chatGPT(6max)
s11357-026-02643-8.pdf, by Darren E. Stewart and William E. Fitzsimmons, examines rapamycin and survival after kidney transplantation.
My assessment: this is a useful long-term observational study showing no detectable survival advantage in a selected transplant population. Its relevance to healthy people taking intermittent rapamycin is limited, and its claim to novelty needs qualification.
The researchers analysed US transplant registry data for people transplanted between 2000 and 2007, following outcomes through December 2025. The main comparison included:
- 4,885 recipients prescribed an mTOR inhibitor at discharge and at their first-year follow-up, predominantly sirolimus.
- 6,589 recipients prescribed no mTOR inhibitor at either visit.
- All participants had survived with a functioning transplanted kidney to the first-year visit and were receiving tacrolimus or cyclosporine.
- Median follow-up was approximately 11.5 years, with 6,443 deaths overall.
The groups initially differed in age, transplant year, accompanying medicines and other characteristics. The authors used propensity weighting to balance 18 measured factors.
Their adjusted results were:
| Analysis | Median survival age: mTOR inhibitor group | Comparison group | Difference, with 95% confidence interval |
|---|---|---|---|
| Follow-up stopped at kidney graft failure | 66.1 years | 66.7 years | -0.6 years, from -2.4 to 0.8 |
| Follow-up continued after graft failure | 64.6 years | 64.3 years | 0.3 years, from -1.6 to 2.0 |
Neither comparison showed a statistically significant difference. Alternative cohort definitions and analyses separating tacrolimus from cyclosporine produced broadly similar findings.
These are estimated survival ages within a selected transplant population. They are not estimates of general population life expectancy.
The main novelty is the extended follow-up and presentation of survival on an age scale. The study brings unusually long human follow-up to a question often discussed using animal experiments and short human trials.
However, the claim that this is the first large retrospective study evaluating rapamycin and human lifespan is difficult to sustain without a much narrower definition:
- Isakova and colleagues, 2013: analysed 139,370 US kidney transplant recipients, including mTOR inhibitor regimens, and examined mortality using adjusted and propensity-matched analyses. PubMed
- Badve and colleagues, 2016: studied 9,353 adult recipients whose grafts survived at least one year. They examined mortality using both time-varying exposure and treatment recorded at fixed visits, finding higher mortality associated with mTOR inhibitor use. This is particularly relevant prior work because of its similar survivor restriction. PubMed
- Knoll and colleagues, 2014: pooled individual patient data from 21 randomised trials involving 5,876 recipients. Sirolimus reduced malignancy but was associated with increased mortality. The BMJ
These studies used different treatment comparisons and follow-up periods, so the new analysis adds information. Nevertheless, the paper should explicitly discuss them and explain why its findings differ. None of these three important mortality studies appears in its reference list.
The study’s strongest features are its large sample, substantial number of deaths, long follow-up, mortality linkage, explicit assessment of measured covariate balance, and several sensitivity analyses. Including follow-up after graft failure is especially valuable.
The principal weaknesses are:
-
The findings have limited applicability to otherwise healthy adults.
These participants had previously experienced kidney failure, undergone transplantation and received combinations of immunosuppressive drugs. Their underlying risks and treatment interactions differ substantially from those of healthy people considering rapamycin.
The study also lacks the dosing information needed to assess intermittent longevity regimens. The authors acknowledge these limitations. Their discussion of lifespan extension at immunosuppressive doses in mice is relevant, but it does not establish equivalent exposure, toxicity or biological effects in humans.
-
Long follow-up does not establish long-term continuous treatment.
Exposure was classified using prescriptions at discharge and around one year. Later treatment information was limited: mTOR inhibitors were recorded at 75% of subsequent annual visits during years two to five in the exposed group, versus 3% in controls.
That demonstrates continuing separation between the groups, but leaves considerable uncertainty about discontinuation, switching, adherence and treatment beyond five years. Such changes could dilute a genuine benefit or harm.
Consequently, the study primarily examines an initial prescribing pattern, rather than a verified course of continuous treatment lasting decades.
-
Propensity weighting cannot remove all confounding.
Treatment was not randomised. Factors such as previous malignancy, infection risk, proteinuria, wound complications and transplant centre practice could influence both treatment choice and survival. The authors acknowledge several of these missing variables.
There is also a subtler problem: adjustment includes first-year kidney function, rejection and hospitalisation, measured after treatment began. These could partly be consequences of treatment. Adjusting for them may remove part of the treatment effect or introduce bias. The intended causal effect needs clearer definition.
-
Selection of first-year survivors narrows the question.
Excluding people who died, lost their graft or changed regimens during the first year selects a particular group of survivors and treatment tolerators.
This is reasonable for investigating subsequent survival among that group. It cannot establish the overall benefits and harms of starting rapamycin at transplantation, because early outcomes have already been excluded. The direction of the resulting selection bias is uncertain.
-
Stopping follow-up at graft failure complicates the primary analysis.
Graft failure predicts subsequent mortality and could itself be affected by treatment. Removing patients from observation when their graft fails can therefore bias the survival comparison.
The analysis retaining deaths after graft failure addresses this concern and is more directly relevant to overall longevity. I would give it greater prominence than the graft-failure-censored analysis.
-
A non-significant result does not demonstrate equivalence.
The full-follow-up estimate was a 0.3-year difference, but its confidence interval ranged from a 1.6-year reduction to a 2.0-year increase. Under the model’s assumptions, that leaves room for effects that could matter clinically.
The defensible conclusion is that the analysis detected no clear survival advantage. It does not establish that the treatments have identical effects.
-
The study cannot explain the biological reason for its result.
It measures survival, without directly measuring ageing rate, autophagy, functional decline or other proposed mechanisms.
A neutral mortality result could reflect absent benefit, opposing beneficial and harmful effects, or exposure misclassification. Cause-of-death information was unavailable for approximately 54% of deaths, limiting attempts to distinguish these explanations. Comparing the proportions of recorded deaths assigned to particular causes also does not establish equal cause-specific mortality rates.
-
There are concrete reporting inconsistencies.
- Figure 1 appears to reverse the group-specific death counts. It places 3,729 deaths under the mTOR inhibitor group and 2,714 under the comparison group; Table 1 assigns those counts in the opposite direction.
- Figure 3’s caption does not match the plotted medians. The adjusted plot shows 64.6 versus 64.3 years, while the caption gives 66.1 versus 66.7 years, corresponding to the graft-failure-censored analysis.
These discrepancies warrant correction and a consistency check against the analysis outputs.
The paper supports a cautious statement: among selected kidney transplant recipients surviving the first year, initial mTOR inhibitor prescribing was not associated with a clear improvement in adjusted long-term survival. Demonstrating an effect on human ageing, or evaluating intermittent rapamycin in healthy adults, requires evidence that this study cannot provide.