Azithromycin and roxithromycin

The senolytic claim for these two antibiotics rests on one paper, published in Aging in 2018 by a group at the University of Salford.

They used a screening assay to look for approved antibiotics that would preferentially kill senescent cells. Two human fibroblast lines, MRC-5 and BJ, were pushed into senescence by eight days of BrdU, a DNA-damaging agent, and viability was read by protein content. Azithromycin and roxithromycin came out as hits. Erythromycin, the closely related parent compound, showed no senolytic activity at all, which the authors themselves flag as evidence of how specific these interactions are. Validated on a second platform measuring electrical impedance, azithromycin preferentially removed about 97 percent of senescent cells, which the authors describe as a near 25-fold reduction. Azithromycin also strongly induced aerobic glycolysis and autophagy in these cells, which is their proposed mechanistic explanation. The paper discloses that two of the authors hold a minority interest in Lunella Biotech.

That is the entire senolytic evidence base. Cells in a dish, one laboratory, one publication. There is no animal lifespan study, no animal healthspan study, and no human trial of either drug as a senolytic.

What there is, in abundance, is human safety data from using these drugs as antibiotics, and it does not read as reassuring for anyone contemplating long-term use.

The ALLOZITHRO trial randomised 480 patients undergoing allogeneic stem cell transplant to 250 mg of azithromycin three times a week or placebo for two years. Thirteen months in, the data and safety monitoring board detected an unanticipated imbalance in haematological relapses and the trial was stopped. Two-year survival was 56.6 percent on azithromycin against 70.1 percent on placebo, hazard ratio 1.5. The two-year cumulative incidence of haematological relapse was 33.5 percent against 22.3 percent. This was long-term, low-dose azithromycin in people, which is exactly the regimen a senolytic use would imply, and it killed more people than placebo.

Separately, a Tennessee Medicaid cohort study found that during five days of azithromycin, compared with no antibiotic, cardiovascular death rose (hazard ratio 2.88) and death from any cause rose (1.85). In absolute terms this was small, an estimated 47 extra cardiovascular deaths per million courses, rising to 245 per million in the highest cardiovascular risk decile. Macrolides prolong the QT interval, which is the accepted explanation.

This is a single in vitro paper being carried a long way.

The result itself is clean and internally consistent, validated on two assay platforms, with a well-chosen negative control in erythromycin. It has not been replicated by an independent group, has never been tested in an animal, and the laboratory that produced it discloses a commercial interest in the space. That is not an accusation, it is disclosed in the paper, and it is the sort of context a reader is entitled to before deciding what one screening paper is worth.

The reason this compound gets a page rather than a line is that the gap between the in vitro claim and the human record is unusually instructive. A senolytic use of azithromycin would mean taking a long-half-life antibiotic repeatedly for years. That exact regimen has been randomised against placebo in people, in ALLOZITHRO, and it was stopped early because more people died. Immunocompromised transplant patients are not a general population, and the finding is still the closest thing to a test that exists.

There is also a cost that falls on other people. Long-term macrolide use drives antibiotic resistance, which is why it is restricted to indications like cystic fibrosis and bronchiectasis where the benefit is established. Using it speculatively for ageing spends a shared resource for an unproven personal gain.

No senolytic of any kind has yet improved a clinical outcome in a randomised human trial. That sentence is the single most useful thing to carry into any senolytic discussion. Senescent cell burden has been reduced in people, by dasatinib plus quercetin, and the field has measured that reduction in skin and fat biopsies. What has not followed is a randomised trial in which people who got a senolytic ended up walking further, breaking fewer bones, or living longer than people who did not. The two randomised trials that tried, in pulmonary fibrosis and in postmenopausal bone turnover, did not separate from control.