Megan Skiendel: Hey, good to see your face — did your week survive?
Zara Reyes: Barely. I had a whole thing with a prescription pickup Friday that actually — lowkey is the reason I wanted to do this topic today.
Megan Skiendel: Oh, say more.
Zara Reyes: So I'm at the pharmacy, totally routine, and the pharmacist just — hands me the bag, smiles, done. And I'm thinking, nobody in this interaction knows anything about whether my body actually processes this the way the standard dose assumes.
Megan Skiendel: And honestly — that invisibility, that exact moment, is what a team at the University of Pittsburgh just measured at scale. They pulled 98,590 people from the NIH All of Us Research Program, looked at whether the genetic variants that quietly change how drugs work in your body are rare edge cases or something else. The study ran in Science Translational Medicine, November 11th, 2026. The answer: one hundred percent of participants carried at least one.
Zara Reyes: Wait — every single one?
Megan Skiendel: Every single one. And look — that's the plain version of the finding, which is almost too clean. What pharmacogenomics is actually saying is that your genes can quietly change how a medication moves through your body, how fast your liver breaks it down, whether it even reaches the target — and almost nobody is checking for that at the pharmacy counter.
Zara Reyes: No but that framing — 'nobody is checking' — that's the part that gets me, because it's not like the science is new. So what does it mean that we now have 98,590 data points saying this is universal?
Megan Skiendel: And that's exactly where you have to slow down before the headline runs off without the data. Because 100% sounds like everyone's in crisis — but the median participant carried nine variants. Nine. Across 16 pharmacogenes. That's not one weird edge-case gene. That's layered.
Zara Reyes: Nine. And across 16 genes that govern — like, not one pathway, absorption, metabolism, transport, the actual drug target.
Megan Skiendel: Right — and 99.49% of participants had what researchers call a pharmacogenomic phenotype. A predicted functional category — poor metabolizer, rapid metabolizer — that CPIC or the FDA already has a specific prescribing recommendation for. The Clinical Pharmacogenetics Implementation Consortium exists precisely to convert that data into clinical action. The guidance is written. It's sitting there.
Zara Reyes: Wait, so the gap isn't even — we don't know what to do. The gap is we're not doing the thing we already know to do.
Megan Skiendel: Honestly — yes. And then Philip Empey, the senior author from University of Pittsburgh and UPMC, was explicit about this: universal carriage does not mean every person needs a test tomorrow or a changed prescription. That 100% is real. What it means clinically is a different question.
Zara Reyes: No but that's the nuance that gets swallowed by the headline every time.
Megan Skiendel: Every time. And then — look, here's the number I can't stop thinking about. Of participants who had both genetic data and clinical records available, 24.38% had already been prescribed at least one medication affected by their flagged variants. The clinic never caught it.
Megan Skiendel: One in four. Retrospectively. Meaning this wasn't caught in real time — researchers cross-referenced the genetic data with pharmacy records after the fact. The clinical system never did that math.
Zara Reyes: And that math was always possible — that's what makes it uncomfortable. Like, the FDA has genetic interaction information on more than 300 medications. Those labels exist. They're not hidden.
Megan Skiendel: Three hundred. And clinicians largely don't know it's there.
Zara Reyes: So who's actually doing this? Because it's not nobody.
Megan Skiendel: Psychiatry, mostly — specifically depression, specifically after a treatment failure. Someone tries two antidepressants, neither works, and suddenly a clinician considers testing. That's reactive, not preemptive. And it's geographically patchy. A September 2026 systematic review looked specifically at PGx testing for antidepressants in primary care and found — uptake is still limited. Guidelines exist. Behavior hasn't followed. I mean, an October 2026 paper from the Pharmacogenomics Global Research Network Implementation Working Group was still defining what basic infrastructure for this in primary care even looks like. That's not a science problem. That's an institution that never built the room.
Zara Reyes: Wait — October 2026, still figuring out what the room looks like?
Megan Skiendel: Still defining core functions. And look, I've watched this exact structure before with electronic health records — the science was clear, the guidelines were written, and nothing moved in practice for a decade because nobody agreed who paid for the workflow change. This rhymes hard. But — and this is the part that actually surprises me — the proof that it can work is already sitting there. Colon cancer chemotherapy. Preemptive genetic testing before prescribing, because up to 8% of patients carry variants linked to serious toxicity. The field figured it out. It just never generalized beyond the cases where the stakes felt too visible to ignore.
Zara Reyes: So the infrastructure isn't missing — it's just rationed by how dramatic the downside looks.
Megan Skiendel: Exactly that. And a prevalence study — even one this clean, even 100% in 98,590 people — historically is not what moves a system like this. Not on its own.
Zara Reyes: And honestly, the equity layer is about to make all of this messier — those 17,000 novel variants in underrepresented ancestry groups carry their own problem that is not just about implementation.
Megan Skiendel: And messier in a specific way — because those 17,000 novel variants aren't evenly distributed. They're concentrated in participants of Americas and African ancestry. Groups that have been underrepresented in PGx research for decades. So the All of Us cohort finally pulls them in, and what do we find? Their genetic data is new to us. Novel. Which sounds like progress.
Zara Reyes: But novel means no CPIC guideline. No FDA label language. No clinical action attached.
Megan Skiendel: Exactly. The variant exists in the database. The prescribing guidance does not. So a clinician who actually ran the test would see — what? A flag with no instruction.
Zara Reyes: And lowkey that's the cruelest version of this. Because the sparse population-frequency data for those ancestry groups is itself one of the documented reasons PGx implementation has been harder to scale. The data was missing, so the guidelines never got written, so now that the data exists — the guidelines still aren't there. It's a loop.
Megan Skiendel: Who's most likely to have the novel, unactionable variant? And who's least likely to have access to testing infrastructure in the first place?
Zara Reyes: Same communities. That's the actual finding buried inside the equity headline. Research inclusion got delivered. Clinical equity — that's a totally separate project that nobody has funded yet.
Megan Skiendel: So picture a patient — say, someone whose family came from West Africa, they're at a clinic that theoretically has testing, they get tested — and the result comes back with a variant that's in zero of the CPIC guidance tables. The clinician has no protocol. The patient walked in, gave their DNA, and left with the same standard dose everyone else gets.
Zara Reyes: And the study gets cited as proof that pharmacogenomics finally included them. That's the narrative trap — the research inclusion becomes the headline and clinical equity never has to show up for work.
Megan Skiendel: And that's — honestly, that's where it settles for me. The one-size-fits-all prescribing model is empirically indefensible now. 98,590 people. The University of Pittsburgh team, Philip Empey, Science Translational Medicine — they made that case clean. But indefensibility has never automatically forced a system to restructure. That's the part nobody wants to say out loud.
Zara Reyes: And the restructuring question — who pays for the test, who mandates it, whether the FDA moves on actual policy versus just label language — that's completely unresolved. Like genuinely open. The 24.38% who were already mismatched before anyone looked — that number just sits there while the policy conversation catches up. Or doesn't.
Megan Skiendel: Yeah. That's the number. Not the hundred percent. The one in four.