Nucleus of the bcaa cluster. Prospective studies repeatedly find that people with higher
circulating branched-chain amino acids (BCAAs — isoleucine, leucine, valine) go on to develop type 2
diabetes — pooled at every follow-up horizon with per-BCAA OR ~2.0
(Ramzan et al., 2022) — which invites the naive read that BCAAs, and
therefore high-protein / BCAA-rich diets, are diabetogenic. This page holds the Mendelian-randomization
(MR) evidence that the causal arrow runs the
other way: genetically higher insulin resistance (IR) raises circulating BCAAs. Elevated
circulating BCAA is largely a readout of insulin resistance, not a dietary cause of it — so the
observational BCAA->diabetes signal does not license cutting dietary protein to prevent diabetes. Two
guards keep this from over-reaching: MR shows the direction (IR -> circulating BCAA), not that
dietary BCAA is harmless or beneficial; and the authors still place BCAA metabolism on the causal
pathway to T2DM as a possible downstream mediator — so BCAA is a marker-and-maybe-mediator, not an
exonerated bystander. (inferred from Wang et al., 2017)
The finding — genetically higher IR raises circulating BCAA (and inflammation)
Two-sample MR: IR (a 53-SNP instrument for the triad higher fasting insulin + higher triglycerides + lower HDL-C, from a meta-GWAS of up to 188,577 Europeans) against 58 NMR metabolites (GWAS of up to 24,925). Effect is SD-difference in each metabolite per 1-SD genetically higher IR (Wang et al., 2017):
- «One standard deviation (SD) genetically elevated insulin resistance (equivalent to 55% higher geometric mean of fasting insulin, 0.89 mmol/L higher triglycerides and 0.46 mmol/L lower HDL-C) was associated with higher concentrations of all branched-chain amino acids, isoleucine (0.56 SD; 95%CI: 0.43, 0.70), leucine (0.42 SD; 95%CI: 0.28, 0.55) and valine (0.26 SD; 95%CI: 0.12, 0.39) as well as with higher glycoprotein acetyls (an inflammation marker; 0.47 SD; 95%CI: 0.32, 0.62) (P<0.0003 for each).» (Wang et al., 2017)
- The estimate is a standardized effect (SD per SD), not a concentration change on an absolute scale, and the “1-SD IR” exposure is itself defined only by the triad above — read the CIs as the identification uncertainty and do not convert the SD figures into a clinical BCAA target.
- The BCAA effects are strong and precise; the effects on the other amino acids (alanine, glutamine, phenylalanine, tyrosine) were weaker and imprecise, so this is a BCAA-specific finding, not a whole-amino-acid-panel one.
- Robust to pleiotropy checks. «The intercepts of MR Egger were of generally small magnitude (absolute values ≤ 0.01 …) with little or no evidence that they departed from zero, providing little evidence for the presence of genetic pleiotropy» (Wang et al., 2017); estimates were concordant across IVW / weighted-median / weighted-mode and across the 53-, 28- and 12-SNP instruments. So the direction claim survives the standard MR sensitivity battery.
The summary claim: «We provide robust evidence that insulin resistance causally impacts on each individual branched-chain amino acid and inflammation» (Wang et al., 2017).
Why this defuses BCAAs / high protein cause diabetes — but only partway
The decision-relevant move is separating three distinct objects the naive read collapses: dietary BCAA intake, circulating BCAA concentration, and diabetes risk. Wang’s own discussion pulls them apart:
- Circulating BCAA is downstream of IR — the MR direction above. The prospective circulating-BCAA -> T2DM association is therefore heavily confounded/reverse-caused by the IR that raises both, exactly the pattern the reverse-causation diagnostic warns about -> The U-Shaped Association Artifact, Insulin Resistance Surrogates and Cardiovascular Risk.
- Dietary BCAA barely moves circulating BCAA, and higher dietary intake tracks a better profile. «observational studies have reported that higher dietary intake of BCAAs is associated with an improved cardiometabolic risk profile including a lower risk of T2DM (28,29). However, dietary BCAAs, both measured in absolute terms or as a percentage of total protein, are only weakly correlated with circulating concentrations of BCAAs (28,29)» (Wang et al., 2017). Those two observational findings are cited by Wang, not tested here — hold them as Wang’s reported literature (unheld sources), directional support only, not as a validated benefit of dietary BCAA. But even at that weight they break the inferential chain: if diet barely sets circulating BCAA, a circulating-BCAA signal cannot be read as a verdict on protein intake.
- The mechanism is impaired catabolism, not intake. «elevated circulating BCAA levels observed in obese and diabetic individuals could arise from impaired BCAA catabolism (11)»; after weight-loss surgery, BCAA-catabolism enzyme (BCKD) rises and BCAAs fall (Wang et al., 2017). So the lever that lowers circulating BCAA is insulin-sensitisation / fat loss, not eating less protein.
The guard against over-defusing. Wang does NOT conclude BCAAs are inert: «this implies that branched-chain amino acid metabolism lies on a causal pathway from adiposity and insulin resistance to type 2 diabetes» (Wang et al., 2017) — i.e. adiposity -> IR -> BCAA -> (possibly) T2DM, with BCAA as a candidate mediator. The BCAA -> T2DM leg rests on a separate MR (Lotta 2016) not held here, and the authors add «further studies are required to understand the exact role of BCAA metabolism in the aetiology of T2DM». So the correct reading is: circulating BCAA is a marker of IR and possibly a downstream mediator — either way the actionable lever is IR/adiposity/catabolism, not dietary protein restriction.
The observational forward association — strong, temporally consistent, and exactly what reverse causation predicts
The prospective signal the reverse-causation argument is about is now held directly, from an independent group. An SR-MA of nine studies (4313 incident-T2DM cases, 10,078 non-diabetic controls; mean BMI 28.4, i.e. overweight) pooled per-BCAA associations with later T2DM (Ramzan et al., 2022):
- «Results: The meta-analysis revealed a statistically significant positive association between BCAA concentrations and the development of T2DM, with valine OR = 2.08 (95% CI = 2.04–2.12, p < 0.00001), leucine OR = 2.25 (95% CI = 1.76–2.87, p < 0.00001) and isoleucine OR = 2.12, 95% CI = 2.00–2.25, p < 0.00001.» (Ramzan et al., 2022)
- The association held at every follow-up horizon — the paper’s novel contribution: «In addition, we demonstrated a positive consistent temporal association between circulating BCAA levels and the risk of developing T2DM with differentials in the respective follow-up times of 0–6 years, 6–12 years and ≥12 years follow-up for valine (OR = 2.08, 1.86 and 2.14, p < 0.05 each), leucine (OR = 2.10, 2.25 and 2.16, p < 0.05 each) and isoleucine (OR = 2.12, 1.90 and 2.16, p < 0.05 each) demonstrated.» (Ramzan et al., 2022) — elevated BCAA precedes diagnosis by up to ~19 years, so it is a genuinely pre-diagnostic marker, not a co-incident one.
- Direction of effect is outcome = incident T2DM (patient-important); exposure = circulating BCAA (a surrogate metabolite) — the same surrogate/target split as the IR markers -> Surrogate Outcomes. Concordant with prior pooled work Ramzan reports but the wiki does not hold (Guasch-Ferre 2016 pooled RR ~1.35–1.36 per BCAA; Sun 2019 leucine RR 1.40, valine 1.26 — unheld, directional only), though those incumbent MAs reported smaller effects; Ramzan’s larger ORs partly reflect case–control OR inflation vs cohort RR and differing per-unit scaling.
Why this does not touch the causal-direction claim above — and why the two designs are consistent, not opposed. Ramzan is careful to claim prediction, never causation: «We suggest the potential utility of BCAAs as an early biomarker for T2DM irrespective of follow-up time.» (Ramzan et al., 2022), and it explicitly credits impaired catabolism as a source of the elevated BCAA: «Recent studies have suggested that defects in the catabolic pathway of BCAAs may also be responsible for the accumulation of BCAAs in the plasma [50,51].» (Ramzan et al., 2022) — the paper does not use the words Mendelian, reverse, or causal anywhere [searched: Mendelian/reverse/causal across chunks 01-02]. A strong, temporally-consistent forward association is precisely the footprint that reverse causation leaves: if subclinical IR (present years before diagnosis) raises BCAA and also progresses to T2DM, then high baseline BCAA will predict T2DM at every horizon — which is what is observed — while causing none of it. So the two designs converge on one picture rather than clashing: BCAA is a validated pre-diagnostic predictor (Ramzan) whose causal arrow runs IR -> BCAA (Wang), i.e. a marker, not a dietary lever. The joined-issue is worked in full, with the parameter table, on Does Circulating BCAA Cause Type 2 Diabetes or Merely Predict It.
The inflammation arm — IR -> GlycA, but not GlycA -> T2DM
The same MR newly implicates IR as causal for the inflammation marker glycoprotein acetyls (GlycA; 0.47 SD): «The association that we identify of insulin resistance with GlycA is novel» (Wang et al., 2017). This is the same marker-vs-lever structure as the CRP crux on Inflammation as a Modifiable Lever: IR raises the inflammation marker, but whether inflammation is itself causal for T2DM is unshown — Wang notes recent MR studies have failed to support an inflammation -> T2DM effect, and GlycA’s own causal role can’t be probed by MR yet (too few genetic variants). So GlycA joins BCAA as a readout of the IR state, not a demonstrated onward lever for diabetes.
Decision relevance
(inferred from Wang et al., 2017) — the decision framing below is the wiki’s own reasoning from the MR direction above, not a claim Wang states.
- Do not cut protein to lower a circulating-BCAA / diabetes signal. A raised circulating BCAA in an insulin-resistant or prediabetic person is a readout of the IR state, not evidence that dietary protein is driving their diabetes risk. The protein decision stays governed by its own outcomes (muscle, satiety, kidney, mortality) -> Dietary Protein and Mortality, and by the distinct leucine-as-anabolic-trigger story -> Anabolic Resistance (which is about benefit of dietary leucine for muscle, a separate question from circulating BCAA as an IR marker).
- The lever is upstream: adiposity and insulin sensitivity. What lowers circulating BCAA in the evidence is fat loss / insulin-sensitising interventions (which restore BCAA catabolism), i.e. the same big-rock levers that address IR itself — a structural-leverage point, not a marker to chase.
- Read circulating BCAA as a marker, like the IR surrogates. It sits with TyG / HOMA-IR on the prognostic-marker side of the Surrogate Outcomes line -> Insulin Resistance Surrogates and Cardiovascular Risk: useful (if at all) to read the IR state, never a validated target to steer toward.
Limits
- MR gives direction, not magnitude-in-clinic or dietary policy. The effect is a genetic-instrument estimate of IR -> circulating metabolite on a standardized scale; it does not estimate what an intervention would do, and does not test dietary BCAA at all.
- Surrogate metabolites, not hard outcomes. The measured outcomes are circulating metabolites, not diabetes events; the BCAA -> T2DM mediator leg is borrowed (Lotta 2016, not held). — a gold SR/MA on whether lowering circulating BCAA (or dietary BCAA intake) changes T2D incidence would test the mediator/decision leg directly; none held.
- European-only; standard MR assumptions. «our analyses were conducted using European datasets which may hamper their translational relevance to non-Europeans, however risk factors for disease tend to show similar relationships across geographical regions» (Wang et al., 2017) — the authors flag the European restriction but argue against it being fatal. Summary-level, so no age/sex subgroups; the instrument was negatively associated with BMI (collider-bias caveat), which the authors argue would bias toward the null, not inflate the estimate.
- Coherence, not validity (R1). This node is internally coherent and source-faithful; no operation here grades it against a realized outcome. The loop is open.