Areas where scientific evidence is lacking or incomplete.
The definitive safety profile for continuous exposure longer than three years has not been established in large-scale RCTs. Prospective clinical monitoring is required to rule out potential late-onset safety concerns.
Implications: Patients on indefinite treatment are operating beyond the evidence window. Post-marketing surveillance is the primary safety net.
All therapeutic benefits for Alzheimer's and Parkinson's diseases are from preclinical animal models. No human CNS trials have been completed. Pathways like alpha-synuclein clearance and NLRP3 suppression remain unproven in humans.
Implications: Neurological claims cannot inform treatment decisions. The translation gap between rodent pharmacology and human cognition is historically large.
It is unknown if tirzepatide is safe or effective in children and adolescents under 18. No paediatric trials have reported.
Implications: Off-label paediatric use has no evidence basis. Growth-period effects on bone accrual and development are unstudied.
SURPASS-CVOT established non-inferiority to dulaglutide for 3-point MACE but did not achieve statistical superiority. The 16% all-cause mortality reduction was driven by non-cardiovascular deaths. No direct lipid particle data (non-HDL, LDL size shift) exists.
Implications: Unlike semaglutide (SELECT trial), tirzepatide cannot claim demonstrated CV risk reduction over an active comparator. Clinicians should not assume CV superiority.
BMD declines of 2.1-2.6% at hip and spine are documented over 52 weeks. Whether this loss persists beyond 52 weeks, is dose-dependent, or is reversible on discontinuation is unknown.
Implications: Long-term skeletal cost of continuous therapy is unquantified. The uncoupling of bone resorption from formation is mechanistically concerning.
SURMOUNT-1 reports 25% lean mass loss, but nSights EHR real-world data shows significantly higher absolute lean mass loss compared to semaglutide (excess 1.1-2.0% over 12 months). This discrepancy between controlled and real-world settings is unexplained.
Implications: The standard mitigation advice (protein + resistance training) may be calibrated to trial populations, not real-world patients who may not adhere to exercise protocols.
An initial decrease in eGFR observed at 12 weeks of treatment recovers over time, but the exact physiological pathways responsible are not defined. It is unclear whether this is haemodynamic, tubuloglomerular feedback, or dehydration-related.
Implications: Without understanding the mechanism, clinicians cannot distinguish a benign haemodynamic effect from early renal injury in individual patients.
Expert disagreements and competing evidence.
SURMOUNT-1 DXA substudy shows approximately 25% of total weight lost is lean mass, consistent with standard caloric restriction ratios.
Post-hoc DXA analysis from SURMOUNT-1 controlled trial.
Source: SURMOUNT-1 DXA substudy
Multi-centre EHR data (nSights network) reveals significantly higher absolute lean body mass loss with tirzepatide compared to semaglutide, with excess losses of 1.1-2.0% over 12 months.
Retrospective EHR analysis across multiple centres.
Source: nSights EHR network analysis
Verdict Note
Both datasets are methodologically valid for their context. The discrepancy likely reflects adherence differences between trial and real-world settings, but no study has directly tested this hypothesis.
GIP receptor activation theoretically supports bone remodelling, and one safety review suggested tirzepatide may have advantages in bone formation.
Mechanistic reasoning from GIP receptor biology.
Source: GIP receptor pharmacology reviews
Weill Cornell retrospective data shows significant BMD declines at hip and femoral neck correlated with weight loss magnitude. Preclinical mouse models show decreased bone mass accrual.
Weill Cornell retrospective study and preclinical mouse models.
Source: Weill Cornell retrospective study
Verdict Note
The clinical data overrides the mechanistic prediction. In practice, tirzepatide is associated with BMD decline. The theoretical GIP bone benefit does not compensate for weight-loss-driven bone loss at the doses and timescales studied.
Post-hoc analysis of SURMOUNT-5 projected that tirzepatide would produce greater 10-year CV risk reduction than semaglutide.
SURMOUNT-5 post-hoc modelling of 10-year risk.
Source: SURMOUNT-5 post-hoc analysis
SURPASS-CVOT confirmed non-inferiority to dulaglutide but did not establish statistical superiority for the primary 3-point MACE endpoint.
SURPASS-CVOT randomised trial, primary endpoint.
Source: SURPASS-CVOT trial
Verdict Note
Predictive models are hypothesis-generating, not confirmatory. Until a dedicated superiority trial reports positive MACE data, tirzepatide does not have demonstrated CV risk reduction advantage over active comparators.
A 2025 meta-analysis found tirzepatide has a higher overall GI adverse event rate (RR 2.94 vs placebo) compared to semaglutide (RR 1.68).
2025 meta-analysis of GI adverse event rates.
Source: 2025 GI adverse event meta-analysis
In the SURMOUNT-5 head-to-head trial, tirzepatide patients had lower treatment discontinuation due to GI events (2.7%) than semaglutide patients (5.6%).
SURMOUNT-5 head-to-head trial discontinuation data.
Source: SURMOUNT-5 trial
Verdict Note
Both claims are accurate. Tirzepatide causes more GI events by count but fewer patients stop treatment because of them. The GIP buffering hypothesis explains this: more frequent but milder events are more tolerable than fewer but severe ones.
Most preclinical studies show tirzepatide repairs brain metabolism, reduces neuroinflammation via NLRP3 suppression, and clears amyloid-beta through autophagy activation.
Multiple preclinical in vitro and in vivo studies.
Source: Preclinical Alzheimer's studies
One study found tirzepatide did not reduce amyloid plaques or improve cognitive deficits in specific transgenic Alzheimer's mouse models.
Single transgenic mouse model study with negative results.
Source: Transgenic AD mouse model study
Verdict Note
The positive preclinical signal is stronger by volume, but the negative result cannot be dismissed as a single outlier. Until human trials report, the neuroprotective claim remains provisional.