Opens the ectopic-fat cluster. The organizing claim, induced across an individual-mechanism source
(Taylor’s personal-fat-threshold hypothesis) and a population-outcome meta-analysis (Kramer’s
metabolically-healthy-obesity MA): cardiometabolic risk tracks WHERE fat is stored — in the liver,
pancreas and viscera rather than subcutaneously (the subcutaneous-overflow mechanism is the wiki’s
adipose-expandability reading, not Taylor’s stated hypothesis — see below) — more faithfully than it
tracks total fat mass or BMI. This reframes three decisions the fat-mass / BMI lens gets wrong: who is
at risk (some normal-weight people are; some obese people are less so, but not durably safe), what the
operative variable is (depot / metabolic status, not the scale), and why the risk reverses (drawing
ectopic fat back down). type-A
The bidirectional energy-toxicity spine
Forward (accumulation). Taylor & Holman define the personal fat threshold (PFT) at the organ level: «It is hypothesized that each individual has a PFT (dotted line) above which excess fat is stored within the liver and the pancreas.» — set by the extent of intra-organ fat and individual susceptibility to its local effects, driving hepatic insulin resistance and β-cell lipotoxicity, and thence type 2 diabetes. (Taylor & Holman, 2014) The subcutaneous-capacity / overflow reading (surplus fills subcutaneous adipose first, then spills into organs and viscera) is a related adipose-expandability model the wiki lays over Taylor’s PFT, not his stated hypothesis — Taylor never mentions subcutaneous fat. (corrected 2026-10-06: overflow-from-subcutaneous framing was tagged to Taylor -> wiki inference; self-critique) Taylor states the threshold as a hypothesis: «We hypothesize that each individual could have a personal fat threshold (PFT) which de- termines their susceptibility to developing T2DM … Gaining sufficient weight to cross their PFT will trigger the condition» — and «the hypothesized PFT is independent of BMI». (Taylor & Holman, 2014)
Backward (reversal). Negative energy balance draws the ectopic depots down and can restore function — the twin-cycle basis of T2D remission: «people with recent onset T2DM could regain normal glucose control and normal β-cell function when the fat content of the liver and the pancreas was decreased by a weight loss diet- ary regimen» — «achievable equally readily by people with lower initial BMI». (Taylor & Holman, 2014) The reversal arm is separately gold-backed by the held RCT evidence -> Total Diet Replacement and Type 2 Diabetes Remission.
The depot distinction — three fat compartments are three different objects type-B
- Subcutaneous — on the adipose-expandability reading, the safe expandable store, not itself pathogenic (no held source in this spine measures subcutaneous storage capacity; Taylor’s ethnicity data are BMI-equivalent risk cut-points, not storage capacity — see below).
- Intra-organ (intra-hepatic + intra-pancreatic) — the pathogenic depot. Liver fat is «pivotal» and drives hepatic IR; pancreatic fat suppresses glucose-mediated insulin secretion (β-cell lipotoxicity).
- Visceral — commonly treated as the villain, but Taylor demotes it to a marker: «Extent of visceral fat accumulation is a surrogate marker for intra-organ fat ex- cess, but is not pathophysiologically related to adverse metabolic consequences» (citing Fabbrini 2009, Kantartzis 2010). (Taylor & Holman, 2014)
Two guidance/epidemiology sources put visceral fat at the causal centre — Taylor’s marker-demotion is
the minority view type-D. A second body now joins INTERHEART (below) against Taylor’s demotion:
AASLD’s MASLD guidance calls visceral fat causally central, not a bystander — «Visceral fat, which is
more metabolically active and inflammatory than subcutaneous fat, mediates the majority of this risk.»
(Rinella et al., 2023) This is genuinely opposed to Taylor’s
«surrogate marker … but not pathophysiologically related» on the same quantity (visceral fat’s causal
role). But the issue is not fully joined: AASLD’s own mechanism routes partly through intra-organ
fat — insulin signalling impairment «promoting the inappropriate release of fatty acids leading to
intrahepatic lipid accumulation» — so AASLD may be describing visceral fat as an upstream driver that
acts via hepatic fat, which is compatible with Taylor’s “marker of intra-organ excess” if the two are
tightly coupled. The live disagreement is narrow and unresolved: is visceral fat causally upstream
(AASLD, INTERHEART) or a correlated readout of the intra-organ depot that does the damage (Taylor)?
Held sources do not settle it; the decision-relevant consequence is small because both readings point to
the same lever (draw the depots down with energy deficit) and the same measurement (waist/visceral fat
remains a useful risk marker either way).
So waist circumference / visceral fat is a surrogate for the surrogate — useful because it correlates with intra-organ fat, but a step removed from the depot doing the damage -> Surrogate Outcomes. A fasting-lab readout of the same insulin resistance the intra-organ fat drives — the triglyceride-glucose (TyG) index — sits one further step removed again: it predicts cardiovascular events but is a marker of the atherogenic-dyslipidemia / IR state these depots generate, not the depot or a treatment target -> Insulin Resistance Surrogates and Cardiovascular Risk.
The same marker-vs-cause doubt reaches the intra-organ depot itself type-F
The tension above asks whether visceral fat is causal or a readout of intra-organ fat. Mantovani’s cohort meta-analysis pushes the same doubt one level further in — onto liver fat, the depot Taylor calls «pivotal». NAFLD predicts incident T2D strongly (pooled HR 2.22, 95% CI 1.84-2.60; fully adjusted 1.85, 1.47-2.22), and the authors argue biological plausibility (hepatic insulin resistance, pro-diabetogenic hepatokines) — yet they state, in the same paper, that “It remains uncertain whether NAFLD causally increases diabetes risk or is a marker of other shared risk factors” and that “the observational design of the eligible studies does not allow for proving causality.” (Mantovani et al., 2018). Three dissociation lines make this a substantive caveat, not boilerplate:
- Genetic (natural experiment): «there may be a dissociation between NAFLD and insulin resistance in humans carrying some genetic variants, such as the patatin‐like phospholipase 3 gene» — possibly high liver fat without the expected IR (Mantovani et al., 2018). (corrected 2026-10-06: carriers show -> may be dissociated, full clause; self-critique)
- Experimental: “specific manipulation of liver fat is insufficient to affect insulin sensitivity/glycemia” in animal models (Mantovani et al., 2018).
- Ethnic: “obese black individuals exhibit a lower prevalence of NAFLD but similar type 2 diabetes prevalence” — liver fat and dysglycemia coming apart across populations (Mantovani et al., 2018).
Why this refines rather than files a new tension. Mantovani holds both positions (plausibility AND the dissociation caveat), so it cannot be set as one pole against Taylor’s «pivotal» without cherry-picking half of a multi-position source (not-joined check (iii)). Read together with the visceral tension above, the pattern is consistent: each fat depot on the way in (visceral, then the intra-organ depots including liver fat) is a candidate treacherous surrogate — a strong predictor whose causal transmission to dysglycemia is only partly evidenced, with genetic/experimental cuts against a clean fat-drives-IR chain. The genetic line is not outside doubt against Taylor, though: Taylor cites the same PNPLA3 study (Kantartzis 2009 — Mantovani’s ref 44, Taylor’s ref 39) beside «pivotal» and absorbs it as variable susceptibility — «The variable effect is illustrated by one known genetic influence in that individuals with the G-allelle of patatin-like phospholipase 3 gene have a higher liver fat level but normal hepatic insulin sensitivity [39].» (Taylor & Holman, 2014) — so the PFT is already extent x susceptibility, not a one-factor chain, and the genetic line refines rather than cuts against it; the animal line and Mantovani’s black-white liver-fat contrast are not in Taylor (his ethnicity data are BMI-equivalent risk cut-points). (corrected 2026-10-06: PNPLA3 line framed as a new doubt against Taylor -> shared primary study Taylor already absorbs; self-critique) The decision-relevant residue is the reversal gap: Mantovani notes “it remains to be definitely proven that improving the liver condition in NAFLD decreases risk of developing diabetes” (Mantovani et al., 2018) — so liver fat earns its place as a prognostic marker and monitoring handle more firmly than as a validated treatment target for diabetes prevention. The lever (draw the depots down with an energy deficit) is unchanged, since it acts on the shared upstream driver regardless of which depot is the proximate cause -> Surrogate Outcomes. (inferred from Mantovani et al., 2018)
Why BMI misleads — the individual-vs-population gap
BMI is a population statistic misapplied to a person. Taylor’s UKPDS distribution is unimodal (no distinct non-obese subtype), yet 36% of newly-diagnosed T2DM had BMI <25 (vs 64% <25 in the contemporaneous UK population). Taylor’s own reading of the strong obesity-T2DM association today is that it reflects the population BMI distribution shifting right, not that T2DM requires obesity — susceptibility is set by whether a person crosses their own threshold, not by a population cut-point (corrected 2026-10-06: was more fat than they can store safely — the overflow framing, not Taylor’s; cold-audit). The threshold itself shifts by ethnicity: «the equivalent degree of risk for a Caucasian of BMI greater than 30 kg/m2 is expressed in South Asians at 25.2 kg/m2 and at 27 kg/m2 in African/Caribbeans». (Taylor & Holman, 2014)
The population signature — Kramer’s MHO meta-analysis reads as the same claim at scale type-A
If risk tracks depot/metabolic status rather than mass, then (a) obesity should not be safe merely because metabolic markers are currently normal, and (b) normal-weight people with a bad metabolic profile should carry high risk. Kramer’s MA of observational cohorts shows exactly this pattern:
- Metabolically healthy obesity is not durably benign: RR 1.24 (1.02-1.55) vs metabolically-healthy normal-weight, in studies with >=10 y follow-up (overall RR 1.19, CI crosses 1) — «there is no healthy pattern of increased weight».
- Metabolic status dominates BMI: metabolically-unhealthy normal-weight carries RR 3.14 (2.36-3.93), «equal to» metabolically-unhealthy obese (RR 2.65). (Kramer et al., 2013)
Opio 2020 refines Kramer on the same question — larger, and it answers the load-bearing sub-question type-F
Opio 2020 is the up-to-date gold SR+MA of the same relationship (23 prospective cohorts, n = 4.49 M; nine new studies, +4.0 M participants over the prior reviews), re-pooling Kramer’s cohort base and reporting findings «consistent with literature by Kramer et al.» — so it refines the dated incumbent, it is not an independent second route. It sharpens the population signature in three decision-relevant ways:
- The excess risk holds even with ZERO metabolic risk factors — the sub-question Kramer could not answer. Restricting to the strictest metabolic-health definition (absence of any risk factor): MHOW RR 1.51 (1.21-1.88, n = 5), MHO RR 2.18 (1.28-3.71, n = 5); «The risk of CVD remained high even when there were no metabolic risk factors.» Opio flags this as the advance over the two prior MAs that tried the stratum and were underpowered (Eckel, Zheng found it non-significant). (Opio et al., 2020)
- No 10-year latency requirement. Where Kramer’s MHO signal emerged only after ~10 y, Opio finds the risk present at both windows with no duration gradient (MHOW <10 y RR 1.34 vs >=10 y RR 1.34, subgroup-difference p = 0.98; MHO if anything larger at <10 y, RR 1.89). So the “small, slow, time-latent” reading below is refined: the excess is not conditional on a decade of follow-up. (Opio et al., 2020)
- Metabolic status still dominates BMI: metabolically-unhealthy normal-weight RR 3.07 (2.27-4.15) — «metabolic abnormality confers an even greater risk of CVD in individuals with normal weight». This near-replicates Kramer’s MU-NW RR 3.14 on an overlapping-but-larger cohort base (agreement on a re-pooled base, not independent-E corroboration). (Opio et al., 2020)
Opio’s own conclusion runs the reframe to its edge: «Hence the term ‘metabolically healthy’ may be a misnomer.» The honest reading is one notch softer than the slogan — the 0-risk-factor estimates rest on five studies each with wide CIs and I2 up to 94%, and most cohorts did not adjust for cardiorespiratory fitness (Ortega attributed much of the MHO-MHNW gap to CRF), so residual fitness confounding could shrink the depot-independent excess (the fitness-vs-fatness distinction).
Same quantity as Kramer? A refinement with one distinction (parameter check). Reference group is the same (MHNW both). The MU-normal-weight contrast is the same quantity and near-identical (3.07 vs 3.14). But the headline MHO number is not a clean 1:1 supersession: Kramer’s estimate combined all-cause mortality with CV events, whereas Opio reports CVD events separately (MHO RR 1.58 overall) and pools a broader set of metabolic-health definitions — so Opio’s larger RR is a related, not identical, quantity. And Opio reports no absolute risk — the absolute anchor (~0.7% over 10-11 y) still comes only from Kramer. Net: a type-F refinement (newer/larger gold answering the 0-risk-factor and durability sub-questions), not independent corroboration and not a numeric replacement of Kramer’s headline.
How the two sources relate (not laundered independence). Taylor and Kramer govern different objects — an individual causal hypothesis vs a population risk contrast — and they measure different proxies (Taylor: intra-organ fat / β-cell function; Kramer: metabolic-syndrome criteria). Their agreement is that BMI is not the operative variable, reached from two directions; it is emergent synthesis, not strict E-independence, because Kramer never measures ectopic fat — metabolic-syndrome status is a coarse stand-in for the depot biology Taylor describes. The convergence raises confidence in the reframe, not in any shared number.
The hard-outcome instantiation — INTERHEART maps the depot distinction onto MI type-F
Taylor and Kramer are glycemic/metabolic-status sources; INTERHEART extends the depot distinction to a hard cardiovascular endpoint (first acute MI, 27,098 people, 52 countries) and reads the same shape at population scale: abdominal girth harmful, larger hips protective (fat or muscle — see below), and BMI the wrong instrument. Top vs bottom quintile OR for MI (adjusted for BMI): waist 1.77 (1.59-1.97) harmful, hip 0.73 (0.66-0.80) protective; waist-to-hip ratio is the single strongest marker (per 1 SD 1.37, 1.34-1.41) while BMI is the weakest (1.10, 1.07-1.13) and its whole MI association vanishes after adjusting for WHR (1.44 -> 1.12) then the other risk factors (0.98). (Yusuf et al., 2005) Full effect table and the load-bearing case-control design caveat live at Waist-to-Hip Ratio and Cardiovascular Risk.
Same surrogate chain, one honest framing difference (a distinction, not laundered agreement). INTERHEART calls waist/WHR «simple and crude surrogate measures for visceral obesity, which is probably the key determinant of metabolic abnormalities». (Yusuf et al., 2005) Taylor goes one step further and demotes visceral fat itself to a marker of intra-organ (hepatic/pancreatic) excess, «not pathophysiologically related to adverse metabolic consequences». So the two do not fully agree on what the pathogenic depot is — INTERHEART stops at visceral-fat-as-key- determinant, Taylor pushes past it to intra-organ fat. What they share, from independent endpoints (MI events vs glycemic/β-cell biology) and independent author lines (Yusuf’s INTERHEART team vs Taylor’s Newcastle group), is the operative claim one level up: fat distribution, not BMI, carries the risk. The hip arm is INTERHEART’s alone (Taylor says nothing about lower-body fat): a larger hip circumference is protective, which the authors themselves say may index gluteal muscle rather than fat — «Third, increasing hip measurements might also indicate increased gluteal muscle and could be a marker of overall skeletal muscle mass.» (Yusuf et al., 2005) — and its case-control design makes that arm the most exposed to reverse causation (acute-illness muscle loss). (corrected 2026-10-06: lower-body fat protective as a shared Taylor+INTERHEART claim -> INTERHEART hip-circumference only, muscle reading named; self-critique) (Taylor & Holman, 2014; inferred from Yusuf et al., 2005)
Decision relevance
- Measure central adiposity (waist/WHR), not BMI, when ranking cardiometabolic risk. The depot claim now has a hard-CV-outcome instantiation, not only a glycemic one -> Waist-to-Hip Ratio and Cardiovascular Risk. In a normal-weight person a high WHR flags elevated MI risk (INTERHEART) and is consistent with intra-organ excess; the PFT itself is not measurable from WHR. Both lenses point past the scale to fat distribution. (corrected 2026-10-06: high WHR “is above their PFT” -> consistent with; self-critique)
- Offer weight loss to normal-weight people with T2D. On the PFT hypothesis they may be above their personal threshold even at BMI <25 — a stratum weight-loss effectiveness studies have «typically excluded» — yet Taylor’s UKPDS data show their fasting glucose responds, and at a smaller required loss: in those presenting with plasma glucose 8-10 mmol/l, normoglycaemia came with ~13% loss at normal body weight vs ~21% for the whole cohort, on a 3-4 month diet-only run-in in which 16% of the cohort reached fasting plasma glucose <6.0 mmol/l (Taylor & Holman, 2014). (DiRECT’s BMI >=27 entry criterion is not in Taylor — via Total Diet Replacement and Type 2 Diabetes Remission.) (corrected 2026-10-06: glucose-band + 16% base added, DiRECT moved off the Taylor tag, PFT stated as hypothesis; self-critique) That is fasting normoglycaemia, not HbA1c-defined remission off drugs, so the lean-stratum remission dose is unmeasured here. This is the sharpest decision-change here. (corrected 2026-10-06: they respond -> fasting-glucose response, Taylor chunk 01)
- Metabolically-healthy obesity is not a free pass — and now on the newer, larger anchor. The durability call re-anchors off the dated Kramer 2013 MA onto Opio 2020 (2020 gold SR+MA, 23 cohorts, n = 4.49 M): the excess CVD risk holds even with zero metabolic risk factors (MHOW RR 1.51, MHO RR 2.18) and does not require a decade of latency (present at <10 y and >=10 y alike, subgroup-difference p = 0.98) (Opio et al., 2020). The superseded reading — Kramer’s time-latent framing, in which the MHO signal emerged only after ~10 y (RR 1.24 at >=10 y, overall null) — is refined, not deleted: Opio shows the risk is not gated on long follow-up. (Kramer et al., 2013) The magnitude is still modest in absolute terms — the only held absolute estimate remains Kramer’s ~0.7% over 10-11 y (Opio reports relative risks only) — and both are observational with high heterogeneity and live fitness confounding, so the honest framing stays a real, non-zero, but small excess, not an emergency: it licenses continued (not urgent) attention, now with less reassurance that a currently-clean metabolic panel buys durable safety -> Layer 1 - Ranking Interventions for a Stratum.
- Steer by depot / metabolic status, and read visceral fat as a marker. Liver fat (MASLD) is the clinically visible edge of the pathogenic depot, with its own dose-responsive drawdown lever -> Fatty Liver MASLD and Weight Loss. Waist / visceral fat is a useful marker of intra-organ excess, not the target itself.
- The lever is negative energy balance, whatever delivers it. Ectopic fat comes down with sustained energy deficit; composition is secondary to the deficit for the depot-drawdown itself -> What Drives Fat Gain - Energy Balance vs the Carbohydrate-Insulin Model.
- Metabolic status is the glycemic lever and a CVD-risk amplifier — but not a rival to apoB for the CVD lipid decision. The ectopic -> IR state raises cardiovascular risk partly by shifting the lipoprotein profile toward more, smaller, apoB-bearing particles (atherogenic dyslipidemia), so within that state the CVD lipid target is apoB particle number, not improve metabolic status read as a competing lever -> LDL ApoB and Cumulative Exposure. Two truths a reader conflates: metabolic status is genuinely causal for the glycemic axis (this spine), and genuinely not the CVD lipid lever (apoB is); a raised TyG / TG-HDL flags the discordance, it does not replace measuring apoB.
Limits
- The PFT is a hypothesis, proposed with a test not yet run (intralipid-infusion return of the absent first-phase insulin response); no direct per-person PFT measurement exists. Admitted directionally under the mechanism-with-human-corroboration rule — its reversal arm is what carries held RCT backing, not the threshold construct itself.
- Kramer and Opio both pool observational cohorts (high heterogeneity, I2 up to 94-99%; largely unadjusted; smoking, follow-up duration, and above all cardiorespiratory fitness as live partial confounders), and both use metabolic-syndrome criteria as a coarse proxy for the depot biology — the MAs cannot see intra-organ fat. So even Opio’s 0-risk-factor result is a modest, subgroup-derived RR on five studies with wide CIs, not a precise or causal hazard; residual fitness confounding could shrink the depot-independent excess.
- The durability dynamics are only partially cashed — a named gap
type-G. Opio settles that the excess is present without metabolic risk factors and without a 10-year latency, but it does not track what happens to metabolically-healthy obese individuals over time — the rate at which MHO transitions to metabolically-unhealthy obesity, which is the mechanism most likely to explain why “healthy” obesity is not durably safe. The dedicated transition meta-analysis is not held: — it would unlock the transition-rate / durability-dynamics arm of this call. - A stress/cortisol upstream driver is nameable but unheld. Besides energy surplus, the
telos proposes a
qol-hparoute — chronic stress -> cortisol/glucocorticoids -> visceral / central fat -> Allostatic Load and Mortality (the HPA spine). Two steps keep it mechanism-not-finding: (i) no held source in this spine measures a cortisol -> visceral-fat relationship; and (ii) the classic cortisol mechanism targets visceral fat, which this page holds is only a marker of intra-organ excess (not the pathogenic depot itself, per the depot distinction above) — so even a real cortisol -> visceral link would still be one step removed from the intra-organ fat that does the damage. A candidate bridge and an openqol-hpaacquire-gap, not a second established accumulation pathway. - Coherence, not validity (R1): the spine is internally coherent and source-faithful; no operation here grades it against a realized outcome.
(Kramer et al., 2013; inferred from Taylor & Holman, 2014; Yusuf et al., 2005)