9月

2026

Obesity and heart disease

この記事の使い方

医療上の免責事項：この記事は教育目的のものであり、医学的な助言ではありません。個別の指導については、必ずかかりつけの医師にご相談ください。.

読みやすい

Does Body Weight Cause Heart Disease? What the Evidence Actually Shows

A plain-language summary of our full evidence review

要約

Carrying excess body fat does raise the risk of heart disease. That much is now well supported — not just by studies showing the two travel together, but by genetic research and by randomized trials, which are much harder to explain away.

But the story is more interesting than “lose weight, protect your heart.” Most of the risk that comes with excess weight travels through things we can already measure and treat: 血圧, 血糖値, and the particles in your blood that carry コレステロール. And the number on the scale turns out to be one of the weaker parts of the picture. Where your body stores fat matters more than how much you weigh.

Weight is not the same thing as fat, and fat is not all the same

BMI — your weight divided by your height squared — is useful for studying large populations. For an individual person, it is a blunt instrument. It cannot tell muscle from fat, and it cannot tell you where the fat is.

That distinction matters a great deal. Fat stored just under the skin, on the hips and thighs, appears to be relatively harmless. Fat stored deep in the abdomen around the organs — called 内臓脂肪See Belly Fat for the full entry. — behaves very differently. So does fat that builds up inside organs that were never meant to store it: the liver, the pancreas, and the tissue surrounding the heart and 冠動脈.

This is why two people who weigh exactly the same can have very different risks. It is also why a slim person is not automatically safe. Someone with a normal BMI can carry significant visceral fat, have a fatty liver, or have inherited high cholesterol particle counts — and develop extensive 動脈プラーク.

A tape measure around the waist tells you more than the scale. A useful rule of thumb: your waist should measure less than half your height.

“Cholesterol” is not one number

This is probably the most important thing in the whole review, and the most commonly misunderstood.

Your standard lipid panel reports LDLコレステロール — the amount of cholesterol carried inside LDL particles. But arteries are not damaged by cholesterol floating loose. They are damaged by particles that carry cholesterol burrowing into the 動脈 wall and getting stuck there. What matters is how many of those particles are circulating.

There is a blood test that counts them: アポリポ蛋白B. Every 動脈硬化惹起性粒子 carries exactly one アポリポ蛋白B molecule, so ApoB is essentially a particle count.

Here is where excess weight comes in. Carrying a lot of visceral fat does not usually send LDL cholesterol soaring. What it does is change the particles. It tends to raise 中性脂肪, increase leftover “remnant” particles, lower HDL, and shift LDL toward smaller particles that each carry less cholesterol.

The consequence is a trap. You can end up with もっと dangerous particles while your LDL cholesterol reading looks perfectly acceptable — because each particle is carrying less cargo. Your lipid panel says fine. Your ApoB says otherwise.

This is why we think ApoB deserves to be measured far more often than it is, particularly if you have a larger waist, elevated triglycerides, 糖尿病, or a fatty liver.

So is excess weight actually causing the problem?

Largely yes, but mostly indirectly.

Genetic studies get around a stubborn problem: people who carry more weight also differ in diet, exercise, income, 喫煙, and access to care, so it is hard to know what is doing the damage. By using inherited variants that nudge body weight, researchers can approximate a natural experiment.

Those studies point the same way as everything else: excess adiposity contributes causally to heart disease, and abdominal fat specifically appears to be the driver. But when researchers ask how it does the damage, roughly two-thirds of the effect appears to run through blood pressure, blood sugar, and blood lipids. Some remains unexplained.

There is a practical lesson in that. Treating weight does not replace treating blood pressure, diabetes, cholesterol particles, or smoking. Those need to be handled directly, whatever the scale says.

Excess weight also raises the risk of 心不全そして心房細動 through routes that have little to do with arterial 歯垢 — extra work for the heart, sleep apnea, changes in the heart’s structure.

Can fitness cancel it out?

It helps enormously. People with 肥満 who are genuinely fit have far lower death rates than unfit people at any weight — in some studies, fitness is a better predictor than BMI.

But fitness does not prove your arteries are clear or your ApoB is normal. It reduces the risk substantially. It does not erase it.

What about weight loss — and the new medications?

Losing 5 to 10 percent of body weight reliably improves blood pressure, triglycerides, blood sugar, and liver fat. Larger losses generally do more.

その GLP-1 medications have genuinely changed what is possible. Semaglutide produces around 15 percent average 体重減少; tirzepatide reaches 15 to 21 percent depending on dose. More importantly, a large randomized trial called SELECT showed semaglutide reduced 心臓発作, 脳卒中, and cardiovascular deaths by about 20 percent in people with existing heart disease and excess weight but no diabetes.

Three honest caveats.

First, these drugs are not cholesterol medications. They lower triglycerides substantially but move LDL and ApoB only modestly — nothing close to what a スタチン does. In SELECT, patients stayed on their usual heart medications. The benefit was on top of standard treatment, not instead of it.

Second, the benefit depends on staying on treatment. When people stop, most of the weight and the metabolic improvements return.

Third, rapid weight loss costs muscle as well as fat. レジスタンス運動 and adequate タンパク質 matter, especially for older adults.

Weight-loss surgery has the longest track record for hard outcomes, and is associated with fewer 心臓発作 and longer survival.

Does excess weight shorten life?

In large studies, a BMI of 30 to 35 was associated with roughly 2 to 4 fewer years of life; 40 to 45 with 8 to 10 fewer. Weight gained early in life appears to matter more than the same weight arriving late, simply because the exposure lasts longer.

You may have heard of the “obesity paradox” — findings that heavier patients with established disease sometimes fare better. This mostly reflects the fact that serious illness causes weight loss, not that body fat protects anyone.

What to actually do

Measure your waist and compare it to your height. Ask for an ApoB test, especially if your triglycerides are up or your waist is large. Know your blood pressure and blood sugar. Build fitness and keep your muscle. And treat each abnormal number directly rather than assuming weight loss alone will fix everything.

This summary is general health education, not medical advice. Decisions about medication, surgery, and testing should be made with your own clinician. Our full referenced review, with all supporting studies, is available on this site.

ディープダイブ

What Is Causal, What Is Correlation, and How Much Does Weight Matter?

PEER-REVIEWED EVIDENCE REVIEW

The short answer

肥満 — especially excess visceral and ectopic fat — is not merely a marker of unhealthy behavior. The totality of prospective, mechanistic, genetic, and randomized evidence supports it as a causal contributor へ心血管疾患. But the causal chain is not simply “more body weight → more LDLコレステロール → more 歯垢.” Adipose-tissue dysfunction can increase 血圧, インスリン抵抗性, 糖尿病, 炎症, 血栓症, 睡眠時無呼吸, 腎臓病, 心房細動, and cardiac remodeling. Several of these pathways matter even when LDL コレステロール (LDL-C) is normal.[1,2]

Obesity does not invariably produce high LDL-C. Its characteristic lipid pattern is usually higher 中性脂肪, VLDL そして remnant particles, lower HDL-C, and more small cholesterol-depleted LDL particles. Because every VLDL, IDL, LDL, and リポ蛋白(a) particle carries one アポリポ蛋白 B molecule, アポリポ蛋白B estimates the total number of circulating 動脈硬化惹起性粒子. A person with visceral obesity can therefore have an LDL-C value that looks acceptable but an ApoB particle number that is unexpectedly high.[10-13]

How much does weight matter? There is no defensible universal increase in LDL-C, ApoB, or triglycerides for each five-unit increase in BMI — or for each 5 or 10 cm of waist — because BMI and waist are proxies, not direct metabolic exposures. In trials of 体重減少, triglycerides generally improve more than LDL-C. A メタ分析 of 73 randomized trials found that, at 12 months, each kilogram lost through lifestyle intervention was associated at the study level with triglycerides lower by 4.0 mg/dL, LDL-C lower by 1.28 mg/dL, and HDL-C higher by 0.46 mg/dL; the relationships differed for medication and surgery and should not be extrapolated linearly to an individual.[14]

Genetic evidence now allows a partial answer to the mediation question. In a Mendelian-randomization mediation analysis, each standard-deviation increase in genetically predicted BMI was associated with a coronary-artery-disease odds ratio of 1.49 (95% CI 1.39–1.60); adjusting simultaneously for genetically predicted 収縮期血圧, diabetes, lipid traits, and 喫煙 attenuated this to 1.14 (95% CI 1.04–1.26), implying that about 66% (95% CI 42%–91%) of the effect operates through those downstream 危険因子 — and that roughly a third does not.[57]

Modern obesity drugs help substantially. In adults without diabetes, セマグルチド produced about 15% mean weight loss at 68–104 weeks in STEP 1 and STEP 5, while チルゼパチド produced about 15% to 21% at 72 weeks across doses in SURMOUNT-1.[25,26,28] More importantly, semaglutide reduced major cardiovascular events from 8.0% to 6.5% over a mean 39.8 months in SELECT — an absolute reduction of 1.5 percentage points and a ハザード比 of 0.80 — in patients with established cardiovascular disease, overweight or obesity, and no diabetes.[30] That is direct randomized evidence of cardiovascular benefit, not proof that every GLP-1 drug prevents events in every population.

First, separate weight from fat — and fat mass from fat location

Body weight includes fat, muscle, bone, organs, and water. BMI — weight in kilograms divided by height in meters squared — is useful for population screening, but it cannot identify what the weight consists of or where fat is stored. It may label a muscular athlete as having obesity, miss an older adult with low muscle mass and excess fat, and classify a person with substantial abdominal fat as “normal weight.”

Waist circumference adds information because it partly captures abdominal and 内臓脂肪See Belly Fat for the full entry.. A common clinical action threshold for people of European ancestry is at least 102 cm (40 inches) in men or 88 cm (35 inches) in women; thresholds near 90 cm for men and 80 cm for women are often used for several Asian populations. These are screening cut points, not biological cliffs, and optimal values vary with sex, ethnicity, age, and BMI.[3] A waist-to-height ratio above about 0.5 is a practical warning signal, but it too is a screening rule rather than a diagnosis.[56]

Subcutaneous fat lies under the skin. When this tissue can expand while remaining relatively インスリン sensitive, it can serve as a safer storage compartment. Visceral fat surrounds abdominal organs and releases fatty acids and signaling molecules into the portal circulation. Ectopic fat accumulates in organs and tissues not designed primarily for energy storage — including liver, pancreas, skeletal muscle, epicardium, and perivascular spaces. Visceral and ectopic fat are more strongly linked to insulin resistance and cardiovascular disease than total fat mass alone.[3,4]

How to measure adiposity in practice

Measure	Main strength	Main limitation
BMI	Inexpensive, standardized, strongly related to outcomes at the population level	Cannot separate fat from muscle or locate fat
Waist circumference	Simple proxy for abdominal/visceral adiposity; adds information to BMI	Technique and ethnicity-specific cut points matter
Waist-to-height ratio	Adjusts waist for body size; easy screening rule	No single threshold perfectly fits every population
Waist-to-hip ratio	Captures body shape and predicted 心筋梗塞 better than BMI in INTERHEART	Hip circumference partly reflects muscle and frame; technique varies[15]
Body-fat percentage	Distinguishes fat from 除脂肪体重 better than BMI	Accuracy depends on method and hydration; location remains uncertain
DEXA	Quantifies total/regional fat, lean mass, and bone	Limited direct resolution of organ and visceral fat; radiation is low but nonzero
CT or MRI	Best clinical-research measures of visceral and ectopic fat	Cost, access, and, for CT, radiation prevent routine screening
Liver-fat MRI or spectroscopy	Sensitive measure of hepatic steatosis	Specialized; not a general coronary-risk test
Fitness testing	Captures functional cardiovascular reserve and modifies prognosis	Does not measure plaque, ApoB, or fat distribution
Muscle mass and strength	Identifies frailty and sarcopenic obesity	Requires additional testing; mass and strength are not interchangeable

Does obesity cause cardiovascular disease?

What each kind of evidence can — and cannot — show

アン association means two findings occur together. A 用量反応 association means risk generally increases as exposure increases. Biological plausibility means credible mechanisms exist. Mediation asks how much of an association passes through measured intermediate factors. An independent statistical association means a relationship remains after variables in a particular model are adjusted; it does not prove independence from all 交絡因子または因果関係. メンデルランダム化 uses genetic variants associated with an exposure as instruments and can strengthen causal inference, but it depends on assumptions about pleiotropy and instrument validity. Randomization best tests whether an intervention changes outcomes, although a drug may have effects beyond weight loss.

Prospective evidence establishes temporality and a graded relationship. In analyses restricted to never-smokers without chronic disease, mortality was lowest around BMI 20–25 kg/m² and rose by 31% for each 5 kg/m² above 25; coronary mortality rose by 49% and 脳卒中 mortality by 38% per 5 kg/m².[5] These are associations, albeit carefully controlled ones. They do not reveal whether fat itself, its metabolic consequences, or correlated exposures produced every event.

A pooled analysis of 97 前向きコホート containing 1.8 million people estimated that blood pressure, cholesterol, and グルコース together explained about 46% of the excess coronary-heart-disease risk and 76% of the excess stroke risk associated with high BMI.[8] That is mediation analysis, not a precise decomposition of causality. Single measurements, measurement error, treatment during follow-up, and unmeasured lifetime exposure leave substantial residual uncertainty.

Genetic evidence moves the inference further, and it now does so in two complementary ways. First, on the question of whether adiposity is causal: Mendelian-randomization analyses have linked genetically influenced adiposity to coronary disease and have found a residual association after accounting for measured blood pressure, 脂質異常症, and glycemic traits.[9] Abdominal adiposity specifically appears causal — a polygenic score for waist-to-hip ratio adjusted for BMI, constructed from 48 variants, was associated with an odds ratio of about 1.46 for 冠動脈疾患 per standard deviation, along with higher triglycerides, lower HDL-C, higher glucose, and higher blood pressure.[58] This is genetic support for the clinical intuition that どこ fat is stored matters, not merely how much of it there is.

Second, on the question of how much operates through conventional risk factors: an MR mediation analysis using genetic data from 140,595 to 898,130 participants estimated a CAD odds ratio of 1.49 (95% CI 1.39–1.60) per standard deviation of genetically predicted BMI. Adjusting for genetically predicted systolic blood pressure alone attenuated this to 1.34 (27% mediated); for diabetes alone, to 1.27 (41% mediated); for lipid traits alone, to 1.47 (3% mediated); and for smoking alone, to 1.46 (6% mediated). Adjusting for all mediators together gave 1.14 (95% CI 1.04–1.26), or 66% mediated (95% CI 42%–91%).[57] Two cautions are essential in reading these numbers. The lipid mediators in that analysis were genetically predicted LDL-C, HDL-C, and triglycerides — not ApoB — so the 3% estimate should not be interpreted as an estimate of ApoB-specific mediation. And the wide 信頼区間 around 66% means the honest summary is “a substantial share, possibly a majority, but not all.”

Genetic instruments approximate lifelong exposure and are not randomized weight-loss trials; nevertheless, their direction agrees with long-term cohort and mechanistic evidence.

Randomized outcome evidence used to be the missing piece. Intensive lifestyle treatment in Look AHEAD improved weight, fitness, glycemia, blood pressure, and several lipid measures in adults with type 2 diabetes but did not significantly reduce its primary cardiovascular-event outcome over a median 9.6 years.[38] That neutral result does not prove adiposity harmless: achieved weight separation narrowed over time, risk-factor treatment was intensive in both groups, and the intervention changed many variables modestly. A post-hoc analysis of the same trial found that participants who lost at least 10% of body weight in the first year had a 21% lower risk of the primary outcome (adjusted HR 0.79, 95% CI 0.64–0.98) than those with stable weight or weight gain — hypothesis-generating rather than confirmatory, because post-hoc weight-change groups are not randomized comparisons and people who lose weight successfully differ from those who do not.[59] In contrast, SELECT demonstrated fewer major cardiovascular events with semaglutide in people with overweight or obesity and established cardiovascular disease but no diabetes.[30] Metabolic-bariatric surgery is associated with fewer cardiovascular events and deaths, but most event evidence is from prospective controlled or matched observational studies rather than blinded randomized trials.[41-43]

Causal conclusion: the converging evidence supports excess adiposity — most strongly visceral and ectopic adiposity — as a causal contributor to cardiovascular disease. Its cardiovascular effect appears substantially, and perhaps mostly, mediated by downstream factors such as blood pressure, diabetes/glycemia, conventional lipid traits, and smoking; the available mediation studies did not measure ApoB-specific mediation and leave a residual effect. Additional risk is expressed through heart-failure, arrhythmia, respiratory, renal, and structural pathways.[1,2,57]

What can distort the association?

Smoking lowers body weight while raising mortality. Chronic illness, cancer, frailty, and preclinical disease can cause unintentional weight loss. Socioeconomic conditions influence diet, activity, stress, sleep, and access to care. Physical activity and 心肺持久力 are often measured poorly. Conditioning analyses on having established disease can create selection or collider bias. BMI misses fat distribution and muscle. Short follow-up ignores decades of earlier exposure. These issues can exaggerate, attenuate, or even reverse observed associations; none is solved merely by adding variables to a regression model.

Does obesity increase “cholesterol”?

“Cholesterol” is not one exposure. 総コレステロール is the cholesterol carried in all リポタンパク質. LDL-C is the cholesterol mass inside LDL particles. Non-HDL-C is total cholesterol minus HDL-C and captures cholesterol carried in all アポB含有リポ蛋白. Remnant cholesterol estimates cholesterol in triglyceride-rich remnant particles. ApoB approximates the number of atherogenic particles.

Obesity often raises total cholesterol, LDL-C, non-HDL-C, remnant cholesterol, triglycerides, and ApoB on average — but not equally, and not in everyone. The most reproducible pattern in visceral obesity and insulin resistance is:

higher fasting and 食後の triglycerides;
more hepatic VLDL production and more remnant particles;
lower HDL-C;
remodeling toward smaller, cholesterol-depleted LDL particles; and
ApoB that is higher than the LDL-C value would predict.[1,2,13]

This starts when insulin-resistant adipose tissue releases more 遊離脂肪酸. The liver receives this substrate, accumulates fat, and exports more triglyceride in VLDL. Hepatic insulin resistance also fails to restrain VLDL production. Clearance of triglyceride-rich lipoproteins can be delayed. コレステリルエステル転送タンパク質 exchanges triglyceride from VLDL into LDL and HDL; 肝リパーゼ then remodels these triglyceride-enriched particles into smaller LDL and HDL particles. The result can be many LDL particles that each contain less cholesterol. LDL-C may therefore be normal or only mildly elevated while ApoB and non-HDL-C reveal a greater 粒子負荷.[10-13]

A second, often-overlooked source of atherogenic particles in obesity is the intestine. After meals, enterocytes assemble カイロミクロン around apolipoprotein B-48. In insulin resistance, both intestinal アポB-48 particle production and clearance are disturbed, adding apoB-48–containing remnant particles that a standard fasting lipid panel does not separately quantify and may underrepresent because it misses the postprandial period. This matters clinically because it is one of the specific abnormalities that incretin therapy appears to correct (discussed below).[72-74]

これは 不一致: LDL-C and ApoB point to different levels of risk. LDL-C answers “how much cholesterol is inside LDL?” ApoB more closely answers “how many atherogenic particles are circulating?” Because particle entry and retention in the arterial wall initiate 動脈硬化, ApoB-containing lipoproteins are causal, supported by concordant randomized, genetic, and epidemiologic evidence.[10-12] The National Lipid Association’s 2024 expert clinical consensus concluded that when LDL-C and apoB are discordant, cardiovascular risk aligns better with apoB, and that apoB should then be the therapeutic target.[61]

How large are the lipid changes?

There is no reliable universal slope for lipid change per 5 BMI units or per 5–10 cm of waist. Cross-sectional slopes combine age, sex, ethnicity, diet, アルコール, genes, diabetes, liver fat, lipid-lowering treatment, and duration of adiposity. They are not equivalent to the effect of gaining or losing that amount of fat. Waist also measures both subcutaneous and visceral tissue, and the same waist can represent different visceral-fat volumes.

The best quantitative synthesis for treatment is a meta-analysis of 73 randomized trials (32,496 participants). At 12 months, its study-level meta-regression estimated the following change per kilogram of weight lost:[14]

Weight-loss method	トリグリセリド	LDL-C	HDLコレステロール	解釈
Lifestyle	−4.0 mg/dL (95% CI −5.24 to −2.77)	−1.28 mg/dL (−2.19 to −0.37)	+0.46 mg/dL (+0.20 to +0.71)	Stronger average TG than LDL response[14]
Pharmacotherapy studied through 2018	−1.25 mg/dL (−2.94 to +0.43)	−1.67 mg/dL (−2.28 to −1.06)	+0.37 mg/dL (+0.23 to +0.52)	TG estimate crossed zero; predates semaglutide 2.4 mg and tirzepatide[14]
Bariatric surgery	−2.47 mg/dL (−3.14 to −1.80)	−0.33 mg/dL (−0.77 to +0.10)	+0.42 mg/dL (+0.37 to +0.47)	LDL estimate crossed zero; procedure and baseline lipids matter[14]

These coefficients are pooled trial-level associations, not instructions to multiply by 20 after a 20-kg loss. Lipid responses are not necessarily linear, and medication changes can obscure them. The synthesis did not provide comparable universal coefficients for total cholesterol, non-HDL-C, remnant cholesterol, or ApoB.[14]

For 5%, 10%, 15%, and 20% weight loss, the scientifically honest summary is graded rather than falsely precise. Around 5% commonly improves triglycerides, glucose, liver fat, and blood pressure; 10% usually produces larger metabolic effects; losses near 15%–20% can transform diabetes, sleep apnea, fatty liver, and heart-failure symptoms in selected patients. Yet LDL-C and ApoB may fall little, substantially, or — on a saturated-fat-rich diet — even rise. STEP and SURMOUNT show that these larger weight-loss ranges are achievable with インクレチン therapy, not that a given percentage loss guarantees a fixed ApoB response.[25,26,28,53]

Summary table: exposures, interventions, and expected effects

Exposure or intervention	Expected lipid/ApoB effect	Cardiovascular effect	Evidence type	自信
Higher BMI	On average higher TG/non-HDL/remnants and lower HDL; LDL-C and ApoB response heterogeneous	Dose-related higher ASCVD, HF, AF, and mortality	Large prospective cohorts, MR, mechanisms	High for risk; moderate for any specific lipid increment
Larger waist/visceral fat	Stronger insulin-resistant pattern and ApoB–LDL-C discordance than BMI alone	Higher events and プラーク負荷 after multivariable adjustment; MR supports causality for abdominal adiposity (CHD OR ≈1.46 per SD WHRadjBMI)[58]	Cohorts, imaging studies, Mendelian randomization	Moderate–high; exact independent effect uncertain
5% weight loss	Usually meaningful TG, glycemic, BP, and liver-fat improvement; LDL-C/ApoB may change only modestly[14,39]	Risk factors improve; event reduction not established for 5% itself	RCTs and consensus syntheses	High for risk factors; low for event attribution
10% weight loss	Usually larger TG/non-HDL and metabolic improvement; ApoB may remain above target[14,39]	Post-hoc Look AHEAD: ≥10% loss associated with HR 0.79 for CV events[59]	RCT post-hoc analyses and cohorts	中程度
15%–20% weight loss	Often substantial TG, remnant, glycemic, BP, waist, and inflammatory improvement; LDL-C/ApoB still variable[25,26,28,53]	SELECT proves fewer MACE with semaglutide in 二次予防; cannot assign all benefit to weight	RCTs	High for semaglutide in SELECT population; moderate for generalization
Exercise with little weight loss	Can reduce waist/visceral fat, TG, BP, and insulin resistance; LDL-C often changes little	Higher fitness strongly associated with lower mortality; outcome trials isolate exercise imperfectly	RCT risk-factor studies; prospective fitness cohorts	High for fitness/risk factors; moderate for events
Metabolic-bariatric surgery	TG usually falls and HDL-C rises; LDL-C/ApoB response depends on procedure and medication[14]	Lower events and mortality in controlled observational evidence	Prospective controlled and matched cohorts; meta-analysis	Moderate for events; 交絡 remains
Semaglutide 2.4 mg	TG, non-HDL, BP, glycemia, waist, and inflammation improve; LDL-C/ApoB reductions are not statin-like	SELECT HR 0.80 for MACE; 6.5% vs 8.0% over 39.8 months[30]	Large blinded CV-outcomes RCT	High for the SELECT population
GLP-1 RAs in type 2 diabetes (class)	Modest LDL-C change; TG, remnants, postprandial apoB-48 improve	Pooled MACE HR 0.86; 全因死亡率 0.88; HF hospitalization 0.89[33]	Meta-analysis of 8 CVOTs	High for T2D populations
マンジャロ	Large weight/waist and metabolic improvements; TG and remnant measures generally improve more than LDL-C	Noninferior to dulaglutide for MACE in SURPASS-CVOT (HR 0.92); HF events improved in obesity-related HFpEF	Weight-loss RCTs; active-comparator CVOT; HF-outcomes RCT	High for weight and HFpEF outcome; incomplete for placebo-referenced ASCVD benefit
Retatrutide (triple agonist)	Triglyceride and non-HDL-C improvements were reported in phase 2; fasting ApoB and cardiovascular outcomes remain insufficiently characterized[75]	No peer-reviewed cardiovascular-outcome evidence	Phase 2 randomized trial[75]	Moderate for short-term weight loss; low for cardiovascular outcomes

If ApoB is the same, does obesity still add coronary risk?

Suppose two people truly have the same ApoB level and the same cumulative lifetime exposure to ApoB-containing particles, but one has obesity. The person with obesity probably still has higher overall cardiovascular risk if obesity has produced 高血圧, diabetes, kidney disease, sleep apnea, inflammation, inactivity, thrombosis, or adverse cardiac remodeling. Obesity also has a particularly strong relationship with atrial fibrillation and 心不全, including HF with preserved 駆出率; a meta-analysis estimated incident HF risk 42% higher per 5 kg/m² of BMI and 28% higher per 10 cm of waist.[24]

For coronary atherosclerosis specifically, the size of any residual effect cannot be quantified confidently. “Same current ApoB” is not “same lifetime ApoB”: one blood draw misses earlier VLDL/remnant and LDL exposure, postprandial particles, lipoprotein(a), treatment history, and duration. In addition, adjustment for ApoB, blood pressure, diabetes, smoking, and activity cannot remove measurement error or unmeasured confounding. Together, the studies suggest that a substantial share — possibly a majority — of obesity’s coronary association operates through measured downstream factors, but the exact proportion is uncertain and neither estimate is ApoB-specific.[8,57] The observational analysis of 97 cohorts estimated roughly 46% using blood pressure, total/HDL cholesterol, and glucose; the Mendelian-randomization analysis estimated about 66% (95% CI 42%–91%) using blood pressure, diabetes liability, conventional lipid traits, and smoking. Readers should treat any single percentage attributed specifically to ApoB as unsupported by these data.

The practical answer is therefore not to choose between weight and ApoB. Treat adiposity and its complications そして lower ApoB to an appropriate risk-based target. Weight loss does not erase the need for スタチン or other lipid-lowering therapy when absolute atherosclerotic risk and ApoB remain high.

Obesity, visceral fat, and plaque

Visceral fat predicts events and 無症候性動脈硬化症 more strongly than BMI in many cohorts. In the フラミンガム心臓研究, each standard-deviation increase in visceral adipose tissue was associated with a 44% higher hazard of incident cardiovascular disease after multivariable adjustment.[16] In a coronary CT study of 427 patients, each standard-deviation increase in visceral fat was associated with 68% higher odds of 非石灰化プラーク presence and 31% higher odds of greater noncalcified-plaque extent; high-risk plaque features were also more frequent.[17] These are adjusted observational associations, not experiments.

Longitudinal imaging adds temporality but not complete causal identification. In an observational cohort of 862 adults, visceral fat predicted coronary-calcium progression, and modeling suggested that adipose-tissue dysfunction/insulin resistance mediated about 51.8% of the association.[18]

Discordance between ApoB and LDL-C is also visible in imaging. In the CARDIA cohort, young adults with high ApoB but low LDL-C had a higher likelihood of coronary calcium 25 years later than those with concordantly low values, indicating that particle number carries information the cholesterol concentration alone does not.[19] The evidence is not uniformly strong, however: in the Multi-Ethnic Study of Atherosclerosis, apoB was associated with coronary calcium prevalence, incidence, and progression, but apoB discordance added only modest predictive value beyond LDL-C or non-HDL-C.[60] The fair reading is that discordance identifies people whose risk is underestimated by LDL-C, while acknowledging that the incremental imaging signal in some cohorts is small.

Epicardial and perivascular fat may exert local inflammatory and paracrine effects on 冠動脈, and epicardial-fat volume is associated with coronary disease in imaging meta-analysis.[4,54] Yet no adjustment model can prove a direct local effect independent of lifetime ApoB, smoking, blood pressure, or glycemia. Measuring lipids once makes “independence from cholesterol” especially uncertain.

Intentional weight loss plausibly reduces plaque-driving exposures, but direct imaging evidence is limited. Small imaging studies can measure coronary calcium, total plaque, noncalcified plaque, or high-risk features; these outcomes are not interchangeable. A rising CAC score does not automatically mean treatment failure: calcium can increase as plaque becomes more calcified and potentially more stable, while lipid-rich or noncalcified components shrink. Clinical events matter more than serial calcium change.

No large randomized trial has established that lifestyle, semaglutide, tirzepatide, or surgery causes coronary プラーク退縮 and that regression mediates fewer events. SELECT shows event reduction without needing to demonstrate plaque regression.[30] Statements that GLP-1 drugs “clean out arteries” go beyond the evidence.

Ten adiposity phenotypes — and why equal BMI does not mean equal risk

Phenotype	Typical measurements	Lipid/ApoB pattern	Cardiovascular risk	Important limitations
General obesity by BMI	BMI ≥30 kg/m², with class I 30–34.9, II 35–39.9, III ≥40[1,2]	Variable; often TG-rich dyslipidemia	Higher average ASCVD, HF, AF, diabetes, and mortality	Misclassifies muscular and sarcopenic people; ignores location
Central/abdominal obesity	Waist, waist-to-height, waist-to-hip	More TG/remnants, lower HDL, frequent ApoB discordance	Often predicts MI and mortality beyond BMI; MR supports causality[58]	Cut points and measurement sites vary[3,15]
Visceral obesity	CT/MRI visceral volume; waist as proxy	Strong insulin-resistant, VLDL-rich pattern	Higher diabetes, plaque, CVD risk	Imaging not routinely indicated; association is not perfect[4,16]
Predominantly subcutaneous obesity	Imaging or body-shape phenotype	Can be relatively favorable if adipose storage remains functional	Lower risk than visceral obesity at the same fat mass, not zero risk	“Safer” does not mean benign; compartment can become dysfunctional
Ectopic fat	Liver MRI/ultrasound; cardiac or vascular imaging	Fatty liver commonly tracks insulin resistance, TG, remnants, ApoB	Liver, epicardial, and perivascular fat identify higher-risk biology	Organ fat is partly a marker of systemic energy surplus[4]
Normal-weight central obesity	BMI <25 with high waist or waist-to-hip ratio	May show insulin resistance, TG/remnant elevation, or normal routine lipids	Higher mortality and CVD than normal weight without central fat	Definitions differ; 逆因果関係 can influence older cohorts[21]
“Metabolically healthy” obesity	Obesity without selected BP, glucose, and lipid abnormalities	May have normal measured LDL-C/TG; ApoB can still be discordant	Lower than metabolically unhealthy obesity, but generally higher than healthy normal weight	Definitions vary; status often converts over time; not reliably benign[20]
Sarcopenic obesity	High fat plus low muscle strength/mass	Metabolic pattern variable	Combines cardiometabolic risk with frailty, falls, and disability	BMI may look only mildly high; diagnostic methods vary[22]
Obesity with high fitness	Exercise testing/estimated cardiorespiratory fitness plus adiposity measures	Fitness may improve TG, インスリン感受性, and BP without normalizing ApoB	Much lower mortality than unfit obesity; risk may approach fit normal-weight groups in some cohorts	Observational fitness data; fitness does not exclude plaque or diabetes[23]
Obesity with diabetes or MASLD	A1c/glucose; liver enzymes and imaging; waist	TG/remnant-rich, low HDL, frequent ApoB discordance	Particularly high ASCVD, HF, kidney, and liver risk	LDL-C can look deceptively ordinary; multiple conditions require direct treatment

“Metabolically healthy obesity” is best understood as a lower-risk, often temporary phenotype — not immunity. Definitions typically use a handful of measurements at one time, and many people transition to metabolic abnormalities during follow-up. Meta-analysis finds cardiovascular risk above metabolically healthy normal weight, although below metabolically unhealthy obesity.[20]

Fitness matters enormously. A 2024 システマティックレビュー and meta-analysis found that cardiorespiratory fitness was a stronger mortality discriminator than BMI categories: fit adults with obesity often had much lower mortality than unfit adults of any weight.[23] But fitness does not demonstrate normal ApoB, absence of plaque, or elimination of obesity-related HF, AF, sleep-apnea, and orthopedic risks. It mitigates risk; it should not be used to declare untreated risk factors irrelevant.

What intentional weight loss changes

Diet and exercise

Calorie reduction drives weight loss regardless of dietary label, while food composition changes lipids independently of weight. In DIETFITS, healthy low-fat and healthy low-carbohydrate diets produced similar mean 12-month weight loss, with wide individual variation.[55] Lower-carbohydrate patterns tend to reduce triglycerides and raise HDL-C more, while lower-fat patterns tend to reduce LDL-C more; 飽和脂肪酸, ファイバー, alcohol, refined 炭水化物, and baseline insulin resistance strongly modify the result.[14,49] Mediterranean and plant-predominant patterns have cardiovascular advantages that cannot be reduced to the number on the scale because they alter dietary fat quality, fiber, sodium, and food processing.

Exercise can reduce waist and visceral fat and improve fitness, blood pressure, glycemia, triglycerides, and 血管内皮機能 even when scale weight changes little.[3] Resistance exercise is especially valuable during weight loss because preserving muscle protects strength, insulin disposal, function, and resting energy expenditure.

The Look AHEAD result supplies an important safeguard: better weight and risk factors do not guarantee a statistically significant reduction in cardiovascular events in every trial.[38] Its post-hoc finding that participants achieving ≥10% weight loss had a 21% lower event risk suggests the magnitude of loss may matter more than the assignment to an intervention — but that comparison is not randomized and cannot exclude confounding by health status, 固守, and motivation.[59] Conversely, a 2017 systematic review of randomized weight-loss interventions found a 15% lower all-cause mortality risk (risk ratio 0.85) despite limited event counts and heterogeneous programs.[40] Lifestyle remains foundational, but event claims should match the population, intensity, durability, and follow-up studied.

Metabolic-bariatric surgery

Surgery usually produces the largest and most durable average weight loss, with major improvements in diabetes, triglycerides, HDL-C, blood pressure, sleep apnea, and fatty liver. The Swedish Obese Subjects prospective controlled study associated surgery with fewer cardiovascular events and lower mortality.[41,42] A 2021 individual-level meta-analysis of 174,772 participants — mostly matched cohorts plus one prospective controlled study — associated surgery with a 49.2% lower mortality hazard and an estimated 6.1-year longer median life expectancy than nonsurgical care.[43] The estimated gain was 9.3 years with diabetes and 5.1 years without diabetes.[43]

Those are striking associations, not fully randomized estimates. Patient selection, healthcare engagement, and residual confounding may contribute. Procedures also differ: malabsorptive operations may lower LDL-C more than purely restrictive operations, and surgery carries perioperative and long-term nutritional risks. Even after major weight loss, ApoB should be remeasured rather than assumed normal.

Do GLP-1 and dual-incretin drugs help — and how, and how much?

This is the question patients ask most often, so it deserves a direct answer before the detail: yes, they help — substantially and in ways that go beyond the number on the scale — but they do not replace lipid-lowering therapy, and the benefit depends on continued treatment.

GLP-1受容体作動薬 reduce appetite and energy intake, slow gastric emptying most prominently early in treatment, and improve glucose-dependent insulin secretion. Tirzepatide activates both GIP and GLP-1 receptors. Their cardiovascular effects plausibly combine less visceral and liver fat, better glycemia, lower blood pressure, lower 食後高脂血症 and triglyceride-rich particles, improved kidney-related risk, and reduced systemic inflammation. Direct vascular or cardiac effects remain biologically plausible but incompletely resolved; weight loss alone cannot be cleanly separated from drug-specific effects in outcome trials.[32-34]

How much weight do they reduce?

Trial and population	Treatment	Mean weight result	What it proves
STEP 1; obesity/overweight, no diabetes	Semaglutide 2.4 mg, 68 weeks	−14.9% vs −2.4% プラセボ[25]	Large randomized weight loss plus lifestyle counseling
STEP 5; obesity/overweight, no diabetes	Semaglutide 2.4 mg, 104 weeks	−15.2% vs −2.6% placebo[26]	Weight loss sustained during continued treatment
SURMOUNT-1; obesity/overweight, no diabetes	Tirzepatide 5, 10, or 15 mg, 72 weeks	−15.0%, −19.5%, and −20.9% vs −3.1% placebo[28]	Dose-related large randomized weight loss
SELECT; established CVD, no diabetes	Semaglutide 2.4 mg	At 208 weeks, −10.2% vs −1.5% bodyweight[31]; at 20 weeks, −6.4% vs −0.8% with waist −5.0 vs −1.1 cm[78]	Sustained long-term separation in a CV-outcomes population
SURPASS-CVOT; T2D with ASCVD	Tirzepatide vs dulaglutide 1.5 mg	−11.6% vs −4.5%[70]	Greater weight and metabolic improvement than an active comparator

Meta-analysis of randomized trials of once-weekly semaglutide in people without diabetes confirms that these weight reductions are sustained across trials rather than being a single-trial result.[50] Treatment usually must continue to maintain most of the effect. In the STEP 1 extension, participants regained 11.6 percentage points of body weight during the year after semaglutide withdrawal — about two-thirds of the prior loss — and cardiometabolic markers drifted toward baseline.[27] In SURMOUNT-4, after an initial 20.9% loss during open-label tirzepatide, those continuing drug lost another 5.5% from week 36 to 88, whereas those switched to placebo regained 14.0%.[29] These trials support chronic-disease treatment, not a short “course.” The corollary for cardiovascular protection is important and under-discussed: because the risk-factor improvements reverse when the drug stops, there is no basis for assuming that a limited course of incretin therapy confers lasting event reduction.

Do they prevent cardiovascular events?

In obesity without diabetes. SELECT enrolled 17,604 people aged at least 45 years with BMI at least 27 kg/m², established cardiovascular disease, and no diabetes. Over a mean 39.8 months, cardiovascular death, nonfatal myocardial infarction, or nonfatal stroke occurred in 6.5% on semaglutide and 8.0% on placebo (hazard ratio 0.80, 95% CI 0.72–0.90).[30] The absolute reduction was 1.5 percentage points — about 67 people treated for the trial’s average duration to prevent one primary event, calculated from the reported event rates.[30] This is the clearest randomized answer that obesity pharmacotherapy can improve major cardiovascular outcomes in a defined high-risk population.

In type 2 diabetes. The GLP-1 cardiovascular-outcomes trial program is large and reasonably consistent. Individual trial results for the three-point MACE endpoint (cardiovascular death, nonfatal myocardial infarction, nonfatal stroke) are shown below.

Trial (year)	Agent	MACE hazard ratio (95% CI)	Superior to placebo?
ELIXA (2015)	Lixisenatide	1.02 (0.89–1.17)*	いいえ
LEADER (2016)	Liraglutide	0.87 (0.78–0.97)[62]	はい
SUSTAIN-6 (2016)	Semaglutide (injectable)	0.74 (0.58–0.95)[63]	Nominally yes; trial designed for noninferiority
EXSCEL (2017)	Exenatide once-weekly	0.91 (0.83–1.00)[66]	No (P=0.06)
HARMONY Outcomes (2018)	Albiglutide	0.78 (0.68–0.90)[68]	はい
REWIND (2019)	Dulaglutide	0.88 (0.79–0.99)[64]	はい
PIONEER 6 (2019)	Oral semaglutide	0.79 (0.57–1.11)[65]	Noninferior only
AMPLITUDE-O (2021)	Efpeglenatide	0.73 (0.58–0.92)[69]	はい
SELECT (2023)	Semaglutide 2.4 mg (no diabetes)	0.80 (0.72–0.90)[30]	はい

*ELIXA used a four-point MACE that additionally included hospitalization for 不安定狭心症.[67]

Across eight earlier GLP-1 cardiovascular-outcome trials in type 2 diabetes (60,080 participants), meta-analysis found 14% fewer 主要心血管イベント (hazard ratio 0.86), 12% lower all-cause mortality (0.88), and 11% fewer heart-failure hospitalizations (0.89).[33] Most agents and doses in that analysis were diabetes treatments, so it should not be treated as an estimate for obesity treatment in people without diabetes. Individual components are informative: in LEADER, cardiovascular death fell (HR 0.78) and all-cause mortality fell (HR 0.85);[62] in SUSTAIN-6, nonfatal stroke was reduced (HR 0.61);[63] in PIONEER 6, cardiovascular death was lower (HR 0.49) though the trial was designed only to exclude harm.[65]

Tirzepatide. SURPASS-CVOT randomized 13,165 adults with type 2 diabetes and atherosclerotic cardiovascular disease to tirzepatide or dulaglutide 1.5 mg — an active comparator already proven to reduce events. Over a median treatment duration of about 47 months, MACE occurred in 12% versus 13% (hazard ratio 0.92), meeting the prespecified noninferiority criterion; formal superiority was not established.[70] Tirzepatide also produced greater reductions in HbA1c, weight, and several cardiovascular バイオマーカー.[70] In a separate post-hoc analysis, an expanded six-component cardiorenal composite (all-cause mortality, myocardial infarction, stroke, coronary 血行再建術, heart failure, and adverse kidney outcomes) favored tirzepatide, occurring in 23.7% versus 27.4% (HR 0.84, 95% CI 0.79–0.90).[77] The interpretive point matters: because there was no placebo arm, SURPASS-CVOT establishes that tirzepatide is at least as cardioprotective as a drug known to reduce events, not the absolute magnitude of its benefit against placebo. A placebo-controlled obesity outcomes trial (SURMOUNT-MMO) had not reported at the time of writing.

Heart failure and kidney disease. For obesity-related HFpEF, semaglutide improved symptoms, physical limitations, walking distance, inflammation, and weight in STEP-HFpEF; weight fell 13.3% versus 2.6% with placebo at 52 weeks.[35] Benefits were also observed in participants with type 2 diabetes.[36] In SUMMIT, 731 patients with obesity-related HFpEF received tirzepatide or placebo for a median 104 weeks. Cardiovascular death or worsening heart failure occurred in 9.9% versus 15.3% (hazard ratio 0.62), driven primarily by fewer worsening-HF events; cardiovascular-death counts were small.[37] In FLOW, 3,533 people with type 2 diabetes and 慢性腎臓病 received semaglutide 1.0 mg or placebo; the primary composite of kidney failure, a sustained ≥50% decline in eGFR, or death from kidney-related or cardiovascular causes fell 24% (HR 0.76, 95% CI 0.66–0.88). Major cardiovascular events fell 18% (HR 0.82, 95% CI 0.68–0.98), and all-cause death fell 20% (HR 0.80, 95% CI 0.67–0.95).[71] These results address HFpEF and kidney disease — not coronary-plaque regression or general 一次予防.

Why do they work? Weight-dependent and weight-independent mechanisms

The benefit is best understood as several modest effects acting together rather than one dominant mechanism.

Weight-dependent effects. Loss of visceral and hepatic fat improves insulin sensitivity and reduces hepatic VLDL secretion, lowering triglyceride and remnant particle burden. Waist circumference falls markedly (−7.7 cm in SELECT at 208 weeks).[31] Blood pressure, glycemia, and sleep-disordered breathing improve.

Partly weight-independent effects. Three deserve emphasis:

Reduced intestinal atherogenic-particle production. Exenatide acutely inhibits intestinal lipoprotein production in healthy humans, an effect demonstrable before meaningful weight change.[72] In patients with type 2 diabetes, liraglutide reduced postprandial hyperlipidemia both by increasing apoB-48 catabolism and by reducing apoB-48 production.[73] Across the class, exenatide, liraglutide, lixisenatide, semaglutide, and dulaglutide have each been shown to lower postprandial apoB-48 in human studies.[74] Because cholesterol-enriched apoB-48 remnant particles — not nascent chylomicrons themselves — are atherogenic, and because a standard fasting panel does not separately quantify them, this is a plausible route by which incretins reduce atherogenic particle exposure without dramatically changing fasting LDL-C.
Anti-inflammatory effects. In SELECT, high-sensitivity C反応性タンパク質 was 37.8% lower with semaglutide than placebo at 104 weeks, with reductions apparent early and across baseline hsCRP categories.[32] In earlier semaglutide trials in type 2 diabetes, exploratory mediation analyses estimated that 20.6%–61.8% of the hsCRP reduction was statistically mediated by changes in HbA1c and body weight; a substantial remainder was not explained by those two variables.[76]
Benefit not fully explained by measured weight change. In a prespecified SELECT analysis, cardiovascular benefit was consistent across baseline adiposity categories; weight loss at week 20 was not linearly associated with subsequent MACE, whereas waist reduction showed a modest association and accounted for an estimated 33% of benefit in mediation modeling. These analyses suggest — but do not prove — mechanisms beyond weight loss.[78] The authors caution that adiposity-change analyses can be confounded by reverse causation, since illness-related weight loss also occurs.

Additional plausible contributors include blood-pressure reduction, natriuresis, reduced epicardial adipose tissue, improved endothelial function, and direct myocardial effects. None of these has been isolated as その mechanism, and mediation analyses can support but never prove a causal pathway.

How much do GLP-1 drugs improve lipids, ApoB, and plaque?

Semaglutide and tirzepatide generally lower triglycerides, remnant-related measures, non-HDL-C, and sometimes LDL-C while improving blood pressure, glycemia, waist, and inflammatory markers.[28,32,53] The LDL-C change is usually modest compared with dedicated lipid-lowering therapy, and ApoB was not a uniform primary endpoint across obesity trials — an important limitation, since it means much of what is said about incretins and “particle burden” is inferred from non-HDL-C, triglycerides, and postprandial apoB-48 studies rather than measured fasting ApoB in large outcome trials. The National Lipid Association has explicitly called for apoB measurement in trials of lipid-altering interventions for exactly this reason.[61]

In STEP 4’s randomized-withdrawal period, non-HDL-C changed by 0% with continued semaglutide versus +10% after switching to placebo, LDL-C by +1% versus +8%, and triglycerides by −6% versus +15%; these quantify maintenance versus withdrawal, not the entire effect from pretreatment baseline.[53]

No major randomized coronary-imaging trial has yet shown that semaglutide or tirzepatide regresses plaque. Their event benefit may reflect multiple small changes in plaque biology, thrombosis, inflammation, hemodynamics, kidney function, and metabolism rather than a visible reduction in total プラーク体積. This is a genuine evidence gap, not a technicality: we have strong event data and weak imaging data, and readers should be skeptical of confident claims in either direction about plaque.

Does a GLP-1 replace a statin? No.

This is the most consequential practical question, and the answer is unambiguous. Incretin therapy lowers triglycerides and remnant particles substantially but usually lowers LDL-C and ApoB far less than 高強度スタチン, PCSK9-based therapy, or combination lipid-lowering regimens; エゼチミブ alone generally produces a more modest reduction.[52,61] In SELECT itself, participants remained on guideline-directed background therapy, so the 20% event reduction was demonstrated on top of statins, not instead of them.[30,52] A person taking an incretin should still receive guideline-directed treatment for ApoB, blood pressure, diabetes, and smoking, and ApoB should be remeasured after weight stabilizes rather than assumed to have normalized.

Newer agents: promising, but not yet mature evidence

Retatrutide, a triple GIP/GLP-1/glucagon receptor agonist, produced up to 24.2% mean weight loss at 48 weeks in a phase 2 randomized trial.[75] Cardiovascular-outcome efficacy has not been established in a peer-reviewed outcome trial; weight loss itself is not a validated surrogate for cardiovascular-event reduction.

Other emerging anti-obesity agents may have peer-reviewed weight-loss evidence, but cardiovascular-event benefit should not be inferred unless confirmed in agent-specific outcome trials. This review uses no unpublished or non-peer-reviewed efficacy result.

Safety, muscle, and practical considerations

Common adverse effects in trials are gastrointestinal, especially during dose escalation; discontinuation is more frequent than with placebo.[25,28,30,51] Systematic review of obesity pharmacotherapy shows this trade-off between efficacy and tolerability is a class-wide feature rather than unique to the incretins.[51] Gallbladder events can occur, and severe hypoglycemia is uncommon without insulin or sulfonylureas.

Rapid weight loss reduces lean tissue as well as fat — a recognized concern with incretin therapy, particularly in older adults and in anyone with baseline サルコペニア. The countermeasures are the same as for any rapid weight loss: adequate dietary protein, レジスタンストレーニング, and attention to strength and physical function rather than the scale alone.[48] This is not a reason to withhold effective therapy from a high-risk patient, but it is a reason to prescribe exercise and nutrition alongside it.

Choice and monitoring require individualized clinical care, particularly with gastrointestinal disease, frailty, pregnancy potential, or interacting glucose-lowering therapy. Cost and access remain substantial real-world determinants of who actually benefits.

Longevity and healthy years of life

The mortality relationship depends on age, smoking, illness, BMI degree, fat distribution, diabetes, and duration. In 239 prospective studies, carefully restricted analyses found a 31% higher all-cause mortality hazard per 5 kg/m² above BMI 25.[5] In 57 studies of about 900,000 adults, BMI 30–35 was associated with roughly 2–4 fewer median years of survival and BMI 40–45 with 8–10 fewer years compared with the optimum range.[6] In 1.46 million White adults, relative to BMI 22.5–24.9, all-cause mortality among healthy never-smoking women rose from a hazard ratio of 1.13 at BMI 25.0–29.9 to 1.44 at 30.0–34.9, 1.88 at 35.0–39.9, and 2.51 at 40.0–49.9; patterns were similar in men.[44]

Waist also matters. In a pooled analysis of 650,386 adults, higher waist circumference predicted mortality within BMI categories; people at the highest waists had several fewer estimated years of life after age 40 than those at the lowest waists.[45] Earlier life-table analyses similarly estimated substantial years of life lost with obesity, with larger losses at younger ages of onset.[46] Because these are observational estimates, they should not be converted mechanically into a personal countdown.

Higher BMI in young and middle adulthood is associated not only with shorter life but with earlier cardiovascular disease and more years lived with it. In a pooled analysis, overweight and obesity were associated with higher lifetime CVD risk, earlier onset, and a greater proportion of life spent with CVD even when differences in total longevity were modest.[7] Childhood or early-adult obesity is therefore more concerning than the same BMI appearing late in life: it implies longer 累積暴露 during a vulnerable life course. Diabetes and severe obesity amplify the risk; high fitness and favorable fat distribution attenuate it.

The “obesity paradox”

Some cohorts of people who already have heart failure, coronary disease, kidney disease, or cancer find lower short-term mortality at higher BMI. This does not show that gaining fat is protective. Disease-related weight loss can move the sickest participants into lower-BMI groups; smoking and frailty do the same. BMI cannot separate muscle from fat, and patients who survive long enough to enter a disease cohort are selected. Earlier diagnosis, greater metabolic reserve during acute illness, fitness, and treatment differences may also contribute. Reviews conclude that reverse causation, selection bias, and body-composition limitations explain much of the apparent paradox.[47]

Intentional and unintentional weight loss must therefore be distinguished. In older adults, weight-loss treatment should protect protein adequacy, resistance exercise, strength, bone health, and function. Lifestyle weight loss can reduce both fat and lean mass; combining exercise — especially resistance training — with adequate nutrition helps limit muscle and bone loss.[48]

Practical interpretation for an individual patient

The most useful assessment is layered:

Measure the body phenotype: weight trajectory, BMI, waist circumference, and preferably waist-to-height ratio. Note age of onset and recent unintentional loss. Assess strength and function in older adults.
Measure the causal and modifiable pathways: blood pressure; a fasting or nonfasting lipid panel; ApoB when triglycerides are elevated, obesity/diabetes/メタボリックシンドローム is present, LDL-C and non-HDL-C disagree, or treatment decisions remain uncertain; A1c or glucose; kidney function; smoking exposure; and symptoms of sleep apnea. Consider liver assessment when clinically indicated.[61]
Estimate absolute ASCVD risk and existing disease: 家族の歴史, lipoprotein(a), age, sex, and prior events matter. Coronary calcium or CT angiography should be used for a clinical risk question — not merely to measure obesity.
Treat every important pathway: improve nutrition, activity, sleep, and fitness; treat obesity when its burden justifies it; lower ApoB directly when needed; control blood pressure and diabetes; and stop tobacco exposure. These treatments complement rather than replace one another.

A normal LDL-C does not guarantee a low particle burden, particularly when triglycerides are elevated, HDL-C is low, diabetes or fatty liver is present, or the waist is large. ApoB is especially useful in these discordant settings.[61] Conversely, a lean person can have genetically high ApoB or lipoprotein(a), hypertension, diabetes, smoking exposure, inflammatory disease, low fitness, or substantial visceral fat and can develop extensive plaque.

Realistic initial weight-loss targets are 5%–10% because they often produce clinically meaningful metabolic improvement and are achievable by several methods.[14,39] Losses of 15%–20% can deliver larger benefits when appropriate, but maintenance, tolerability, cost, nutrition, and muscle preservation become central.[25,26,28] The treatment method matters: diet composition has direct lipid effects; exercise improves fitness and visceral fat with little scale change; incretins have metabolic and possibly weight-independent effects; surgery changes gut physiology and produces the largest durable average losses. None guarantees ApoB normalization or plaque regression.

What the evidence actually shows

Does obesity increase cholesterol? Often, but the clearest changes are higher triglycerides, VLDL/remnants, non-HDL-C, and ApoB — not necessarily a large rise in LDL-C.[1,2,13]
How much? No credible universal per-BMI or per-waist increment exists. In randomized weight-loss trials, lipid response varies by method; the best pooled 12-month lifestyle estimate per kilogram lost was TG −4.0 mg/dL, LDL-C −1.28 mg/dL, and HDL-C +0.46 mg/dL.[14]
Can LDL-C be normal while ApoB is high? Insulin resistance can create many cholesterol-depleted LDL and remnant particles, producing LDL-C–ApoB discordance, which is associated with coronary calcium decades later.[11-13,19,61]
Is obesity causal for coronary disease? The totality of dose-response, mechanistic, Mendelian-randomization, cohort, and randomized treatment evidence supports a causal contribution, especially from visceral and ectopic fat.[1,2,8,9,30,57,58]
How much risk is mediated through measured downstream risk factors? A substantial share, possibly a majority. Blood pressure, measured cholesterol, and glucose together accounted for about 46% of the high-BMI–coronary association in one large pooled observational analysis;[8] Mendelian-randomization mediation analysis estimated about 66% (95% CI 42%–91%) mediated by blood pressure, diabetes, lipids, and smoking combined.[57] Neither analysis measured ApoB particle number, so no defensible figure exists for ApoB-specific mediation; both estimates also leave a residual fraction unexplained.
Which fat is most dangerous? Visceral and ectopic fat — particularly liver, epicardial, and perivascular depots — usually signal greater metabolic risk than predominantly subcutaneous fat, and genetic evidence supports abdominal adiposity as causal.[3,4,58]
Can a lean person be high risk? Normal-weight central obesity, insulin resistance, high ApoB or lipoprotein(a), hypertension, smoking, diabetes, and low fitness can coexist with a normal BMI.[21]
Can fitness neutralize obesity? High fitness markedly lowers risk and may outweigh BMI as a mortality discriminator, but it does not prove absence of plaque or eliminate every obesity-related complication.[23]
Does weight loss reduce plaque? It improves many plaque-driving factors. Direct randomized evidence of coronary plaque regression is limited, and CAC progression alone is not treatment failure.
Does weight loss reduce events? SELECT proves semaglutide reduced major events in secondary prevention without diabetes; bariatric surgery is associated with fewer events and deaths; lifestyle event evidence is more mixed, with post-hoc data suggesting ≥10% loss may be the relevant threshold.[30,38,41-43,59]
Do GLP-1/dual-incretin drugs help, and how much? They produce roughly 10%–21% mean weight loss in major trials and improve metabolic risk. Semaglutide reduced MACE by 20% in SELECT (6.5% vs 8.0%; NNT ≈67 over ~40 months); across eight diabetes CVOTs the pooled MACE reduction was 14%; semaglutide also reduced kidney events and death in FLOW; tirzepatide reduced worsening-HF/CV-death events in obesity-related HFpEF and was noninferior to dulaglutide for MACE. Benefit was not fully explained by measured early weight change in a prespecified SELECT secondary analysis,[78] and randomized withdrawal/maintenance trials show that weight-management benefits generally depend on continued treatment.[27,29]
How much can obesity shorten life, and what should be treated? Moderate obesity has been associated with about 2–4 years and severe obesity with 8–10 years shorter median survival in large cohorts; measure waist, ApoB when informative, BP, glycemia, smoking, fitness, and muscle — and treat each abnormality directly.[6]

Appendix: Evidence-based graphics package (production briefs)

This appendix is intended for production use rather than for the reader-facing narrative. These briefs avoid presenting extrapolation as measured data.

Causal lipid pathway — conceptual model. Visceral adiposity → adipose insulin resistance/free-fatty-acid flux → hepatic VLDL production and impaired remnant clearance → higher ApoB/remnants → arterial retention and plaque. Add a parallel intestinal limb: insulin resistance → increased apoB-48/chylomicron production → postprandial remnants. Label: “Conceptual model supported by human metabolic, genetic, and clinical evidence; arrows are not effect-size estimates.”
Partly ApoB-independent pathway — conceptual model. Obesity → hypertension, diabetes, inflammation/thrombosis, sleep apnea/kidney disease, and cardiac/epicardial-fat remodeling → ASCVD, AF, and HF. Label HF and AF separately from plaque-mediated ASCVD.[1,2]
BMI × waist risk heat map — observed association. Use sex- and ethnicity-appropriate waist categories within BMI categories from a selected individual-participant cohort; do not splice risk estimates from unrelated cohorts. Label: “Observed adjusted association, not causal effect.”[3,45]
Changes across 5%, 10%, 15%, and 20% loss — illustrative evidence ranges. Plot weight itself exactly; show TG, glucose, BP, liver fat, and ApoB as bands rather than points. Use trial-specific panels for lifestyle, semaglutide, tirzepatide, and surgery. Label: “Illustrative ranges synthesized across heterogeneous trials; not a dose-response measured in one trial.”[14,25,26,28,53]
Fat-depot comparison — conceptual/observational hybrid. Compare subcutaneous, visceral, liver, epicardial, and perivascular fat by location, measurement method, metabolic signature, and strength of outcome evidence. Do not rank pancreatic fat as a proven direct coronary cause.[3,4,54]
GLP-1 cardiovascular outcome trial forest plot — observed randomized data. Plot MACE hazard ratios and 95% confidence intervals for ELIXA, LEADER, SUSTAIN-6, EXSCEL, HARMONY, REWIND, PIONEER 6, AMPLITUDE-O, and SELECT, with the pooled estimate shown separately and SELECT distinguished as the only non-diabetes population. Label: “Observed randomized trial results; trials differ in population, comparator, and endpoint definition.”[30,33,62-69]

Note on evidence quality and verification

Every reference in the list below is a peer-reviewed journal publication, and each numerical claim is attributed to the original study wherever possible. Three points of transparency:

No unpublished or non-peer-reviewed efficacy result is used. Every quantitative claim in this review rests on a peer-reviewed publication listed below.
Study designs are identified. Post-hoc and secondary analyses (Look AHEAD weight-magnitude analysis, SELECT hsCRP and adiposity analyses, SURPASS-CVOT cardiorenal composite) are named as such, and observational evidence is distinguished from randomized evidence throughout.

Findings published after September 3, 2026 — including ongoing incretin cardiovascular-outcome trials, coronary-imaging studies, and placebo-controlled obesity outcome trials such as SURMOUNT-MMO — could refine or revise several conclusions in this review, particularly those concerning tirzepatide’s placebo-referenced cardiovascular benefit and the effect of weight loss on plaque.

Editorial note: This review summarizes populations and averages; it is not a diagnosis or an individualized treatment plan. Medication and surgical decisions require clinical assessment of benefits, contraindications, adverse effects, nutrition, and patient preferences.

参考文献

All sources below are peer-reviewed journal publications. Primary trials and original cohort studies are used for quantitative claims whenever available.

透明性に関する注記： この記事はAIツールの支援を受けて作成されました。最終的なコンテンツは著者によって慎重に確認および編集されており、その正確性については著者が責任を負います。提供される情報は教育目的のみのものであり、医学的なアドバイスを構成するものではありません。.