glucose-dependent insulinotropic polypeptide. Like GLP-1, it is a hormone your gut releases within minutes of food arriving, and like GLP-1, it tells the pancreas to make insulin. But GIP is different from GLP-1 in two critical ways. First, it acts directly on fat tissue, improving how efficiently fat cells absorb and store energy instead of leaving that energy to circulate in the blood. Second, and commercially more important, GIP appears to raise the threshold at which the brain triggers nausea. That means a drug hitting both receptors can be pushed to a higher dose before patients quit. Most of tirzepatide’s advantage over semaglutide comes from that headroom.
Hitting both receptors produced $9.9 billion of Mounjaro sales and $4.9 billion of Zepbound sales in Q2 of this year alone. The molecule’s composition-of-matter patent, the patent covering the substance itself and the hardest kind to work around, runs in the United States until January 5, 2036. But long before this patent expires, Lilly will have innovated much better drugs. And sold billions of pills.
The next innovation is retatrutide, which is a GLP-1 peptide that adds a third receptor. Glucagon is insulin’s mirror image. When your blood sugar falls, the pancreas releases glucagon, and the liver answers by breaking down stored glycogen and pushing glucose back into the blood. Adding a glucagon receptor to a weight-loss drug sounds backwards, since you are deliberately stimulating the hormone that raises blood sugar. What you buy is energy expenditure. Glucagon signaling raises the rate at which the liver burns fat and lifts resting metabolic rate.
So tirzepatide works the intake side of the equation: it reduces how much a patient eats and improves how well the body handles what does get eaten. Retatrutide adds the output side. The patient eats less and burns more. Lilly expects to file for approval in Q1 2027. Retatrutide gives Lilly an edge that many customers will pay up to receive.
Meanwhile, today Eli Lilly trades at 28x next year’s consensus earnings, which is only a small premium to the market.
Action to Take: Buy Eli Lilly (NYSE: LLY) below $1,500 per share
As with our other major pharma recommendation – Novartis (NYSE: NVS) – we recommend using an 80/20 allocation on this position, where you invest 80% of your capital into the equity and 20% of your capital into a long-dated call option with a strike price that is between 5% and 10% “out of the money.”
Moderna’s New Personalized Cancer Vaccine
In 1891, a New York surgeon named William Coley began injecting live streptococcal bacteria directly into tumors he could not cut out. He had noticed that patients who developed raging post-surgical infections sometimes watched their cancers shrink. He treated more than 1,000 patients and documented cures above 10% in cancers that otherwise killed everyone. Parke-Davis, later absorbed into Pfizer (PFE), sold a version of his toxin from 1899 until 1951.
The logic never changed: the immune system can kill cancer, and it will do so if you show it the right target.
Researchers identified the first usable molecular target on tumor cells in 1991. Dozens of companies then spent 20 years and billions of dollars building vaccines against targets like it. Nearly all of them failed for the same reason. Those targets were the patient’s own proteins, present at low levels in healthy tissue and merely overproduced by the tumor. Your immune system spends enormous effort learning not to attack your own proteins, because attacking them causes autoimmune disease. Training it to break that rule means overriding a safety system, and the safety system usually wins.
Subscribers who followed my work in the mid-2000s will remember Antigenics, which trades today as Agenus (AGEN). Antigenics manufactured a vaccine called Oncophage individually from each patient’s own surgically removed tumor. Genuine personalized medicine, 20 years too early. Its Phase III trial in kidney cancer randomized 818 patients. The hazard ratio for recurrence came in at 0.923, with a confidence interval running from 0.729 to 1.169 and a p-value of 0.506.
|
(Editor’s note: There are three critical numbers to know from any clinical trial. First, the hazard ratio compares how fast bad events happen in the treated group against how fast they happen in the control group. Read it like this: 1.0 means both groups suffered at identical rates and the drug did nothing at all. A 0.51 hazard ratio means the treated patients relapsed at roughly half the rate. Lower is better. A confidence interval brackets the range the true answer probably falls inside, given how few events the trial actually observed. If that range crosses 1.0, the drug might have done nothing, and the headline number could be noise. The p-value estimates the odds of seeing a result that large by pure chance if the drug were inert. Regulators generally want it below 0.05.)
|
The point estimate suggested 8% fewer recurrences. The confidence interval crossed 1.0 in both directions, meaning the vaccine might have helped by 27% or harmed by 17%. The p-value said a coin flip explains the whole result. Antigenics collapsed, renamed itself Agenus in January 2011, and built a business around the adjuvant Antigenics used to provoke an immune response (QS-21).
Dendreon (formerly DNDN) came next. Its personalized cancer vaccine, Provenge, was grown from each patient’s own immune cells. It won its Phase III trial, extending survival in prostate cancer by 4.1 months at a hazard ratio of 0.78. The FDA approved it in 2010. It showed no effect whatsoever on how fast the cancer progressed.
Then Dendreon priced it at $93,000 a course and… no one bought it. The company filed for bankruptcy in November 2014, and on May 6, 2015, the European Commission withdrew Provenge’s marketing authorization across the entire European Union. Not for safety. Not for efficacy. Dendreon requested the withdrawal itself, because it could not get European payers to reimburse for the drug.
In 2007, Genitope ran 287 patients through a lymphoma vaccine trial and delivered 19 months of progression-free survival versus 23 months on control. Worse than nothing. The stock halved the day the results came out. It delisted it nine months later.
After all these failures, most people assumed that personalized cancer vaccines would never work. But then three things changed.
First, the targets changed. Instead of vaccinating against the patient’s own normal proteins, laboratories now sequence the tumor’s DNA alongside the patient’s healthy DNA, subtract one from the other, and isolate the mutations that exist only inside the cancer. Those mutated proteins are genuinely foreign. These proteins are called neoantigens.
Second, checkpoint inhibitors arrived. Tumors survive by displaying molecular signals that switch off immune cells as they approach. Checkpoint inhibitors like Merck’s Keytruda block those signals. The old vaccines were pushing immune responses into a system the tumor could turn off at will, which is precisely why they generated beautiful laboratory data but didn’t save patients. Now that switch gets disabled.
Third, as we’ll be discussing for years, the engineering has gotten much better. mRNA (messenger RNA) solved the delivery problem: it carries the instructions directly into a cell, so the patient can make precisely the right antibodies. Encode dozens of neoantigens and the patient’s own body will manufacture all of them. Moderna’s product encodes up to 34 antigens, matched to that individual’s specific cancer. Production time fell from 50 or 60 days to about six weeks.
The individualized product cut recurrence at a hazard ratio of 0.51, with a confidence interval of 0.294 to 0.887 and a p-value of 0.0075.
In other words, the vaccine reduced cancer’s recurrence by about 50% – an outcome that was almost impossible by chance alone.
While the intellectual property around these vaccines and their delivery systems is, of course, complicated (like all IP in biotechnology) Moderna’s durable advantage rests on two assets that offer a moat beyond patents.
The first is the algorithm that picks which mutations go into the vaccine. In the largest published benchmark, ranking candidate neoantigens by how tightly they stick to the immune system’s display molecules identified genuinely useful targets less than 20% of the time. Filtering instead for whether the fragment actually appears on the cell surface, and whether a T-cell can recognize it, lifted accuracy above 70%. The clinical numbers are harsher still. Only 2% of computationally predicted neoantigens provoke a response in the laboratory. And in a pancreatic trial run by BioNTech SE (BNTX), only 11% of the selected targets produced a detectable immune response in patients.
Moderna’s vaccine holds 34 slots. At an 11% hit rate, four of them are doing something; at 70%, 24 are. Moderna’s algorithm is producing a better vaccine. And that advantage widens with every patient Moderna treats and learns from.
The second asset is Moderna’s factory. Every manufacturing batch treats exactly one patient. Moderna’s annual report states that the process scales out in parallel rather than up: you cannot build a bigger vessel and serve more people; you have to replicate the entire line. Moderna’s total dedicated capacity consists of one building: 140,000 square feet in Marlborough, Massachusetts. It cost around $400 million to build. Across the 10 solid tumors where this approach could work, American surgeons treat roughly 815,000 operable cases every year. Virtually every patient would want a personalized cancer vaccine, if having one meant a 50% better chance of extending cancer-free remission over a meaningful period (years). So, for now, it’s Moderna’s one building, against 815,000 American patients a year.
Currently, Moderna has no close competitor at all. BioNTech runs the only comparable program at scale and expects no pancreatic results before 2031. Genentech, the Roche Holding AG (SWX: ROG) subsidiary, abandoned its cancer vaccine partnership with the Norwegian developer Nykode Therapeutics.
Over the trailing 12 months through June 30, Moderna revenue was $2.23 billion and gross profit was $521 million, a gross margin of 23.4%. Research spending consumed $2.88 billion, 5.5x gross profit. The operating loss was $3.32 billion, or $7.99 a share. Moderna holds $6.91 billion of cash and investments with no reported debt, and it burns $1.24 billion a year, leaving 5.6 years of runway.
Analysts expect Moderna to lose $4.93 a share next year on $2.40 billion of revenue. Consensus is modeling almost no cancer-vaccine revenue at all through the end of 2027. Strip out the cash, and the market is valuing this business at $51 billion.
Moderna’s partner in this pivotal trial was Merck, which is seeking ways to extend the patent life of its best-selling checkpoint inhibitor, Keytruda. Merck books the sales and the two companies split the profits equally. Sell-side analysts put peak melanoma sales at $5.4 billion per year. Our base case assumes the vaccine reaches melanoma and lung cancer, which gets us to $11.4 billion in peak franchise revenue per year. Assuming 38% operating margin, that throws off $4.3 billion of profit per year. Moderna keeps half. As an estimate (we don’t know how long it will be before competitors emerge), we’ll use a 13-multiple on these earnings. That gets us to a future total value of $28 billion. But this market could easily grow much larger. Broad adoption across solid tumors at costs below $45,000 gets us to $92 billion.
Discounting those future values back to today yields a potential present value of $11 billion to $36 billion. On a Moderna per-share basis, that’s $27.56 or $90.42. That’s a very wide range of potential value for an investor to estimate today – welcome to biotech investing! And it’s even more difficult in this case because this entire approach is new, is extremely expensive, and doesn’t scale easily.
Manufacturing costs more than $100,000 per patient today. And this drug gets given to people whose cancer has already been cut out, who show no detectable disease anywhere in their bodies, and who feel completely fine. That’s a lot of money for an insurance company to approve… to someone who is currently healthy.
Provenge worked. It won its trial, and the FDA approved it. Then it was pulled out of Europe entirely, because payers would not shell out $93,000 for a drug given to patients who were not visibly, terminally ill. Moderna and Merck are proposing a six-figure therapy for patients who, while at great risk for further disease, are not currently sick. Who will pay that much?
One more thing. After Merck and Moderna announced that the trial met both of its endpoints, the market added $36 billion to Moderna’s value. That’s the present value of my most aggressive forecast for this product. I estimate that fair value on Moderna’s other business lines (an approved COVID vaccine, an approved RSV vaccine, a late-stage flu program) and its associated intellectual property is worth something around $15 per share. Add this to my estimates on the present value of the cancer vaccine, and you get something between $45 and $110. But the stock is trading above that – at $140 per share! Too expensive to recommend right now.
I suspect we will see a major pullback in Moderna’s shares at some point as the difficult economic reality of a $100,000 cancer vaccine plays out.
But what about Merck? Merck collects the other half of the vaccine profits; it sells the Keytruda that the vaccine must be given alongside, and it trades at only 16.1x what analysts expect it to earn in 2027. Keytruda generated 48.7% of Merck’s 2025 revenue. And Keytruda loses United States exclusivity in 2028. The stock has been trading very cheaply because of the risks Merck faces with the Keytruda patent cliff. I bought it personally earlier this year primarily because of its dividend and because it’s a very high-quality, low-volatility stock that fits with my “Honeycomb Strategy.” I didn’t buy it because of the cancer vaccine – in fact, before I took up this work in July, I didn’t know anything about Moderna’s cancer vaccine. But the shares jumped on that news and are now up 40% year-to-date. No, it’s not an expensive stock. But it’s not cheap anymore. I’m going to pass for now. If there’s a big drawdown, I’ll do a complete review of it.
Portfolio Review And Actions
Present value is our estimate of what a share is worth today on the evidence in hand. It’s not a price target, as we believe in letting our winners run. We don’t want to sell a great business when we have a good entry price. Buy below is the highest price at which we would put new money to work.
Eli Lilly enters at our 14% cap for integrated and tollbooth businesses, sized by inverse volatility alongside Novartis (NVS). Moderna would enter under our 10% cap for companies with marketed products, once it trades below $110.
Big Month Of News For uniQure
uniQure (Nasdaq: QURE) is a Hold.
Its AMT-130 is a gene therapy for Huntington’s disease. Huntington’s is caused by a single inherited defect that makes a mutant protein which slowly kills neurons in a part of the brain called the striatum. Patients lose motor control, then cognition, then everything, over 15 to 20 years. There is no treatment that changes the course of the disease.
AMT-130 is delivered by a neurosurgeon, through the skull, guided by MRI, directly into the striatum, in a single procedure. A viral vector carries instructions that make neurons produce a small RNA that silences the mutant gene.
Two things will happen this month. uniQure will present four-year data with a June 30 cutoff, covering 12 high-dose and 12 low-dose patients. The three-year data (previously published) is why the stock has tripled. On the primary measure – a composite score built from motor, cognitive, and functional tests – treated patients declined 0.38 points against 1.52 for an external control group, a 75% slowing, at p equal to 0.003. Functional capacity slowed 60% at p equal to 0.033. That is the most convincing result anyone has produced in Huntington’s disease.
September’s presentation is evidence about durability, and we suspect it will report data that’s stronger than the three-year data. We also believe this drug will receive approval this year, despite the very small number of patients and the lack of a real control group.
There’s a lot of history here. On November 3, 2025, FDA’s pre-submission feedback said it no longer agreed the Phase I/II data were adequate as primary evidence and the data were “currently unlikely” to support a filing. The shares fell from $67.49 to roughly $26. In June 2026, after a follow-up meeting, the agency reversed and agreed that the three-year analysis is acceptable as the primary basis for accelerated approval.
Here’s the real problem: uniQure has no placebo arm. These trials compared treated patients against a matched set drawn from Enroll-HD, an observational registry, selected by a statistical technique that pairs patients on baseline characteristics.
In February 2026, the FDA rejected a gene therapy for Hunter syndrome under similar circumstances: uncertainty in the eligibility criteria, comparability of the natural-history control, and the appropriateness of the biomarker. The agency noted that the control group was mismatched on age and baseline cognition.
uniQure’s filing rests on the same architecture: a small, treated cohort against a matched external control.
In addition, this division of the FDA has spent 2026 tightening the screws on gene therapy delivered to the brain. In January, it halted two REGENXBIO trials after a tumor appeared in an asymptomatic five-year-old dosed four years earlier, and the resected tumor showed viral genetic material integrated next to a cancer-causing gene called PLAG1. Causality was never established, and no other patient in either trial had a tumor. In August, the agency halted the same program again after spine imaging in five patients turned up small nodules, all judged benign. And in November 2025 the FDA put a boxed warning on Sarepta’s Duchenne gene therapy for fatal liver injury and pulled the indication for non-ambulatory patients.
The FDA is very worried about using viral vectors for gene therapy… and it’s easy to understand why.
But we think it’s likely that uniQure will get approval because there’s no other possible therapy and these results look strong. In 2023, the FDA granted accelerated approval to Qalsody for a form of ALS on the strength of a reduction in neurofilament light, a protein that leaks out of dying nerve cells, in 108 randomized patients. uniQure’s neurofilament data was 8.2% below baseline at three years on the high dose. We think there’s enough proof that the therapy is slowing the disease to warrant approval.
And so, we’ll hold.
Silence Therapeutics Closer To Divesiran Approval
Silence Therapeutics (Nasdaq: SLN) remains a Speculative Buy with a raised buy price.
Polycythemia vera is a blood cancer in which the bone marrow makes too many red cells, the blood thickens, and thick blood clots. The standard treatment has not fundamentally changed in a century: a nurse drains blood out of the patient, therapeutic phlebotomy. The body already has an off switch for making red blood cells. It’s a hormone called hepcidin that controls how much iron is available to the marrow. Iron is the raw material for red cells, so turning hepcidin up starves the marrow of what it needs and fewer red cells get made.
There are two ways to raise hepcidin. You can inject synthetic hepcidin, or you can help the patient produce more of his own. Protagonist Therapeutics (PTGX) and Takeda Pharmaceutical (TAK) took the first route with a peptide called rusfertide, and on August 28 the FDA approved it as Mimrylo.
Silence is taking the second approach. Divesiran is a small interfering RNA that silences a gene called TMPRSS6, whose protein is the brake on the body’s own hepcidin production. Release the brake, and the patient makes more of their own hepcidin.
Until August 28, the case for divesiran rested on a theory: that raising hepcidin controls polycythemia vera. The FDA has now approved a drug that does exactly that, on the strength of a randomized Phase III trial in 293 patients. The mechanism is no longer a hypothesis but a labeled indication. Mimrylo won approval on a single endpoint: the share of patients who did not become eligible for phlebotomy between weeks 20 and 32. Eligibility was defined precisely, as a confirmed hematocrit at or above 45% that also ran at least three points above the patient’s own baseline, or a hematocrit at or above 48%. Silence now knows the endpoint, its exact definition, and the trial size it will require to gain approval.
This new drug also begins to build the market for this kind of therapy. Somebody has to persuade hematologists to stop draining blood and start writing a prescription, and somebody has to get insurers to pay for it. Takeda has both the sales force and every incentive to do that work. Silence arrives after the category exists… but it will be second to market. What does that mean? Mimrylo is a once-weekly injection. Divesiran is one injection every twelve weeks. That’s fifty-two injections a year against four.
Regarding efficacy, Silence’s Phase II produced 88% clinical responders against 19% on placebo, at p below 0.0001, in 48 phlebotomy-dependent patients. Mimrylo produced 76.9% responders against 32.9% on placebo.
Silence sold $175 million of stock at $13.50 (right above our buy-up-to price) on the strength of the divesiran data.
We are raising our buy-below price from $13 to $16 and our present value estimate to $22.
Keeping A Watch On Arrowhead Pharmaceuticals
Arrowhead Pharmaceuticals (Nasdaq: ARWR) remains on the watchlist..
Plozasiran is a small interfering RNA (siRNA) that silences APOC3, a gene whose protein blocks the body from clearing fat out of the blood. Silence it, and triglycerides fall. It is injected under the skin once every three months. The FDA approved it on November 18, 2025, as REDEMPLO, for familial chylomicronemia syndrome, a genetic disorder affecting roughly 6,500 Americans. Arrowhead’s management has set the wholesale price at $45,000 per patient per year. In the June quarter, REDEMPLO produced about $2.4 million of revenue across more than 400 prescribers, roughly double the prior quarter.
Expanding the label is the company’s strategy. Roughly 2.3 million American adults carry triglycerides above 500 milligrams/deciliter (“mg/dL”), based on two decades of federal health survey data. Arrowhead will file for that population before the end of this year. And, on August 30, at the European Society of Cardiology congress in Munich, the company presented the trial data to support the filing. SHASTA-3 and SHASTA-4 randomized 757 adults to four quarterly doses or placebo. Triglycerides fell 79% and 81% at 12 months, both at p below 0.0001, and 85% in patients who started above 880 mg/dL.
Very high triglycerides cause acute pancreatitis, a sudden inflammation of the pancreas that hospitalizes patients and sometimes kills them. Pooled across both trials, pancreatitis events fell 78%, In patients who had already survived one attack, events fell 91%. Safety was clean: serious adverse events were 8.3% on drug against 10% on placebo, and 1.4% of patients discontinued against 0.8% on placebo.
Arrowhead isn’t the first to market here. On June 24, the FDA approved Ionis Pharma’s olezarsen as Tryngolza for the same population. Arrowhead’s competitive answer is dosing and depth: olezarsen is a monthly autoinjector, and plozasiran is quarterly. Olezarsen cites a 66% triglyceride reduction against Arrowhead’s 79% to 81%, and an 85% pancreatitis reduction against Arrowhead’s 78%.
Arrowhead carries roughly $1.22 billion of debt against $1.57 billion of cash and investments. The largest piece is a $400 million senior secured loan from Sixth Street. Interest runs at 15%, payable in kind rather than in cash, secured by a first lien on substantially all assets including the intellectual property itself. The company also has $700 million of zero-coupon convertible notes due January 2032 with a conversion price near $87.07.
Our $85 present value is built in four steps. We start with Morgan Stanley’s $3.2 billion peak plozasiran estimate, which that firm doubled from $1.7 billion in April, and value it at three times peak sales. We add $2 billion for the partnered royalty streams on olpasiran with Amgen and fazirsiran with Takeda, plus the obesity programs. Then we add the cash and subtract the debt. At $81.38, you are paying 3.5x Morgan Stanley’s peak-sales estimate for the whole enterprise, on a drug that is second to market with an interest bill that compounds at 15%.
We are keeping a watch on Arrowhead Pharmaceuticals until we see a price decline. We hope there’s a pullback in biotech this fall that allows us a good entry price.
Halozyme: Adding Another Tollbooth
Halozyme Therapeutics (Nasdaq: HALO) joins the recommended list with a Buy Below of $110, with a present value of $124.
Most antibody drugs have to be dripped into a vein, which means the patient sits in an infusion chair for hours and the hospital bills for a nurse, a chair, and a bag. Halozyme owns an enzyme called rHuPH20 that temporarily loosens the mesh of tissue under the skin, so the same drug can be pushed in with a syringe in minutes. It licenses that enzyme to drug companies under the name ENHANZE and collects a royalty.
Halozyme runs no trials, treats no patients, and files for no approvals. Roche Holding (SIX: ROG), Johnson & Johnson (JNJ), Takeda Pharmaceutical (TAK), Argenx (ARGX), and Bristol-Myers Squibb (BMY) do all of that. Halozyme cashes checks.
In June, royalty revenue rose 50% to $307.7 million, of which $152.2 million came from DARZALEX, the subcutaneous myeloma drug Johnson & Johnson sells.
Management raised full-year guidance for the second time this year, to $1.835 billion to $1.910 billion of revenue and non-GAAP earnings of $8.65 to $9.00 a share, up from $7.75 to $8.25. And it bought 4.8 million of its own shares in the quarter for $332.8 million, at an average price of $69.30, under a $1 billion authorization that runs through 2028.
One U.S. patent covers the composition of rHuPH20, which is licensed in every ENHANZE agreement. And it expires on September 23, 2027. That’s why Goldman Sachs rates the stock a Sell with a $58 target. But the royalty steps down, it does not stop. Halozyme’s filings reveal that when the rHuPH20 patent expires in a country, the royalty rate is reduced rather than terminated, and the company says the reduction never exceeds 50%. Apply the worst case to the exposed share of the royalty base, and we get a hit of roughly $1.85 a share against consensus 2027 earnings of $10.31. Halozyme guides DARZALEX royalties as secure through 2032.
And there is a second platform that is not in any existing license. MDASE is a separate family of modified hyaluronidases with 99 patents and applications, exclusive through 2032 outside the United States and 2034 inside it. Every new agreement Halozyme signs can be written on MDASE rather than ENHANZE. The company has signed five new agreements this year against a goal of three, including a global oncology collaboration with GSK (GSK) in May.
And there’s a potential kicker – a big one. Halozyme sued Merck in April 2025, claiming that subcutaneous Keytruda infringes 15 MDASE patents. A German court granted a preliminary injunction against subcutaneous Keytruda in December 2025.
If you’ve watched Veeva Systems (NYSE: VEEV) over the last month – up more than 40% – you’ll understand why I prefer to own biotech’s tollbooths.
Action to Take: Buy Halozyme Therapeutics (Nasdaq: HALO) up to $110 per share
BioNTech’s Failure
On August 28, BioNTech SE (BNTX) terminated the Phase II trial of autogene cevumeran, its individualized mRNA cancer vaccine partnered with Genentech, the Roche Holding subsidiary. The shares fell 8%. We do not own BioNTech, but what happened in this failed trial is important.
The independent Data Safety Monitoring Board found a numerical imbalance in overall survival between the two arms and concluded that continuing would not change the efficacy result. Reuters reported that analysts read that imbalance as more deaths in the vaccine group than in the control group.
BioNTech enrolled patients with high-risk stage II or stage III colorectal cancer who had already had surgery, then three months of chemotherapy, and who still had circulating tumor DNA in the blood afterward. Circulating tumor DNA is what it sounds like: fragments of cancer genome floating in the bloodstream. Finding it after surgery means the surgeon did not get everything.
But the vaccine was given as monotherapy, without a checkpoint inhibitor. Moderna’s product is given with Keytruda, Merck’s checkpoint inhibitor.
Bicycle: You Are Being Paid To Own The Pipeline
Bicycle Therapeutics (Nasdaq: BCYC) is our most uncomfortable Buy.
Bicycle holds $510.1 million of cash against $96.3 million of total liabilities and no debt at all, leaving $413.7 million of net cash, or $5.96 a share. So you are paying $4.14 for $5.96 of cash, which puts the enterprise value at negative $127 million. The market is paying you $1.83 a share to buy the stock.
Bicycle makes bicyclic peptides: short chains of amino acids chemically stapled into two rigid loops, which lets a very small molecule bind a target with the precision of an antibody and then clear the body in hours rather than weeks. Attach a chemotherapy payload, and you get a tumor-seeking missile that does not linger in healthy tissue. The lead drug, zelenectide pevedotin, targets a protein called Nectin-4 that sits on the surface of bladder-cancer cells.
There is already a Nectin-4 drug, and it is a good one. Padcev, from Pfizer (PFE) and Astellas, produced a confirmed response rate of 67.7% with pembrolizumab in first-line advanced bladder cancer and did more than $1.9 billion of sales in 2025. Zelenectide’s confirmed response rate in the equivalent setting was 58% to 62% in 26 patients.
Padcev’s label carries a boxed warning for severe and fatal skin reactions. Skin reactions occurred in 70% of patients on the combination, 17% at grade 3 or 4, with fatal bullous dermatitis in 0.2%. Peripheral nerve damage occurred in 67%, and 87% of those patients still had it at last evaluation. Zelenectide, at its selected dose, produced skin reactions in 17% and neuropathy in 33%, with no severe skin reactions at any grade.
On March 17, regulators told Bicycle that its pivotal trial design would not support accelerated approval in metastatic bladder cancer. The company converted the trial to a randomized Phase II, cut 30% of its staff, and set out to reduce operating expenses by half.
Its partners left. Genentech gave notice in July 2025, and Novartis followed later that year. Bayer terminated one target program in November 2025 and the entire agreement in July 2026, effective this month. Those three deals carried $45 million to $50 million of upfront money each and up to $1.7 billion of milestones apiece. Bicycle had collected $240.2 million of partner cash across its history and now has only one active collaboration, with Ionis.
In short, the outlook looks dark. But that’s why the stock is cheap. And we believe the underlying technology holds a lot of promise. The drug is not dead, and neither is the differentiation. Bicycle is still running the randomized Phase II and expects data in the second half of this year. In the Phase I first-line study, in patients too sick for cisplatin, zelenectide plus pembrolizumab produced a 59% response rate and median progression-free survival of 13.0 months. In late-line monotherapy, 38% responded with a median duration of response of 10.0 months. Those are not failing numbers.
Meanwhile, Bicycle is moving its resources to radioconjugates: the same targeting peptide carrying a radioactive isotope instead of a chemotherapy payload. It has locked up the hard part of that business, which is not the chemistry but the fuel. Bicycle signed a 15-year contract with the UK Nuclear Decommissioning Authority for access to up to 400 tonnes of reprocessed uranium, and it is working with the UK National Nuclear Laboratory to extract thorium-228 from it. It has exclusive lead-212 generators from SpectronRx and a supply agreement with Eckert & Ziegler. Novartis built a $2 billion drug in Pluvicto and an $816 million drug in Lutathera on this idea, and the constraint on the whole field is isotope supply. Bicycle’s first company-sponsored radioligand trial starts in 2027.
Bicycle Therapeutics is a Speculative Buy, below $5.00, with a present value of $7.50.
Good investing,
F. Porter Stansberry
Stevenson, Maryland
For important disclosures and disclaimers, please visit our Legal Disclosures page.
The post The Age Of Discovery appeared first on Porter & Co..
Login
