21st Century Cures, 20th Century Accounting: Why a $3M Gene Therapy Just Broke Insurance

A 6-month-old named KJ Muldoon just received the first personalized CRISPR gene therapy ever made — designed, manufactured, and administered for his exact mutation in six months. Nature named him to the Top 10 People Who Shaped Science of 2025.

The miracle is real. The financing model isn't.

Gene therapies run $2M to $3.5M each. Insurance contracts are annual. Gene therapy benefits last a lifetime. The employer who pays in year one rarely sees the savings — average commercial plan tenure is three years.

In this episode:

— Why the "gene therapy tsunami" narrative is overstated (EBRI's numbers tell a different story)

— The free-rider problem and why annual contracts can't price lifetime cures

— The concierge medicine paradox: we pay $3M to cure you, but won't pay $300/mo to keep you healthy

— Four financing models worth knowing for 2026: gene-therapy-specific stop-loss, outcomes-based agreements, risk-pooling platforms, and performance-based annuities

— How AI is compressing drug development timelines — and why that compounds the budget problem

We have 21st century cures and 20th century accounting. Eventually, one of those has to change.

Watch the full video: https://youtu.be/ap2XjNf3LN0

Read the Substack companion piece: https://open.substack.com/pub/danmccoymd/p/21st-century-cures-20th-century-accounting

Full sources and the deep dive: danmccoymd.substack.com

Transcript

SPEAKER_00 0:00

In February 2025, a six-month-old baby named KJ Muldoon became the first person to receive this kind of personalized CRISPR-based gene editing therapy. KG was born with CPS-1 deficiency, a metabolic disorder so rare it affects roughly one in a million people. Without treatment, half of infants with this condition don't survive their first year. The ammonia builds up in their blood because their bodies can't process protein, and it's devastating. A team at Children's Hospital of Philadelphia designed, manufactured, and administered a gene therapy specifically for KJ. And in six months, originally they'd estimated 18 months to do this. They did it in six. And here's the part that still gets me. Nature, the peer-reviewed journal that typically honors career scientists, named KJ to their top 10 people who shaped science of 2025, a baby. Because what happened to him represents something genuinely new in medicine. KJ is now walking, hitting developmental milestones, tolerating a normal diet. The drug that saved his life was made for one patient, tailored to his exact mutation, and not directly usable for anybody else without redesign. It was quite literally personalized medicine in its most extreme form. The miracle is real, the financing model for it, well, that's another story entirely. Let me back up because not everyone watching this has been following the gene therapy space that closely. Traditional medicine for genetic disease is management. You have a broken gene, so you take drugs forever to compensate for what that gene isn't doing. Hemophilia patients infuse clotting factors regularly, sickle cell patients manage pain crisis and organ damage. Cystic fibrosis patients do breathing treatments every day of their lives. Gene therapy is different. It goes to the source. You fix the gene or you replace it or you edit it, and the disease stops at the root. The price tags match the ambition. Hemgenics for hemophilia B costs $3.5 million. Zolgenzma for spinal muscular atrophy in infants runs about $2 million. Casgivi and Lipgenia, the new sickle cell therapies, are over $3 million each. As of 2025, the FDA had approved over 40 cell and gene therapies. The pipeline suggests dozens more by the early 2030s, and that's just the ones we can see coming. This isn't a distant future, this is now. Here's the part that most people in benefits get wrong. The problem with gene therapy isn't that it costs $3 million. Health plans deal with expensive treatments all the time. The problem is that gene therapy inverts everything the insurance model was built on. First, there's the timing mismatch. Insurance contracts are annual. Gene therapy benefits last a lifetime. You pay $3 million in year one, and the patient, if the therapy works, never needs a treatment again. But by year three, that patient might be on a different employer's plan. The employer who paid gets none of the downstream savings. This is what economists call the free rider problem. Average tenure on a commercial health plan today is about three years. The employer who pays for the cure rarely benefits from it long term. Second, there's what I think of as the lightning strike problem. For a large self-insured employer with 50,000 employees, a $3 million claim is significant, but manageable. For a small employer with 500 employees, that same claim might represent 10 or 15% of their entire annual health spending. It's the kind of variants that can blow up a benefits budget. Third, and this is the one that keeps the actuaries awake at night, you can't predict it. Traditional insurance pricing depends on actuarial tables. You know roughly how many heart attacks, how many cancer diagnoses, how many joint replacements to expect in a population of a given size in a given year. But gene therapy targets are rare diseases. A condition that affects one in 50,000 people, one in a hundred thousand, one in a million, which rare disease will be cured next? Will you happen to have an employee with it? There's really no table for that. There's a case that crystallizes this. Mosaic Life Care, a health system in Missouri, initially denied coverage for Zulkinsma for employee family twins with spinal muscular atrophy. The price tag around $4.2 million for both children. After significant public attention, read social media, they announced a multi-million dollar philanthropic fund to support such genetic treatments going forward. Now here's where I want to push back on the conventional narrative, because the conventional narrative may be scaring employers for the wrong reasons. The story you hear most often is that gene therapy is a coming tsunami that will overwhelm employer-sponsored insurance. MIT economist Jonathan Gruber, one of the architects of the ACA, used exactly that phrase, and it sounds, frankly, terrifying, but look at the actual numbers. The Employee Benefit Research Institute published data in February showing that cell and gene therapy users represent fewer than one-tenth of 1% of commercial enrollees. Their total share of spending, about half a percent. Current utilization is about 9 per 100,000 members. That's not nothing, but it's also not a tsunami, at least not yet. And here's another data point worth considering. Only a small minority of American workers are in truly exposed small self-insured firms, the ones where a single multi-million dollar claim could genuinely be catastrophic. Most workers are either in fully insured plans where the carrier bears the risk, or in large self-insured plans with enough scale to absorb the variants. Companies already cover rare, expensive events, organ transplants, premature infants in the NICU for months, burn victims. These are financially devastating too, and somehow the system handles them. So the tsunami narrative might be overstated or at least mistin'. But and this is important, there are things employers are not thinking about that they probably should be. First is durability. These therapies are too new to have long-term safety data. We don't know the 10-year outcomes, the 20-year outcomes. Some early CAR T patients had developed secondary cancers five years post-treatment. That doesn't mean the therapy was wrong, but it does mean the full cost picture isn't really clear yet. Second, and this is important, lasering. If you're relying on stop loss insurance, be aware that carriers are getting aggressive about excluding known high-risk conditions. If you have employees with known high risk or preexisting genetic conditions, your stop loss might not cover the gene therapy when it's needed. And third, commercial failure. Just because a therapy gets FDA approval doesn't mean it survives in the market. Bluebird Bio withdrew Centeglo from Europe entirely, not because it didn't work, but because pricing negotiations with national health systems failed. Pfizer discontinued BECFAS, a hemophilia gene therapy with virtually no commercial uptake. Commercial viability is not the same as clinical viability. And now we add artificial intelligence to the equation, which is where this gets both more exciting and more concerning. Traditional drug development takes 10 to 15 years from initial discovery to FDA approval. That's been the baseline for decades. AI, though, is compressing that timeline dramatically. Stanford Medicine developed something called CRISPR GPT, essentially a gene editing copilot that helps researchers design therapies, analyze data, troubleshoot failures. It's accelerating the work that used to require months of trial and error. And remember Baby KJ? That therapy was expected to take 18 months to develop, it took six. The team used AI-assisted design to compress what would have been a multi-year process into half a year. The pipeline reflects this acceleration. There are thousands of gene therapy clinical trials underway globally, with thousands of more therapies in development. And in February of this year, the FDA launched a new regulatory pathway they're calling the Plausible Mechanism Framework. I love that name, specifically designed for diseases so rare they may affect only a handful of people. Diseases for traditional randomized clinical trials just aren't feasible. The FDA estimates that 30 million Americans have some form of rare disease. Many of those diseases have the same underlying structure, a single gene gone wrong, which means, at least in principle, they're candidates for gene therapy. Here's the financing implication. More therapies targeting more conditions approved faster. The budget pressure really compounds. What a rare event today might be a quarterly occurrence in five years. I also want to linger on something that I think captures the absurdity of where we are. Health plans routinely refuse to cover concierge medicine. A patient pays a few hundred dollars a month for enhanced access to their primary care physician. Longer appointments, same-day availability, proactive care management. And the plan says not medically necessary, not covered. But those same plans are now finding ways to cover $3 million gene therapies. So think about what gene therapy actually is. Baby KJ's drug was made for one patient tailored to his exact mutation. It's not directly usable for anybody else without redesign. It's not just personalized medicine, it's individual medicine. It's hyper concierge. It's N of one. And somehow that's covered. We've built a system that will pay millions to cure you, but won't pay hundreds to keep you healthy. We'll cover the most individualized, most expensive intervention imaginable, but not the basic preventative relationship that might have caught the problem earlier. I don't know what that is, but to me, it's not rational. So if you're an employer or a benefits consultant or a CFO looking at your health plan, what are the actual options here? Well, stop loss insurance. It's the most obvious. For 2026, carriers like BCS Financial are quoting gene therapy-specific stop loss in low single-digit dollars per employee per month. That's not trivial, but it's manageable. The catch, watch for lasers, watch for exclusions, confirm that the therapy you might need is actually covered, not just gene therapies in general. Outcomes-based agreements are emerging at the federal level. CMS launched the Cell and Gene Therapy Access Model in early 2025. That's a multi-year program where manufacturers agree to rebates if therapies don't work as promised. 33 states plus DC are participating for the sickle cell therapies. It's a model where payment is tied to results, not just administration. Then there's risk pooling platforms. These are new and worth watching. A company called Aerodime launched in December with $20 million in Series A funding specifically to help employers pool gene therapy risk. The model, employers pay into a common fund, claims are paid from the pool, outcomes are tracked even when patients change employers, and manufacturers provide rebates if therapies fail. They're piloting with early customers in 2026. Then there's performance-based annuities. These are another option, spreading a $3 million payment over five to 10 years, with payments stopping if the therapy fails. It aligns the payment timeline with the benefit timeline. Some manufacturers are starting to offer this too. The uncomfortable truth, none of these options are perfect. Stop loss has exclusions, outcomes agreements require data infrastructure, risk pools are new and frankly improving, and annuities require manufacturer cooperation. But all of them are better than doing nothing and hoping the lightning doesn't strike. Here's the through line I keep coming back to. Gene therapy isn't a financing problem. It's a financing system problem. We built the entire infrastructure of American healthcare around chronic disease management, recurring revenue, annual contracts, predictable populations. The insurance model, the payment model, the actual model, all of it assumes you'll be treating a condition for years, not curing it in an afternoon. Gene therapy inverts all of that. One-time cost, lifetime benefit, a different employer pays than benefits from the cure, and a different insurer inherits a healthy patient they didn't invest in. The medicine has leapt forward, the accounting hasn't followed. The question isn't whether gene therapy is worth $3 million. For a child who would otherwise die, or a patient who would otherwise spend decades managing a brutal disease, the answer is obviously yes. The question is whether we can build a financing system that matches the medicine we now have. Because right now, we have 21st century cures and a 20th century accounting system. And eventually one of those has to change. If you found this useful, subscribe. The full research sources are on my Substack. Links in the description