Medicine Is Evolution

How I think about biomedical mechanisms and why evolution is usually my first stop

When I’m trying to understand a biomedical mechanism I always ask the same question: why hasn’t evolution gotten rid of this? It sounds simple. It’s not. The answer almost always forces me to think about tradeoffs, timing, and selection pressure in ways that pure mechanistic thinking misses.

The framework I keep coming back to is this: when I encounter a gene, a protein, or a cellular behavior that looks purely pathological, I try to ask three questions before I do anything else. When was this useful? Who did it protect? And critically, at what life stage? Evolution is always running a local cost-benefit calculation. It doesn’t optimize for long-term comfort. It optimizes for survival and reproduction, which means it will cheerfully maintain something that helps you at 25 and kills you at 60, because by 60 your genes have already done their job.

Three examples that have shaped how I think about this:

Amyloid beta: the CNS bodyguard that overstays its welcome

The Alzheimer’s field spent a long time focused on amyloid as the problem to be cleared (same with Parkinson’s disease and alpha-synuclein). What shifted my thinking was a simpler question: why would a protein be so conserved across species if its only function were to destroy brain tissue? The answer is that amyloid beta has potent antimicrobial properties. It appears to function as part of the brain’s innate immune defense against pathogens breaching the blood-brain barrier. Asking the evolutionary question reframes the whole picture. It’s not a malfunction. It’s not just a toxic byproduct.

The accumulation that causes dementia happens decades after peak reproductive age, which means selection pressure against it is minimal to nonexistent. Evolution doesn’t care much about what happens to you at 75. It cared intensely about what killed you at 25, and at 25, having an antimicrobial peptide patrolling your CNS perimeter was probably very useful. The immunological logic is elegant. The timing is everything.

Sickle cell: paying the price in homozygotes to protect the herd

Carrying one copy of the sickle cell allele provides significant protection against Plasmodium falciparum malaria, a disease that historically killed people young and often. In environments where malaria was endemic, heterozygotes survived and reproduced at higher rates than either homozygote. The costs of the allele (severe disease in homozygotes, some effects in carriers) were outweighed by the survival advantage it conferred where the alternative was dying of malaria before you had children. Classic balancing selection. The “bad” gene stuck around because the selection pressure it was solving for was front-loaded in life, exactly where evolution is paying the most attention.

CFTR mutations: when protection against cholera becomes cystic fibrosis

Cystic fibrosis mutations are common enough in European populations (roughly 1 in 25 people carry a CFTR variant) that neutral drift doesn’t explain their prevalence. Something was maintaining them. The leading hypothesis is protection against cholera and typhoid, both of which kill through massive chloride-driven fluid secretion in the gut. CFTR mutations disrupt that exact mechanism. Heterozygous carriers appear to have been significantly more resistant to the dehydrating diarrheal diseases that were mass killers throughout human history, particularly in early childhood when such infections are most lethal. Early-life selection pressure, strong enough to maintain an allele, with the cost (severe lung and pancreatic disease) paid predominantly by the rare homozygote who inherited two copies.

The pattern across all three examples is the same. A mechanism that looks purely pathological in the clinical context where we encounter it was doing real protective work in a different context, at a different life stage, against selection pressures that no longer dominate the landscape the same way. For me, recognizing that doesn’t change the treatment. But it changes how I think about the biology and it changes what I expect when I try to understand how interventions will behave.

The pharmacology corollary

There’s a corollary to this framework that I find myself returning to when I’m thinking about drug mechanisms: evolutionary conservation predicts compensatory architecture. The more ancient and fundamental a pathway, the more redundancy evolution has built around it. That redundancy is exactly what a drug runs into.

If a pathway has been conserved for hundreds of millions of years, the organism has had hundreds of millions of years to build fail-safes against its disruption. mTOR is a good example: so fundamental to nutrient sensing and cellular growth that inhibiting it triggers multiple compensatory feedback loops simultaneously, some of which partially restore the signaling you were trying to block. The VEGF story in oncology is another. Cut off one pro-angiogenic signal and tumors route around it through alternative pathways, often faster than you’d expect. The renin-angiotensin system has so much redundancy baked in that single-point inhibition rarely produces the clean outcome the whiteboard predicted.

The way I’ve come to think about this: when a drug hits a compensatory wall, evolution is showing you exactly how much it valued that function. The strength of the feedback is roughly proportional to how lethal losing that function was, historically. That reframe makes the compensatory response informative rather than just frustrating. It points somewhere. The question I try to ask is: what did evolution build to make sure this process keeps running even when disrupted? The answer to that question is usually where the next layer of biology is hiding.

Why I keep coming back to this

I find this lens useful because it imposes a kind of discipline on mechanistic thinking. It forces a before-and-after: before this was pathology, what was it? It pushes back against the assumption that because something causes harm in the context we observe it, harm is all it does. And it grounds expectations about intervention in something more durable than a pathway diagram. It grounds them in the history of what that pathway was actually selected to do.

The gut check stays the same regardless of the specific biology. When I encounter something that seems purely destructive, I ask: when was this useful, who did it protect, and at what life stage? And then: what did evolution build to keep this running even when disrupted? The answers aren’t always there. Sometimes things really are neutral drift, or developmental artifacts, or genuine misfires in environments our genomes weren’t built for. But asking the question changes how I approach the problem. In my experience, that’s usually worth the time.

---

Stay Curious,

Andrew