Most people treat coffee as a stimulant with a side of habit. A 2026 open-access review in the Journal of Translational Medicine is more interesting, and more cautious, than that.
Jiang and Ding argue that coffee is a mix: caffeine, chlorogenic acids, trigonelline, diterpenes, and roasting melanoidins. Those compounds hit overlapping mitochondrial pathways (AMPK, SIRT1, PGC-1α, Nrf2, mitophagy). The authors call this a working hypothesis, not a proven causal model. That distinction matters. A lot of the molecular story still comes from cells and animals, often at concentrations you never see in human plasma.
What I actually use from this paper
Coffee is not just adenosine blockade. Trigonelline can be converted toward nicotinic acid and NAD+. Chlorogenic acids are mostly handled by the gut microbiome, so two people drinking the same cup are not getting the same metabolites. Unfiltered coffee (French press, Turkish, boiled) keeps cafestol and kahweol, which can raise LDL. Paper filters take out about 90% to 95% of those diterpenes and leave the water-soluble pieces.
The dose is not “more is better.” Observational meta-analyses cluster around 3 to 5 cups a day (roughly 300 to 400 mg of caffeine) for all-cause mortality and cardiovascular associations. The curve is U-shaped for the heart. Liver outcomes look more linear up to 5 to 6 cups. Past that, anxiety, insomnia, and blood pressure can show up in people who are sensitive. A “cup” is not standardized across studies, so treat those numbers as a range, not a prescription.
Genotype is not a parlor trick. CYP1A2 slow metabolizers (about 40% to 50% of people in this literature) keep caffeine around longer. Some studies link more than 2 to 3 cups a day in slow metabolizers to higher hypertension or myocardial infarction risk, a pattern not seen the same way in fast metabolizers. ADORA2A variants change sleep and anxiety. Hormonal contraceptives slow caffeine clearance. If coffee wrecks your sleep, that is data.
What this is not
It is not a license to drink a venti mocha and call it mitochondrial medicine. UK Biobank data in the review show the mortality association is strongest with unsweetened coffee. Sugar and syrups blunt it. High-fat creamers add a saturated-fat load that can cancel the cardiometabolic story.
It is not proven to prevent Parkinson’s, Alzheimer’s, type 2 diabetes, or NAFLD. Those inverse associations are real in cohorts. They are still observational. Reverse causality and healthy-user bias are on the table. The authors say so.
Pregnancy is a separate rule: EFSA and ACOG keep caffeine at or under 200 mg a day (about 1 to 2 cups).
Practical take, from a physician who drinks coffee and measures lipids
If you tolerate it, paper-filtered coffee, black or with a little milk, in the 2 to 4 cup range is a reasonable default. If you are a slow metabolizer, anxious, hypertensive, or sleeping badly, drop the dose and stop after noon. If you love French press, check LDL. If you load it with sugar, you are studying dessert, not coffee.
Jiang Z, Ding Y. Coffee as a polypharmacological modulator of mitochondrial health: from molecular mechanisms to translational implications. J Transl Med. 2026;24:1109.
https://doi.org/10.1186/s12967-026-08662-5
This is educational, not medical advice for your specific case.
