<- Home

Methodology

The rubric, neutrality rule, and what a grade does and does not mean. Data as of 2026-07-14.

The robustness rubric

High evidenceLarge, well-run randomized trials and peer-reviewed pivotal or regulatory-grade records.

Moderate evidenceSmaller randomized trials, strong observational studies, and prespecified secondary analyses.

Low evidenceEarly-phase trials, post-hoc analyses, small substudies, and animal-dominant evidence.

Very low evidencePress releases, conference toplines, registry results, and in-vitro work. Registered studies with no results are listed separately and are not graded as results.

What a grade is not

A grade rates the strength of the evidence behind a finding. It does not rate the drug, and it does not say an effect is good for you. Colour encodes how much weight a source can bear, never benefit or harm.

The neutrality rule

The library records what has been observed and leaves disagreements standing side by side. It never ranks drugs, never recommends, and never frames a finding as a treatment decision.

Standing caveats

C1: Mouse tirzepatide is not a GIP attribution

The mouse GIPR is weak/partial for tirzepatide (~3-60x reduced potency vs human; El K et al., Nat Metab 2023). Mouse tirzepatide effects largely reflect the GLP-1 component. Apply every time mouse tirzepatide data appears.

C2: Acute vs chronic glucagon energy-expenditure differ

Acute human glucagon EE shows no detectable acute BAT recruitment in the one imaging study (Salem et al., DOM 2016); chronic rodent glucagon EE involves a UCP1/FGF21/sympathetic BAT component. Different time-frames and species; do not conflate.

C3: Human GIP-on-EE: separate two epistemic states

(a) Human GIP-on-EE is measured-null under acute infusion (REE shows no effect across isolating acute-infusion datasets). (b) It is unmeasured under chronic selective agonism (LY3537021 / T2-003 never measured EE/substrate/BAT). Do not fuse 'measured negative' with 'no data'. A low-power vasodilation-confounded skin-temperature surrogate does NOT establish thermogenesis. Treat the chronic-selective-agonism gap as permanent until a selective-agonist calorimetry study says otherwise.

C4: Retatrutide human EE is inferred, not measured

No published retatrutide trial has directly measured energy expenditure by calorimetry; the EE story rests on glucagon physiology plus a pair-fed mouse.

C5: Supra-additive EE synergy is unproven

The energy-expenditure effect is best described as additive/glucagon-driven, not supra-additive.

C6: Only the GCGR arm is isolated within retatrutide itself

The Coskun GCGR-antibody experiment isolates the GCGR arm. GLP-1R and GIPR contributions within the molecule are proxy-based; several are permanently unresolvable in humans.

C7: Commercial/vendor sources are marketing, not evidence

Never cited as primary (e.g. peptide-vendor 'activates brown fat in humans' claims).

C8: Rodent EE/adaptation datasets carry no recorded housing/ambient-temperature condition

Load-bearing rodent EE/adaptation sources do not record ambient housing temperature. Sub-thermoneutral housing (standard ~22 C) systematically inflates rodent adaptive thermogenesis and UCP1 share. Mouse-human EE discordance must not therefore be read as purely a species difference; part of it may be a housing artefact.

C9: Retatrutide lean-mass reassurance is provisional, proportional, and combination-level

It rests on a single Phase 2 T2D, fat-primary-endpoint, sponsor-authored, completer-only DXA substudy. Absolute lean-kg lost is unquantified; proportional parity can hide the largest absolute lean loss of any agent. DXA 'lean mass' conflates muscle with organ/bone/fluid. Treat 'no disproportionate lean penalty' as provisional, proportional-only, and combination-level - not as demonstrated equivalence.

C10: In-vitro receptor potency does not map to in-vivo arm contribution

Relative EC50/potency ratios describe in-vitro pharmacology only. They do NOT establish in-vivo functional arm-dominance: at therapeutic doses receptors approach saturation so the ratio stops governing occupancy. Never carry a potency ratio into a functional/arm-contribution conclusion without this flag.

ACB: Asymmetric-confidence baseline (do not flatten into one confident voice)

GCGR-driven energy expenditure: HIGH (acute pharmacology) / UNRESOLVED (chronic) - the HIGH attaches only to the acute human EE rise; the chronic decision-bearing claim is unstudied. GLP-1-driven browning: MODERATE (rodent isolating only; weak human translation). GIP-on-EE / thermogenesis: UNRESOLVED (no isolating human evidence).

Not medical advice. This page describes how the evidence is graded. It does not diagnose, recommend, give dosing, or replace a qualified clinician.