Tyrosine is marketed as a cognitive enhancer. It is a dopamine precursor. The assumed pathway: more precursor, more neurotransmitter, better cognition. A Mendelian randomization study of 270,000 people found that genetically elevated tyrosine shortens male lifespan by eleven months. The actual pathway: metabolic load, insulin resistance, accelerated aging. The body does not use the molecule the way the label describes.
The shingles vaccine is marketed as preventing herpes zoster. The assumed pathway: immune targeting of varicella-zoster virus, reduced shingles incidence. A study of 3,884 adults over 70 found that vaccination slows biological aging across seven molecular clocks, with effects persisting four or more years. The actual pathway: blocking chronic low-grade inflammation from subclinical viral reactivation, reducing the background inflammatory burden that drives epigenetic aging. The body benefits from the vaccine through a mechanism broader than the one it was designed for.
Both interventions operate through pathways different from their marketed ones. Tyrosine's real mechanism is metabolic, not cognitive. The shingles vaccine's real mechanism is anti-inflammatory, not specifically anti-viral. In both cases, the label describes the intended target. The body routes the intervention through a different system.
This is not unique to these two cases. It is a structural feature of interventions in complex systems. A molecule entering a body does not follow instructions. It interacts with every system it contacts. The intended pathway is one possibility. The actual pathway is determined by the organism's physiology, not the intervention's design. When the intended and actual pathways diverge, the surprise can go either direction. Tyrosine harms where it was supposed to help. The shingles vaccine helps where it was not supposed to reach.
The methods that revealed both findings matter. Mendelian randomization strips out confounding by using genetic variants as natural experiments. Epigenetic clocks measure biological aging independent of disease categories. Neither method would have detected the divergent pathway through the intervention's own framework. You need a measurement system that is indifferent to the intervention's intended mechanism to see what the intervention actually does.