1. The Hypothesis and How This Review Tests It
The proposition under examination is that a nutritionally adequate, entirely whole-food プラントベースの食事, very low in total fat (hereafter VLF-WFPB), offers cardiovascular protection beyond that of other high-quality dietary patterns, for prevention, for treatment of established disease, and for regression. Four separate questions are kept apart throughout: (Q1) Does VLF-WFPB improve outcomes compared with a typical Western diet? (Q2) Does it outperform other high-quality diets, such as DASH, Mediterranean, or Portfolio? (Q3) Are complete exclusion of animal foods, very low total fat, and minimal processing each independently necessary for any advantage? (Q4) Do such diets protect arteries through pathways beyond lowering ApoB-containing リポタンパク質? Q4 is treated as the central question (Section 5), because a benefit that runs entirely through アポリポ蛋白B could in principle be matched by any equally effective means of lowering ApoB.
The positive case rests on convergence. Randomized trials show that plant-based diets lower causal 危険因子; intensive lifestyle trials that included very-low-fat ベジタリアン食 showed functional and angiographic improvement; and populations with lifelong low animal-food intake had low コレステロール and low coronary mortality. Each line is weak alone. Together they make it more credible that sustained dietary lowering of ApoB-containing lipoproteins can slow coronary disease, which bears mainly on Q1. They cannot answer Q2 or Q3, because in every line the diet travels with other exposures—体重減少, physical activity, co-interventions, drugs—that also change risk. Convergence is therefore used here to judge plausibility, not to enlarge sample sizes or to compute a probability of superiority. Unrelated populations are not pooled.
Narrative reviews by Wang and colleagues, Freeman and colleagues, and Kahleova and colleagues served as background and as routes to original studies; numerical claims are taken from the original reports, and repeated citation of the same cohort is not treated as independent corroboration [1–3].
1.1 Scope and methods
This is a targeted critical review, not a registered システマティックレビュー. Sources were identified through PubMed-indexed records, publisher pages, full-text articles where accessible, guideline repositories, and reference tracing, with a final update search on 12 September 2026. Numerical claims were checked against original reports where accessible; where only an abstract or an official summary could be obtained, that is stated at the relevant passage. No new メタ分析 or patient-level reanalysis was performed.
Interests deserve attention without being disqualifying. Several lifestyle programs were evaluated by the investigators who developed them; the direct vegan–Mediterranean comparison was conducted by an organization that advocates plant-based diets; and the major Mediterranean secondary-prevention trial was funded principally by olive-oil foundations. These are reasons to emphasize allocation, retention, prespecification, blinded outcomes, and independent replication, not reasons to reject results.
1.2 How to read the evidence: five tiers of data
Not all evidence answers the same question. This review sorts studies into five tiers according to how well each design can show cause and effect, not according to whether its results support the hypothesis. Lower tiers still count—evidence is evidence—but they answer different questions and carry different risks of misleading us (Figure 1).
Tier 1: randomized trials with clinical events. People are assigned to a diet by chance, and 心臓発作, 脳卒中, or deaths are counted. Chance assignment reduces 交絡 in expectation, though not necessarily in any single trial, so a difference in events is the strongest available evidence that the diet caused it. Examples in this review are PREDIMED, コディオプレブ, 、そして リヨン地中海食心臓研究.
Tier 2: randomized trials of intermediate outcomes. Assignment is still by chance, but the outcome is a measurement that predicts events: LDL-C, ApoB, 血圧, coronary narrowing, プラーク体積, or myocardial blood flow. These trials show that a diet changes the measurement; they do not by themselves show fewer events. Examples are the ライフスタイル心臓トライアル, EVADE CAD, and the lipid meta-analyses.
Tier 3: 前向きコホート studies. Large groups of individuals report what they eat and are followed for years. These studies capture real disease in real people over decades, which trials rarely can, but they show association: people who choose a diet differ in many other ways, and statistical adjustment can only partly remove those differences (Section 9.1). Examples are Adventist Health Study-2, EPIC-Oxford, and the Danish nitrate cohort.
Tier 4: population comparisons, cross-sectional studies, and uncontrolled case series. Whole populations are compared (rural China, Okinawa, the チマネ), people are measured once, or treated patients are followed without a comparison group (the Esselstyn cohorts). These data generate hypotheses and show what is achievable, but many explanations can fit the same pattern.
Tier 5: mechanistic studies in cells and animals. Experiments can isolate a single pathway under controlled conditions, which makes them the best way to test how something might work. Whether the same pathway matters in human arteries over decades has to be shown separately.
Design is a starting point, not a verdict: confidence also depends on risk of bias, precision, how directly the comparison addresses the question, and how outcomes were ascertained. Tiers also apply to outcomes rather than to whole studies, so a trial can sit in one tier for imaging and another for clinical events. The case for a whole-food plant-based diet is strongest where several tiers point the same way and weakest where it rests on one tier alone. Section headings throughout the review note the main tier of evidence discussed.

Figure 1. Five tiers of evidence used in this review, the question each tier can answer, and examples of studies in each tier. Higher tiers are better at showing cause and effect; lower tiers add breadth, duration, and biological explanation.
2. Defining the Diets and the Comparisons
The proposed diet. VLF-WFPB consists of vegetables, legumes, intact whole grains, fruit, and minimally processed starchy foods, with no animal foods and little or no added oil, operationally about 10–15% of energy from fat. The 2014 Esselstyn protocol also excluded avocado, nuts, excess salt, and sugary foods, later also caffeine and fructose, and advised a multivitamin, vitamin B12, and flaxseed meal [4]. The earlier 1985 Esselstyn cohort was not vegan: patients eliminated oil, fish, fowl, meat, and dairy products except skim milk and nonfat yogurt, targeted 10% of energy from fat, and received individualized cholesterol-lowering medication [5]。 オーニッシュ・ライフスタイル心臓治験 diet was a 10%-fat whole-foods vegetarian diet that permitted nonfat dairy and egg whites [6, 7]; achieved fat intake was 6.2% of energy at one year and 8.5% at five years [6].
Related but different diets. Higher-fat whole-food vegan diets include nuts, seeds, avocado, or オリーブオイル; the low-oil phase of the 2024 Recipe for Heart Health trial still provided 32% of energy from fat [8], and the vegan arm of the EVADE CAD trial ended at about 30% [9]。一般的なビーガンダイエットは、食品加工や脂肪を制限することなく動物性食品を除外します。従来の低脂肪ダイエットは雑食性です。CORDIOPREVの低脂肪群では、赤身の肉と低脂肪乳製品を含めて脂肪分30%未満が処方され、32.1%を達成しました [10].
比較対照。 DASHコンビネーションダイエットは、飽和脂肪酸と総脂肪を減らしつつ、果物、野菜、低脂肪乳製品を重視しました [11]. 地中海式ダイエット ランダム化試験でテストされたものは、脂肪分が大幅に高くなっていました。PREDIMEDの各群には、エクストラバージンオリーブオイルまたはナッツが補充されました [12]; CORDIOPREVの地中海食群では、少なくとも35%の脂肪分が処方され、40.5%を達成しました [10]; また、リヨン介入試験では、バターやクリームの代わりにα-リノレン酸を豊富に含むマーガリンを提供し、地中海型のパターンに従うよう助言を行いました [13, 14]。 ポートフォリオ・ダイエット ナッツ、植物 タンパク質、粘性 ファイバー、および植物 ステロール を低飽和脂肪の背景に追加したものです [15]。表1は、実際にテストされたパターンをまとめたものです。
表1. テストされた食事パターン。
| パターン | 主要な試験における定義上の特徴 | 脂肪(エネルギー比%) | ラベルだけでは確立されないこと |
| VLF-WFPB | 豆類、全粒穀物、野菜、果物、でんぷん質の食品。動物性食品はなし。油はほとんどまたは全く使用しない。Esselstyn 2014では、ナッツとアボカドも除外されました [4] | 目標は約10〜15%。Ornish(ビーガンではなくベジタリアン)は6.2〜8.5%を達成しました [6] | 十分なB12、タンパク質、必須脂肪、または持続性 固守 |
| 不飽和脂肪酸が多めのホールフード・ビーガン | 動物性食品なし。ナッツ、種子、アボカド、またはオリーブオイルを含む | 32〜48% [8]; EVADE CADでは約30% [9] | 無油ビーガンパターンに対する劣等性 |
| DASH | 果物、野菜、低脂肪乳製品。飽和脂肪と総脂肪を削減。DASH-Sodiumにおけるナトリウム削減 [11, 16] | 削減されているが、極めて低くはない | その利益が乳製品に依存すること。試験におけるイベント減少 |
| 地中海 | 野菜、豆類、全粒穀物、ナッツ、エクストラバージンオリーブオイル、魚 [10, 12] | CORDIOPREVで40.5%を達成 [10] | 固定された三大栄養素比率、または無制限のエネルギー |
| ポートフォリオ | ナッツ、植物性タンパク質、粘性繊維、植物ステロールを低飽和脂肪ダイエットに追加 [15] | 試験によって異なる。ナッツを含むが、必ずしも脂肪分が多いわけではない | イベントの減少。中心的なエビデンスは 脂質低下;テストされたパターンは脂肪分10〜15%に制限されていませんでした |
| 従来の低脂肪(雑食性) | 赤身の肉、低脂肪乳製品、複合 炭水化物 [10] | <30%を処方。32.1%を達成 [10] | VLF-WFPBとの同等性 |
3. 三つのレバー:動物性食品の除外、総脂肪、および食品加工(ティア2〜3)
VLF-WFPBは、切り離し可能な3つの変化を束ねています。すなわち、動物性食品の除外、総脂肪をエネルギーの約10〜15%に制限すること、そしてホールフードまたは最小限に加工された食品から食事を構成することです。Esselstynと Ornishプログラムは、 脂肪制限とホールフードの食事を極限まで追求し、すべてまたはほぼすべての動物性食品を除外します(初期のプロトコルでは無脂肪乳製品を許可しており、Ornishのものは卵白も許可していました)。このため、これらはパッケージ全体の自然な実験場となりますが、同じ理由で、それ自体ではどの部分に効果があるのかを割り振ることはできません(図2)。「植物ベース」というラベルは最初のレバーのみを説明しており、エビデンスはそれだけでは不十分であることを示しています。

図2. 3つの切り離し可能な食事のレバーと、それらが冠動脈に影響を及ぼす経路 動脈硬化。ほとんどの介入プログラムは3つのレバーすべてを同時に変更し、共同介入を追加するため、試験では単一のレバーに利益を帰属させることはできません。
植物ベースの食事における食事の質。 計約209,000人の医療従事者を対象とし、8,631件の冠動脈イベントが発生した米国の3つのコホートにおいて、全体的な植物ベースの食事指数は、リスクの低下とわずかにしか関連していませんでした(ハザード比 for extreme deciles 0.92; 95% CI 0.83 to 1.01). A healthful index that rewarded whole grains, fruits, vegetables, nuts, legumes, oils, tea, and coffee was associated with 25% lower risk (0.75; 0.68 to 0.83), whereas an unhealthful index that rewarded refined grains, potatoes, sweetened beverages, juices, and sweets was associated with 32% higher risk (1.32; 1.20 to 1.46) [17]. Two points follow. Plant-based eating is not protective in itself; quality determines the direction. And the healthful index scored nuts and vegetable oils as beneficial, so this evidence favors a whole-food pattern rather than specifically a very-low-fat one.
Processing within plant foods. In 126,842 UKバイオバンク participants followed for a median of 9 years, each 10-percentage-point increase in energy from plant-sourced foods that were not ultra-processed was associated with 7% lower 心血管疾患 risk (hazard ratio 0.93; 95% CI 0.91 to 0.95) and 13% lower cardiovascular mortality (0.87; 0.80 to 0.94). Plant-sourced 超加工食品 showed the opposite association (1.05; 1.03 to 1.07 for disease and 1.12; 1.05 to 1.20 for mortality), and total plant-food intake, ignoring processing, showed no association [18].
Randomized evidence on processing. Processing also has randomized evidence, although for energy balance rather than atherosclerosis. In an inpatient crossover trial, 20 weight-stable adults received ultra-processed or unprocessed diets for two weeks each, with meals matched for presented calories, エネルギー密度, macronutrients, sugar, sodium, and fiber. On the ultra-processed diet they ate 508 ± 106 kcal/day more and gained 0.9 kg; on the unprocessed diet they lost 0.9 kg [19].

Figure 3. Diet quality and processing. (a) Plant-based diet indices and incident 冠動脈疾患, extreme deciles. (b) Plant-sourced foods by processing, per 10% of energy, UK Biobank. (c) Weight change in an inpatient crossover trial of ultra-processed versus unprocessed diets (mean ± SE). Panels come from different designs and are not comparable in magnitude. Sources: Satija 2017; Rauber 2024; Hall 2019.
What this changes. The evidence on processing is consistent in direction: associations for plant foods reverse with processing, and in a randomized feeding trial the ultra-processed pattern increased energy intake and weight relative to the unprocessed pattern, with nutrients matched as far as the design allowed. These findings support attention to food quality and processing; they do not establish the importance of processing relative to animal-food exclusion, a comparison no trial has made. This reframes the Esselstyn and Ornish programs. What they share with each other, and with rural China, Okinawa, and the Tsimane (Sections 6–8), is not zero animal food but a diet built almost entirely from whole, minimally processed foods. That feature, and the question of whether its benefits run only through ApoB (Section 5), are the focus of what follows.
4. The Biological Case: ApoB and the Lipid Pathway (Tiers 1–2)
4.1 Organizing framework: cumulative exposure to ApoB-containing particles
LDL and other ApoB-containing lipoproteins cause atherosclerotic cardiovascular disease, and genetic, epidemiologic, and trial evidence indicate that risk rises with both the magnitude and the duration of exposure [20]. Particles enter the arterial 内膜 in proportion to their concentration and are retained in susceptible sites, so the relevant quantity is 累積暴露 rather than a single measurement. This framework is used here as an organizing principle, not as a validated numerical risk equation.
The framework has two implications for diet. First, a modest reduction sustained from early adulthood could matter more than the same reduction begun after disease is established; this is the logic behind the population evidence in Sections 6 to 8. Second, in established disease the relevant benchmark is the statin-trial slope: about a 22% relative reduction in major vascular events per 1 mmol/L (38.7 mg/dL) of LDL-C lowering over roughly five years [21]. If that slope applied to the pooled dietary LDL-C difference of 0.30 mmol/L reported below, it would predict a relative reduction of about 7% over five years (1 − 0.78^0.30). That figure is an extrapolation, not a measured dietary effect: dietary trials last weeks to months, adherence erodes, and a diet also changes body weight, blood pressure, and other factors that the スタチン slope does not capture.
4.2 Randomized evidence on lipids and risk factors
The largest meta-analysis of randomized trials comparing vegetarian or vegan diets with omnivorous diets included 30 trials. Plant-based diets lowered 総コレステロール by 0.34 mmol/L (95% CI 0.23 to 0.44; about 13 mg/dL) and LDL-C by 0.30 mmol/L (95% CI 0.19 to 0.40; about 11.6 mg/dL). ApoB fell by 12.92 mg/dL (95% CI 3.20 to 22.63), a 14% reduction, but only six trials contributed ApoB data and heterogeneity was substantial (I² = 71.7%); 中性脂肪 did not differ overall [22]. The ApoB figure is the published pooled estimate; because it rests on six trials, its precision depends on those trials being independent randomized comparisons of diet alone, which could not be confirmed from the accessible sources; pooled analyses in this literature also require checking for multiple reports of a single cohort. These results support lipid-mediated plausibility. They do not establish event reduction, プラーク退縮, an individual’s response, or a special benefit from excluding 不飽和脂肪酸, because the trial diets and comparators varied widely.

Figure 4. Randomized evidence that plant-based diets lower LDL-C and ApoB. Between-diet differences with 95% 信頼区間; Koch 2023 values converted from mmol/L (× 38.67). Estimates share constituent trials and are not additive. Sources: Koch 2023; Wang 2023; Barnard 2021 (participants without medication changes).
A second meta-analysis restricted to people with, or at high risk of, cardiovascular disease pooled 20 trials (1,878 participants; mean duration 25.4 weeks). Vegetarian diets lowered LDL-C by 6.6 mg/dL (95% CI 3.1 to 10.1), HbA1c by 0.24 percentage points (95% CI 0.07 to 0.40), and body weight by 3.4 kg (95% CI 2.0 to 4.9), with no effect on 収縮期血圧 (−0.1 mm Hg; 95% CI −2.8 to 2.6) [7]. The article’s Key Points box states 6.8 mg/dL and 0.25%; the results section and forest plots give 6.6 mg/dL and 0.24%, which are used here. Two further details matter for Q2. Against usual diets, LDL-C fell 12.9 mg/dL; against active dietary comparators, the LDL-C difference was not statistically significant. And baseline LDL-C explained the between-trial heterogeneity [7]. Only four trials enrolled patients with established cardiovascular disease, three of them Ornish-type programs.
Blood pressure. Earlier meta-analyses, as summarized by Wang and colleagues, reported systolic reductions of about 2.5 mm Hg with vegetarian diets [7]. The null result in higher-risk patients is plausibly explained by background 降圧薬 therapy and by medication reductions during trials, which the authors note could mask diet effects. Blood-pressure benefit is not specific to plant-exclusive eating: the DASH diet, which includes low-fat dairy, lowered pressure substantially in controlled feeding [11, 16], and in a direct crossover comparison a Mediterranean diet lowered systolic pressure more than a low-fat vegan diet [23]. Blood pressure is therefore not a demonstrated advantage of VLF-WFPB. Estimates from the two lipid meta-analyses are not added together; they share trials.
4.3 Dietary cholesterol
食事性コレステロール raises serum cholesterol independently of 飽和脂肪酸. In a meta-analysis of egg-feeding studies cited by Freeman and colleagues, each additional 100 mg/day raised LDL-C by about 1.9 mg/dL and HDL-C by about 0.3 mg/dL; the response is larger when baseline intake is low and varies between individuals with intestinal absorption capacity [2]. Even people habituated to a very low intake respond: in eight Tarahumara men whose customary diet supplied little cholesterol, a 1,000 mg/day diet raised plasma cholesterol from 113 to 147 mg/dL after a cholesterol-free phase [24]. VLF-WFPB supplies essentially none; in EVADE CAD, dietary cholesterol fell to a median of 0 mg/day in the vegan arm versus 142 mg/day in the AHA arm [9]. The average contribution to LDL-C lowering is modest; the individual contribution can be larger. The AHA’s 2026 dietary guidance states that, for most people, dietary cholesterol is no longer a primary target for cardiovascular risk reduction [25]; eliminating it is therefore a minor, not a defining, part of the case for VLF-WFPB.
4.4 Saturated fat and what replaces it
In the current Cochrane review, reducing saturated fat lowered combined cardiovascular events (risk ratio 0.83; 95% CI 0.70 to 0.98; 12 trials, 53,758 participants), and larger reductions in saturated fat, reflected in larger cholesterol reductions, produced larger benefits. Subgroup analyses did not show a significant difference between replacing saturated fat with 多価不飽和脂肪酸 or with carbohydrate [26]. The May 2020 issue of that review reported a risk ratio of 0.79; the corrected version of record is used here. VLF-WFPB drives saturated fat very low (4.5% of energy in the EVADE vegan arm versus 6.6% in the AHA arm [9]), and it replaces it predominantly with starch and fiber from whole foods. Plant protein, viscous fiber, and plant sterols are further plausible LDL-lowering components; current 脂質異常症 guidance points patients toward reducing saturated fat and increasing fiber-rich plant foods [27].
The Portfolio evidence shows a different route to the same target. In a meta-analysis of controlled trials (439 participants), adding nuts, plant protein, viscous fiber, and plant sterols to a cholesterol-lowering NCEP Step II diet lowered LDL-C by about 17%, with reductions in ApoB and 非HDLコレステロール as well [15]. Because the Portfolio diet contains nuts, it demonstrates that excluding nuts is not a prerequisite for substantial dietary LDL-C lowering.
5. Beyond ApoB: The Central Question (Tiers 2–5)
If every benefit of VLF-WFPB ran through lower ApoB, the diet would be one of several interchangeable ways to lower ApoB, drugs included, and the case for its particular rules would reduce to how far and how durably it lowers 動脈硬化惹起性粒子. The distinctive claim of whole-food plant-based medicine is that it does more: that whole plant foods act on the 内皮, the gut, immune signaling, blood pressure, and energy balance in ways a statin does not. That claim is biologically serious, and it is the central question of this review.
Four kinds of evidence bear on it, in increasing order of strength: (1) a mechanism demonstrated in cells or animals under conditions where lipids are unchanged or controlled; (2) human physiological or バイオマーカー effects not explained by LDL-C change; (3) human outcome associations that persist after adjustment for lipids; and (4) randomized comparison of diets at matched ApoB. The subsections below 等級 each candidate pathway against these criteria (Figure 6). No study yet provides the fourth kind.
5.1 Fiber, gut microbes, and butyrate
Viscous fiber lowers LDL-C, but fermentable plant polysaccharides may also act through the 腸内マイクロバイオーム. Across 83 genetically diverse, atherosclerosis-susceptible mouse strains, the abundance of the butyrate-producing genus Roseburia was inversely related to 病変 size and was not correlated with cholesterol. In germ-free アポリポ蛋白 E–deficient mice colonized with defined bacterial communities, Roseburia intestinalis lowered systemic 炎症 and atherosclerosis only when the diet was rich in plant polysaccharides, and intestinal delivery of 酪酸塩 itself reduced endotoxemia and atherosclerosis [28]. This is the clearest demonstration in this review of a diet-dependent, cholesterol-independent atheroprotective mechanism. It comes from mouse models, and no human trial has shown that raising butyrate production changes 歯垢 or events.
5.2 Leafy-green nitrate, nitric oxide, and the endothelium
Leafy green vegetables, central to Esselstyn’s protocol, are the main dietary source of inorganic nitrate, which the body can convert to 一酸化窒素 through an enterosalivary pathway. In 53,150 Danish adults followed for up to 23 years (14,088 cardiovascular events), a moderate vegetable-nitrate intake (median 59 mg/day, roughly a cup of leafy greens) compared with the lowest quintile (median 23 mg/day) was associated with 15% lower cardiovascular disease risk (hazard ratio 0.85; 95% CI 0.82 to 0.89), and with lower risks of 虚血性心疾患 (0.88; 0.82 to 0.94), 虚血性脳卒中 (0.83; 0.76 to 0.91), and 末梢動脈疾患 (0.74; 0.67 to 0.83). The association plateaued above about 60 mg/day and persisted after adjustment for reported 高コレステロール血症 and other dietary factors, which is not the same as controlling measured, cumulative ApoB exposure; a 媒介分析 estimated that baseline systolic blood pressure explained 21.9% of it [29]. The authors note that the largest nitrate trial did not lower blood pressure in older adults with elevated pressure, and that observational data cannot separate nitrate from vegetable intake in general; in this cohort lettuce and potato supplied most vegetable nitrate [29]. The finding supports the leafy-green element of the whole-food lever; it is not specific to a vegan diet.
血管内皮機能 (acute and short-term human physiology; relevance to events unproven). A single high-fat meal transiently impaired brachial flow-mediated dilation in healthy volunteers [30]. In the same investigators’ ten-person experiment, an olive-oil meal reduced flow-mediated dilation acutely, by 31%, and the reduction was smaller when the meal included 抗酸化物質 vitamins or salad with balsamic vinegar [31]. Sustained feeding points the other way: in a meta-analysis of eight trials, olive-oil interventions increased flow-mediated dilation by 0.76 percentage points (95% CI 0.27 to 1.24) [32], and in 805 CORDIOPREV participants a Mediterranean diet improved flow-mediated dilation more than the low-fat diet after one year [33]. Over eight weeks in EVADE CAD, EndoPAT-measured endothelial function did not change in either arm [9]. Neither acute nor short-term vascular-function findings establish effects on events.
5.3 Inflammation and innate-immune memory
炎症 (human randomized biomarker evidence). EVADE CAD randomized 100 patients with angiographic coronary disease to eight weeks of a vegan or AHA-recommended diet, with groceries, sample menus, and dietitian support for both. At baseline 94–96% took statins and more than half took high-dose statins. The vegan diet produced a 32% lower high-sensitivity C反応性タンパク質 (β 0.68; 95% CI 0.49 to 0.94; P = 0.02), consistent after adjustment. LDL-C was 13% lower (adjusted β 0.87; 95% CI 0.78 to 0.97), which the investigators classified as nonsignificant under their Bonferroni threshold (α = 0.0015) for secondary endpoints. Weight, HbA1c, other lipids, leukocyte activation markers, and quality of life did not differ between arms [9]. Crucially, EVADE did not test very-low-fat eating: median reported fat intake at eight weeks was 29.9% of energy in the vegan arm and 30.2% in the AHA arm, both groups were encouraged to use unsaturated oils and olive oil appeared in recipes for both, and the principal dietary contrast was plant versus animal protein, with lower saturated fat and higher fiber in the vegan arm [9]. The hs-CRP result is a biomarker finding; it does not establish fewer events or an ApoB-independent clinical benefit.
Animal work suggests how diet could leave a lasting inflammatory imprint. In LDL-receptor–deficient mice, Western-diet feeding induced systemic inflammation that disappeared from the blood after a return to standard chow, yet myeloid progenitor cells remained reprogrammed, with heightened innate-immune responses; mice also lacking NLRP3 were protected [34]. Because these mice are severely hypercholesterolemic and 酸化LDL was implicated in the human arm of the study [34], this mechanism is not independent of lipoproteins; it suggests instead that the history of dietary exposure, not only current lipid levels, may shape plaque biology. The Tsimane point the other way: about half had hs-CRP above 3 mg/L from infectious burden, yet coronary calcium was the lowest recorded, which the investigators interpret as inflammation possibly not driving atherosclerosis when LDL is low [35].
5.4 Energy density, processing, body weight, and glycemia
Body weight and energy density (human randomized biomarker evidence). Vegetarian diets reduced weight by 3.4 kg in higher-risk patients [7], and a low-fat vegan diet produced greater weight loss than a Mediterranean diet in a crossover trial [23]. In the Lifestyle Heart Trial, the experimental group lost 10.9 kg at one year and remained 5.8 kg below baseline at five years [6]. Weight loss itself lowers LDL-C, blood pressure, and glycemia, so these effects overlap with the lipid pathway and cannot be added to it.
インスリン感受性 and glycemia (human randomized biomarker evidence). HbA1c fell 0.24 percentage points overall and 0.36 points in people with type 2 糖尿病 [7]. This may add to lipid effects in people with diabetes, but again partly through weight.
The processing trial (Section 3) shows one non-lipid route directly: the same nutrient targets delivered as unprocessed rather than ultra-processed food reduced spontaneous energy intake by about 500 kcal/day [19]. Blood pressure is a further separate, causal pathway. It is not a demonstrated advantage of plant-exclusive eating (Section 4.2), but nitrate-rich vegetables and potassium-rich, low-sodium patterns such as DASH lower it [11, 16, 29].
5.5 Trimethylamine N-oxide
Gut microbiome and trimethylamine N-oxide (human association plus mechanistic hypothesis). Gut microbes convert dietary precursors including コリン そして カルニチン to trimethylamine, which host liver enzymes then oxidize to TMAO [36]; circulating TMAO predicted cardiovascular events in cohort data [37]. メンデルランダム化 analyses have not found genetically predicted TMAO to be associated with 冠動脈疾患, 心筋梗塞, or stroke [38], which weakens a causal interpretation, and no trial shows that lowering TMAO reduces events. The 2014 Esselstyn report asserts that its participants were unlikely to harbor TMAO-producing flora, but TMAO was not measured in that cohort [4].
5.6 Human signals that ApoB does not fully explain—and those it does
Most measured benefits of plant-based diets can run through ApoB, weight, blood pressure, and glycemia. A diet may have a real total effect even if nothing remains after conditioning on these mediators; the question of an additional effect is separate. Three observations bear on it. First, in the Lifestyle Heart Trial, experimental-group ApoB fell from 1.000 g/L at baseline to 0.769 g/L at one year but was 1.014 g/L at five years, while percent 直径狭窄症 continued to improve; LDL-C remained 20% below baseline at five years [6]. Because the program also included exercise, stress management, and group support, and because a year-5 snapshot does not capture the lower cumulative exposure during the preceding years, this cannot be read as diet acting independently of ApoB. Second, within STARS, progression correlated with saturated and total fat intake after adjustment for LDL-C, an observational within-trial analysis [39]. Third, in cohort data roughly one-fifth of the vegetarian–IHD association may be mediated by ボディ質量指数 [40], and in EPIC-Oxford the vegetarian–IHD association weakened substantially after adjustment for self-reported cholesterol, blood pressure, diabetes, and body mass index [41]. No trial has compared dietary patterns at matched ApoB, blood pressure, and weight, which is the design needed to establish an effect beyond the mediators.
Figure 5 displays the Lifestyle Heart Trial data. Although it cannot establish a dietary effect independent of ApoB, it remains the best clinical exhibit for the hypothesis: the experimental group took no lipid drugs, and the change in percent 狭窄 continued from −1.75 points at one year to −3.07 points at five years, after ApoB had returned to 101.4 mg/dL (baseline 100.0), while controls worsened [6]. Two further observations sharpen the picture. First, the Tsimane had mean ApoB of 97 mg/dL at scanning, not especially low, yet almost no coronary calcium; their LDL-C averaged about 71 mg/dL (1.84 mmol/L) from 2004 to 2011 and has risen since, and they spend hours a day in physical activity [35], so lifetime exposure and activity may explain this as well as any dietary factor. Second, the Esselstyn 1995 cohort, often cited for this argument, cannot support it: every patient took cholesterol-lowering drugs and mean LDL-C was 71.6 mg/dL [42], so concomitant lipid lowering is a plausible explanation for part of the improvement, and the uncontrolled design cannot isolate the diet’s contribution or show that any pathway beyond ApoB was involved.

Figure 5. Lifestyle Heart Trial. (a) LDL-C and ApoB in the experimental group, which took no lipid-lowering drugs (ApoB converted from g/L × 100). (b) Change in percent diameter stenosis on quantitative angiography, plotted from the 1998 five-year report for the 35 participants with five-year angiography; the 1990 one-year report covers all 48 participants and gives slightly different baselines, so its values are not plotted here. 60% of controls started lipid drugs between years one and five.
5.7 Weighing the beyond-ApoB case
The additional-benefit hypothesis is therefore conditional: a stricter pattern could outperform another healthy diet if it produces a larger and sustained improvement in ApoB, blood pressure, weight, or 代謝の健康 without poorer adherence or nutritional disadvantage. “Vegan,” “no oil,” and “10% fat” are not themselves validated 代用エンドポイント. Nor can a short-term dietary lipid change be converted mechanically into an Esselstyn-specific event reduction using drug-trial slopes.
On balance, the beyond-ApoB case is biologically plausible and partly supported. Animal models demonstrate one diet-dependent, cholesterol-independent mechanism (fiber and butyrate) and one mechanism of lasting inflammatory memory; human cohorts link leafy-green nitrate to lower cardiovascular risk after adjustment for hypercholesterolemia, and minimally processed plant foods to lower risk; and a randomized trial shows that processing alone changes energy intake [18, 19, 28, 29, 34]. Against this, the vegetarian–coronary association in EPIC-Oxford weakened substantially after adjustment for conventional risk factors [41], genetic evidence does not support TMAO as causal [38], and no trial has compared diets at matched ApoB. Figure 6 summarizes the evidence by pathway; the proposed trial (Section 17) is designed to supply the missing test.

Figure 6. Candidate mechanisms beyond ApoB and the type of evidence for each. Author synthesis of sources cited in Section 5; “supportive” indicates the direction of evidence, not proof of a causal effect on human atherosclerosis.
6. Rural China (Tier 4)
6.1 What was measured
China Study I combined county mortality data with dietary, blood, and urine surveys in 65 rural counties and 130 villages, sampling 50 adults per village; diet and blood were collected in 1983–1984 [43]. The average rural diet supplied 14% of energy from fat, 71% from carbohydrate, 5% from アルコール, and 10% from protein, of which about 11% was of animal origin (roughly 1% of total energy); fiber intake was 33 g/day and mean body mass index 20.5 [43]. Mean serum total cholesterol was 127 mg/dL, compared with 203 mg/dL in US adults aged 20–74 [43]. The diet was low in animal food, not vegan; the authors describe rural diets of 1950–1980 as containing 3–6% animal-based foods [43].
6.2 What the coronary figures are
The widely quoted coronary comparison uses mortality, not incidence, recorded in 1973–1975 and truncated at ages 0–64: 4.0 per 100,000 men and 3.4 per 100,000 women in rural China versus 66.8 and 18.9 in the United States, taken from a 1989 WHO statistics annual, giving ratios of 16.7 and 5.6 [43]. Four features limit interpretation. The mortality period precedes the dietary and blood survey by about a decade. Excluding deaths after age 64 removes the ages at which most coronary deaths occur. Death certification and coronary diagnosis in rural China in the 1970s differ from US practice, so under-ascertainment cannot be excluded. And competing mortality from infections and other causes of death, which the authors describe as clustering in the same counties, removes people before coronary disease manifests [43]. No プラーク負荷 was measured.
6.3 What the county correlations show
Across counties, coronary mortality correlated positively with plasma ApoB (r = 0.37) and with an index of salt intake (r = 0.42), and inversely with green-vegetable intake; ApoB in turn correlated with animal protein and meat intake [43]. Coronary mortality also correlated with wheat flour intake (r = 0.67), which the authors attribute partly to co-varying milk, salt, triglycerides, and body weight [43]. The authors themselves note that the county, not the individual, is the unit of analysis, so the data cannot show whether low disease rates reflect uniformly low animal-food intake or fewer individuals eating more [43].
6.4 Confounding and alternative explanations
Energy intake per kilogram was about 30% higher than in the United States with far less 肥満, which the authors attribute to greater daily energy expenditure such as cycling to work [43]. Physical activity, body weight, tobacco use, socioeconomic conditions, health care access, and diagnostic ascertainment all differ between these populations and the United States. The coronary paper does not report ischemic or 脳出血, and this review makes no claim about stroke subtypes in these counties. This review also did not re-verify post-1990 trend data; claims that China’s later dietary transition proves a causal diet effect should rest on contemporaneous individual-level cohorts.
6.5 Checking the secondary summary
Freeman and colleagues’ 2017 review is a useful route to these data but should not be quoted in place of them [2]. Its table correctly gives the 0–64 age restriction and the 127 mg/dL cholesterol, but lists macronutrients of 14% fat, 71% carbohydrate, and 10% protein, which sum to 95% because the 5% alcohol in the original is omitted [2, 43]. Its Tarahumara entry reports 528 people surveyed and a mean adult cholesterol of 136 mg/dL [2]; the original abstract reports 523 people aged 5 to 70 with a mean of 125 mg/dL overall and 116 mg/dL in children, a diet of 12% fat, 2% saturated fat, 71 mg/day cholesterol, and 75% carbohydrate, and a virtual absence of 高血圧, obesity, and the usual age-related rise in cholesterol [44]. The difference in denominators could not be resolved without the full text. The Tarahumara study measured lipids and diet, not coronary outcomes.
6.6 What rural China contributes
The rural Chinese data are consistent with the cumulative-exposure hypothesis: populations with lifelong low animal-food and saturated-fat intake had low cholesterol, and county coronary mortality tracked ApoB. That strengthens plausibility for Q1. It says nothing decisive about Q2 or Q3, because the diet was neither vegan nor compared with a high-quality alternative, and because activity, body size, and competing mortality are inseparable from diet in these data.
7. Traditional Okinawa (Tier 4)
The traditional 沖縄の食文化 is known chiefly from a 1949 survey, conducted during post-war scarcity and US administration, as analyzed by the Willcox group. Sweet potato supplied about 69% of energy; total energy intake was about 1,785 kcal/day; fat supplied roughly 6% of energy; and meat intake was a few grams per day, with small amounts of fish, soy, and seaweed [45, 46]. The Willcox group estimates that adults ate about 11% fewer calories than needed to maintain weight until the late 1960s, with a lean average body mass index of 21 [47]. Interpretation is contested: one critic has argued that the Okinawan data reflect severe malnutrition, whereas Gavrilova and Gavrilov, responding, attribute the later loss of longevity advantage to westernization of the diet and note that a low infectious burden may also have contributed [48]. A single post-war survey cannot establish lifelong intake.
Coronary outcomes should be judged directly rather than through longevity. The same group reports that older Okinawans have about 80% less coronary mortality than the US population, based on age-adjusted vital statistics rather than individual dietary linkage [47]. The advantage has not persisted: Okinawans who did not experience the energy-restricted era now have higher body mass index, more type 2 diabetes, and worse cardiovascular risk factors than other Japanese, and the prefecture’s life-expectancy advantage is now confined to older ages [47]. That transition is compatible with a dietary contribution but equally with changes in energy balance, activity, and other exposures.
Okinawa informs the hypothesis in a limited way. It shows that a very-low-fat, high-carbohydrate, plant-predominant diet is compatible with low coronary mortality. It does not show that complete exclusion of animal foods is necessary, because the traditional diet included pork and fish, and it cannot separate low fat from caloric restriction, low body size, physical labor, or 遺伝学. Its shared features with DASH and Mediterranean patterns are high vegetable and legume intake, low saturated fat, and low energy density; its distinctive features are very low total fat, a single dominant starchy staple, and chronic mild energy restriction.
8. Other Traditional Populations: The Tsimane and Kitava (Tier 4)
The Tsimane. The Tsimane, forager-horticulturalists of the Bolivian Amazon, provide the only traditional-population data in this review with direct 冠動脈画像診断. Of 705 adults aged 40 to 94 scanned in 2014–2015, 596 (85%) had no coronary 動脈 calcium, 89 (13%) had scores of 1–100, and 20 (3%) had scores above 100; among those older than 75, 31 (65%) had none and four (8%) had scores of 100 or more [35]. Mean LDL-C was 91 mg/dL and HDL-C 39.5 mg/dL, and obesity, hypertension, 高血糖, and regular 喫煙 were rare. Compared with the US MESA cohort, the Tsimane reached a nonzero カルシウムスコア about 24 years later and a score of 100 or more about 28 years later [35] (Figure 7).
The Tsimane diet is low in fat and minimally processed but not plant-exclusive: about 14% of energy from protein, 14% from fat, and 72% from carbohydrate, an estimated 38 g of fat per day including 11 g of saturated fat and no トランス脂肪酸, with rice, plantain, manioc, and corn as staples and meat and fish obtained by hunting and fishing [35]. Men and women average 6–7 and 4–6 hours of physical activity per day. The limits are important: calcium scoring cannot detect 非石灰化プラーク, the design is cross-sectional, and outcome data are thin, with one possible myocardial infarction among 50 recent adult deaths ascertained by verbal autopsy. LDL-C has risen by about 0.16 mmol/L per year since 2011 as motorized river travel improved access to market food [35], a natural experiment whose coronary consequences are not yet known.

Figure 7. Coronary artery calcium in the Tsimane. (a) Proportion with a calcium score of zero by age group, Tsimane versus US MESA, as labeled in Figure 2 of Kaplan 2017. (b) Distribution of scores among 705 Tsimane adults. Calcium scoring does not detect noncalcified plaque.
Kitava. On Kitava in the Trobriand Islands, where tubers, fruit, fish, and coconut are dietary staples, semi-structured interviews with 213 adults identified no case corresponding to stroke, sudden death, or 狭心症, and resting electrocardiograms showed few abnormalities [49]. Smoking was common: 76% of men and 80% of women over 20 smoked in the risk-factor survey [50]. The evidence rests on interviews and electrocardiograms rather than imaging or death registration.
What these populations add. None was vegan, and their fat intakes were not uniformly low in saturated fat. The Kitavan diet supplied about 21% of energy as fat and 17% as saturated fat, mostly lauric and myristic acid from coconut [51], a clear exception to the pattern. What rural China, traditional Okinawa, the Tsimane, and Kitava share is diets built from whole, minimally processed staples, leanness, and in most cases high physical activity. The pattern supports the cumulative-exposure argument and the whole-food lever; it provides no support for the claim that complete exclusion of animal foods is required. Fat intake ranged from very low in Okinawa to low among the Tsimane, and coconut was a Kitavan staple.
9. Prospective Cohort Evidence (Tier 3)
The most comprehensive cohort meta-analysis included 13 prospective cohorts with 844,175 participants. Compared with non-vegetarians, vegetarians had lower risk of cardiovascular disease (相対リスク 0.85; 95% CI 0.79 to 0.92; 8 cohorts) and ischemic heart disease (0.79; 95% CI 0.71 to 0.88; 8 cohorts), but not of total stroke (0.90; 95% CI 0.77 to 1.05; 12 cohorts). For vegans the ischemic heart disease estimate was 0.82 (95% CI 0.68 to 1.00; 6 studies), and the cardiovascular disease estimate 0.92 (95% CI 0.79 to 1.06). Risk of bias was moderate in eight cohorts and serious in five [40]. Incidence estimates were used in preference to mortality where both were reported. The IHD association was weaker when early follow-up was excluded, about one-fifth of it appeared attributable to body mass index, and the E-value was 1.86 (lower confidence limit 1.49), meaning an unmeasured 交絡因子 would need associations of that strength with both diet and disease to explain it away [40].
The “40% lower coronary risk” figure sometimes attached to vegetarian diets traces to a meta-analysis restricted to Seventh-day Adventist cohorts, which reported a relative risk of 0.60 (95% CI 0.43 to 0.80) for coronary events [2, 52]. That estimate is population-specific and should not be generalized. Several Adventist and Oxford cohorts recur across meta-analyses, so repeated citation is not independent corroboration. Cohort vegetarian and vegan categories are defined by exclusion of animal foods, not by total fat, so none of this evidence tests the very-low-fat component.
Endpoint-specific data matter. In EPIC-Oxford (48,188 participants followed for 18.1 years), vegetarians including vegans had a 22% lower rate of ischemic heart disease than meat eaters (hazard ratio 0.78; 95% CI 0.70 to 0.87), equivalent to about 10 fewer cases per 1,000 people over 10 years; adjustment for self-reported high cholesterol, high blood pressure, diabetes, and body mass index attenuated the estimate to 0.90 (95% CI 0.81 to 1.00). The same group had a 20% higher rate of total stroke (1.20; 95% CI 1.02 to 1.40), about three more cases per 1,000 over 10 years, mostly hemorrhagic, and this association did not attenuate with risk-factor adjustment [41]. Coronary benefit and stroke risk must therefore be reported separately, and the attenuation of the coronary association is consistent with benefit running largely through conventional risk factors.

図8. ベジタリアン食およびヴィーガン食に関する前向きコホートの推定値。従来の危険因子を調整すると、冠動脈疾患との関連性は弱まり(EPIC-Oxford)、脳卒中リスクの低下は見られない。推定値はメタ解析間でいくつかのコホートが重複しているため、統合(プール)せず並べて示している。出典:Dybvik 2023; Tong 2019; Orlich 2013; Kwok 2014。
9.1 健康者バイアス:なぜ関連性の解釈が困難なのか
コホート研究は異なる食事を選択した人々を比較するが、菜食を選択する人々は通常、他の健康的な習慣も併せ持っている。これは健康者バイアスと呼ばれる交絡の一種であり、観察された疾患の差の一部または全部が、食事ではなく他の習慣に起因している可能性がある。アドベンチストのデータはこの問題を明確に示している。アドベンチスト健康研究-2の非菜食主義者と比較して、ヴィーガンは喫煙経験がない可能性が高く(85.0%対75.7%)、アルコールを全く摂取しない可能性が極めて高く(98.8%対83.4%)、週に少なくとも151分以上の激しい運動をする可能性が高く(24.8%対17.2%)、大学院の学位を保持している可能性が高く(19.5%対14.1%)、より痩身であった(平均BMI 24.1対28.3)[53](図9)。著者らは、菜食という意識的なライフスタイルの選択自体が結果に影響を与える可能性があり、制御されていない交絡が残っている可能性があると指摘している [53].
研究者はこのバイアスを軽減するためにいくつかの手法を用いているが、それぞれに限界がある。 似た者同士の比較: アドベンチスト健康研究-2とEPIC-Oxfordの両方において、比較対象となる非菜食主義者グループ自体が比較的健康意識が高く、喫煙や飲酒をするアドベンチストは極めて少ない。これによって差は縮まるものの、完全にはなくならない [53]. 統計的調整: モデルは喫煙、運動、教育、および同様の要因について調整を行うが、それは測定された要因のみであり、その測定の正確さに依存する。 調整の影響を確認する: EPIC-Oxfordがモデルにコレステロール、血圧、糖尿病、BMIを追加したところ、菜食と虚血性心疾患のハザード比は0.78から0.90に変化した [41]。これは、恩恵の多くがそれらの危険因子を介していることを示唆しており、食事が機能している場合に予想される通りである。また、推定値がモデリングの選択にどれほど敏感であるかも示している。 頑健性の定量化: Dybvikのメタ解析では、関連性を完全に説明するためには、未測定の交絡因子が食事と心疾患の両方に対して1.86の相対リスクを持つ必要があるとされる [40]. 期間と年齢: 5つのコホートの統合分析では、菜食主義者における虚血性心疾患死亡率の低下は、その食事を5年以上続けている人に限定されており、また若年層でより顕著であった(65歳未満で45%低下、65〜79歳で31%低下、80〜89歳では8%低下で有意差なし) [54]。食事の期間との関連は、因果効果があれば生じるものであるが、長期の菜食主義者は他の長期的な習慣も異なっている可能性がある。
さらに2つの観察結果が両面性を示している。あるメタ解析では、非アドベンチストのコホート(相対リスク0.84; 95% CI 0.74〜0.96)よりも、アドベンチストのコホート(相対リスク0.60)において、虚血性心疾患とのより強い関連が見られた [52]。この差は、アドベンチストにおけるより強力な健康者効果を反映している可能性もあれば、アドベンチストの菜食主義者が実際に何を食べているかを反映している可能性もある。アドベンチストのヴィーガンは1日あたり約46〜47gの食物繊維を摂取していると報告しているが、EPIC-Oxfordのヴィーガンは約26〜28gである [53]。そして 逆因果関係—初期の疾患の後に食事を変える人々—は、両方のコホートが行ったように、登録時に既存の心血管疾患を持つ人々を除外することで軽減されるが、完全には排除されない [41, 53].
これらはどれも、コホートの証拠を軽視すべきであることを意味しない。それらは大規模で長期的であり、国を越えて一貫しており、脂質のランダム化比較試験や生物学的な方向性と一致している。これはティア3の証拠として解釈されるべきである。すなわち、関連性については強力であり、因果関係については支持的だが決定的ではなく、それ単体では、ある健康的な食事を別の食事と比較してランク付けすることはできない。

図9. 健康な利用者バイアス。(a) ライフスタイルの選択が、食事そのもの以外の経路を通じて、いかに食事と心疾患リスクの低下を結びつけるか。(b) アドベンチスト・ヘルス・スタディ2におけるビーガンと非菜食主義者のベースライン特性。Orlichらの2013年の報告に基づき、年齢、性別、人種で標準化。
9.2 ビーガンは他の菜食主義者よりも良好な結果を得ているか?
ビーガン食はすべての動物性食品を排除します。ラクト・オボ・ベジタリアン食は乳製品と卵を含みます。 ペスコ・ベジタリアン 食には魚が含まれます。もし動物性食品を完全に排除することが保護効果を高めるのであれば、ビーガンは他のグループよりも良好な結果を示すはずです。しかし、コホートデータはそのようには示していません(図10)。
アドベンチスト・ヘルス・スタディ2。 非菜食主義者と比較して、虚血性心疾患死のハザード比は、ビーガンで0.90(95% CI 0.60〜1.33)、ラクト・オボ・ベジタリアンで0.82(0.62〜1.06)、ペスコ・ベジタリアンで0.65(0.43〜0.97)であり、統計的に有意な減少を示したのはペスコ・ベジタリアンのみでした [53]。男性では、ビーガンのハザード比は虚血性心疾患死で0.45(0.21〜0.94)、心血管疾患死で0.58(0.38〜0.89)でした。一方、女性では対応する推定値は1.39(0.87〜2.24)および1.18(0.88〜1.60)でした [53]。コホート全体では平均5.79年間で372件の虚血性心疾患死が記録されたため、これらのサブグループの推定値は不正確であり、菜食と心血管死亡率の関連は性別によって有意に異なっていました [53]。88,400人の参加者を対象とした2024年のより長期の追跡調査では、菜食主義者全体で虚血性心疾患死亡率が低かったものの、ビーガン食は、 全因死亡率 男女合計における低下とは関連していませんでした。ビーガンの男性は、若年層においてのみ死亡率が低くなっていました [55].
EPIC-Oxfordおよびプールされたコホート。 肉食者と比較して、EPIC-Oxfordにおける虚血性心疾患の発症率は、ビーガン(67例)で0.82(0.64〜1.05)、ラクト・オボ・ベジタリアンで0.77(0.69〜0.86)、魚食者で0.87(0.77〜0.99)でした [41]。5つのコホートのプール解析では、定期的に肉を食べる人と比較して、虚血性心疾患の死亡率はビーガンで26%低く、ラクト・オボ・ベジタリアンと魚食者の両方で34%低くなっていました [54]。ビーガン食に関する系統的レビューでは、少なくとも7,380人のビーガンを含む3つのコホート研究のいずれにおいても、主要な心血管アウトカムについてビーガンのリスクが有意に高い、または低いとは報告されていないことがわかりました [56].
これが示すこと、示さないこと。 ビーガンの数が少ないため、その推定値の幅は広く、「効果なし」というよりは「結論が出せない」状態です。このレビューにおいてより重要な点は、これらのコホートのビーガンはVLF-WFPBを実践していなかったことです。EPIC-Oxfordのビーガンはエネルギーの28.1%を脂質から摂取し、1日に約26gの食物繊維を摂取していました [41]。これは脂質が非常に低いわけでも、特に一物全体食が多いわけでもありません。これらのコホートは、ビーガンの心血管リスクがラクト・オボ・ベジタリアンや魚食者よりも低いことを証明するものではありません。また、各グループは他のグループと直接比較されるのではなく肉食者と比較されているため、同等性や劣等性を証明するものでもありません。なお、EPIC-Oxfordでは、ビーガンの点推定値は数値上、魚食者よりも低くなっていました。アドベンチスト・ヘルス・スタディ2内では、最高五分位と最低五分位を比較すると、ナッツや種子由来のタンパク質は心血管死亡率の40%低下と関連し、肉由来のタンパク質は61%の上昇と関連していました。これらの関連は、菜食のタイプを調整した後も持続しました [57]。このパターンは植物性タンパク質とナッツを支持するものです。ナッツや、すべての動物性食品を排除することを支持するものではありません。

図10. ビーガンと他の菜食主義グループの比較。(a) アドベンチスト・ヘルス・スタディ2(死亡)およびEPIC-Oxford(発症例)における食事グループ別の虚血性心疾患。それぞれ非菜食主義者または肉食者と比較。(b) アドベンチスト・ヘルス・スタディ2の性別ごとのビーガン。信頼区間が広いのはビーガンの数が少ないことを反映している。出典:Orlich 2013 (Table 4); Tong 2019 (Supplementary Table 3)。
9.3 中国および台湾の菜食主義者
The search for this review found no prospective study from mainland China comparing vegans with omnivores for heart disease outcomes. The available Chinese and Taiwanese evidence concerns vegetarians, most of whom eat dairy and often eggs. In Taiwan, two Buddhist Tzu Chi cohorts (13,352 participants in total) found lower stroke risk among vegetarians: in the first cohort, the hazard ratio for ischemic stroke was 0.26 (95% CI 0.08 to 0.88); in the second, overall stroke 0.52 (0.33 to 0.82), ischemic stroke 0.41 (0.19 to 0.88), and hemorrhagic stroke 0.34 (0.12 to 1.00) [58]. Participants were members of a Buddhist foundation, a population with its own healthy-user profile, and an exploratory analysis suggested that vitamin B12 intake modified the association [58]. The hemorrhagic stroke result runs opposite to EPIC-Oxford’s [41], a reminder that stroke subtypes behave differently across populations.
ある 横断研究 from Xiamen, 169 healthy Chinese lacto-vegetarian men had lower blood pressure, LDL-C, ApoB, triglycerides, and fasting グルコース, and thinner carotid 中内膜厚, than 126 omnivorous men [59]. This is Tier 4 evidence on 代わりのマーカー in dairy-eating vegetarians.
Hong Kong data add a caution. Vitamin B12 deficiency was reported in about 80% of Hong Kong vegans, who rarely used fortified foods or supplements; B12-deficient vegetarian groups showed impaired arterial endothelial function and thicker carotid walls, and B12 supplementation improved these vascular measures in Hong Kong vegans [60]. An unsupplemented vegan diet can therefore undermine the vascular benefit it is meant to provide, which reinforces the nutrient guidance in Section 15.
10. Esselstyn’s Studies Reassessed (Tier 4)
10.1 The 1985 Cleveland Clinic cohort
Three reports describe this cohort, and their denominators differ. The full 1995 report describes 22 patients (21 men, 1 woman) with severe, angiographically documented coronary disease enrolled between 1985 and 1988; the 11 participants whose results are reported all had triple-vessel disease and were nondiabetic, nonhypertensive nonsmokers [42]. The diet derived less than 10% of energy from fat and excluded oils, meat, fish, fowl, and dairy except skim milk and nonfat yogurt. Every participant also received an individualized cholesterol-lowering drug, most often cholestyramine 4 g twice daily with lovastatin 40–60 mg daily; relaxation and meditation training was offered but abandoned within weeks, and exercise was not prescribed [42]. In the 11 imaged participants, mean total cholesterol fell from 246 mg/dL at baseline to 132.4 mg/dL during treatment, with mean LDL-C 71.6 mg/dL and HDL-C 36.3 mg/dL [42].
Of 38 lesions with more than 20% stenosis in those 11 participants, three treated by 経皮的冠動脈形成術 and four native-vessel lesions proximal to bypass grafts were excluded a priori—the latter because, as the authors note, such lesions were expected to progress, and they did—and six more could not be matched at follow-up, leaving 25 [42]. Two technicians masked to 血管造影 sequence read the films. By percent diameter stenosis, 11 of 25 lesions regressed and 14 were stable, mean stenosis fell from 53.4% to 46.2% (estimated decrease 7 percentage points; 95% CI 3.3 to 10.7), and 8 of 11 participants were classified as regressing. By minimal ルーメン diameter, the less reference-dependent measure, 6 lesions regressed, 14 were stable, and 5 progressed, and the mean increase was 0.08 mm (95% CI −0.06 to 0.22; not significant) [42].

Figure 11. Esselstyn 1995 cohort: enrollment, attrition, lesion selection, and angiographic results by the two methods reported. Source: Esselstyn 1995 (full text).
The same paper reports attrition two ways. Its abstract states that 5 of 22 participants dropped out within two years and 17 maintained the diet; its dropout analysis states that 11 left within two years (three moved, four had work conflicts, three could not maintain the diet, and one chose バイパス手術) [42]. The five dropouts who resumed their previous diet reported 10 心臓発作. 画像診断を受けた11名の参加者は、登録前の8年間に37件の心血管イベントを経験していた。追跡調査中に新たな心筋梗塞を起こした者はいなかったが、2名は研究中に冠動脈処置を必要とし(1名は再血管形成術、もう1名はバイパス手術)、バイパス手術を受けた患者は、 駆出率 が20%未満であり、後に不整脈で死亡した [42].
1999年の更新報告では、コホートは24名の患者(男性23名、女性1名)として記述されている。非遵守の患者6名は12〜18ヶ月以内に除外され、標準治療に戻った。18名は5年間遵守し、そのうち11名が5年目の血管造影を受け、11名全員に進行停止、8名(73%)に狭窄率による退縮が認められた。平均コレステロール値は5年間で237から137 mg/dLに低下し、12年目には145 mg/dLであった。遵守した18名の患者は、登録前の8年間に49件の冠動脈イベントがあったが、追跡調査中はゼロであったと報告されている。一方、除外された6名の患者には1998年までに13件の新規イベントが発生した [5]。「冠動脈イベントなし」という表現は、報告書独自のイベント分類を反映したものである。同じコホートには、5年目の血管造影後に心筋梗塞を伴わない心室性不整脈死が含まれていた [42]。2014年の報告書ではまた別の要約が示されており、22名中17名が遵守し、 反転 が12名中4名で血管造影により確認されたとしている [4]。ベースラインのコレステロール値(246対237 mg/dL)、コホート規模(22対24)、および過去のイベント数(1995年の議論では11名で37件、1999年では18名で49件)は、報告書間で異なっている。2つの異なる73%という数字が流通している。狭窄率による退縮が見られた画像診断済み患者11名中8名と、1995年時点で食事療法を継続していた当初の患者22名中16名である [5, 42].
登録後のイベントとそれ以前の8年間のイベントを比較することは、平均への回帰の影響を受けやすい。なぜなら、患者は通常、イベントが頻発した後にこのようなプログラムに参加するからである。この療法は食事療法と薬物療法を組み合わせていたため、食事療法の独立した寄与を推定することはできない。
10.2 198名の患者からなる2014年のコホート
設計と登録。 心血管疾患を持つ200名の自発的なボランティアに対しカウンセリングが行われた。2名が追跡不能となり、198名が残った(91%が男性、平均年齢62.9歳、平均追跡期間44.2±24.1ヶ月)。195名で冠動脈疾患が確認され、うち180名は血管造影またはCT血管造影によるものであった。44名に心筋梗塞の既往があった。全員が非喫煙者で、161名に高脂血症、60名に高血圧、23名に糖尿病があった。介入は1回5時間のセミナーと、その後の電話または電子メールによるフォローアップであり、患者は通常の心臓病薬(記録はされていない)を継続した。運動は推奨されたが、必須ではなかった [4].
遵守状況とデータ収集。 肉、魚、乳製品をすべて避け、意識的に添加油も避けた患者を遵守者と分類した。データは2011〜2012年に電話で収集され、死亡者については親族から収集された。脂質値の報告はなかった [4].
アウトカムと分母。 遵守した177名の患者のうち、112名がベースラインで狭心症を報告し、そのうち104名(本文では93%、表では105名の94%)が改善した。「逆転」は、画像診断または負荷試験により177名中39名(22%)で確認された。177名中18名(10%)が悪化したと分類された。調査員は、これらのイベントのうち9件は食事とは無関係であると判断した(医師の説得により無症状でバイパス手術を受けた2名、 ステント 血栓症 クロピドグレルの中止後、およびワルファリンの拒否後の脳卒中を含む)。4件は疾患の進行(脳卒中1件、バイパス手術2件、再ステント留置1件)と判断された。非心臓死が5名あり、心臓死はなかった。見出しにある0.6%というイベント率は、脳卒中のみを進行関連の重大イベントとしてカウントしている。同じ論文の付録表では、代わりに遵守者のイベント率を2.2%と記載している [4]。非遵守の患者21名のうち、13名(62%)に少なくとも1件のイベントが発生した。2名の 心臓突然死、心臓移植1件、虚血性脳卒中2件、ステント留置処置4件、バイパス手術3件、および内膜剥離術1件である [4].
なぜその対比が因果関係ではないのか。 このグループはランダム化されておらず、参加者は自ら求めてプログラムに参加しました。遵守状況は自己申告に基づいて事後に分類されました。研究者は電話によって事象を確認・判定し、どれが食事に関連しているかを決定しました。画像診断はプロトコルに基づくものではなく臨床上の必要性から行われ、薬剤の使用状況は記録されていませんでした。また、非遵守群のイベントのうち7つは、 血行再建術 処置(プロシージャ)であり、それらは症状や医師の判断に依存するものでした。遵守群と非遵守群の患者はベースライン時にも差があり、例えば男性の割合は93%対76%でした [4]。62%対0.6%という単純な対比は分割可能ですが、その商は妥当な因果関係を示す相対リスクあるいは 治療必要数比ではありません。グループはランダム化されておらず、それらに適用された事象の定義は比較可能ではありません。2007年以降に登録された参加者は、プログラムの一環として筆頭著者の著書のコピーを受け取りました [4].
10.3 エセルスティンの研究が裏付けられること、裏付けられないこと
エセルスティンのプログラムが注目に値する理由は3つあります。それは、ホールフード(未精製食品)という手段の最も集中的な現実世界での表現です。2014年のコホートでは、油なし、動物性食品なし、加工食品なし、そして豊富な葉物野菜が摂取されました。これは、重度の疾患を持つ意欲的な患者の中には、このような食事を何年も継続できる人がいることを示しています。1995年時点でも当初の患者22人のうち16人が継続しており [42]、2014年のコホートでは自己申告による遵守率が89%でした [4]。そして、葉物野菜の硝酸塩と一酸化窒素、内皮保護、油の排除といった、セクション5で独立した証拠に照らして評価される、特定の検証可能な仮説を生み出しました。
裏付けられないのは、「ApoBを超えた」という主張や優越性の主張に対する因果関係の重みです。1995年のコホートでは食事にコレスチラミンとロバスタチンを併用し、平均LDL-C 71.6 mg/dLを達成しました。このレベルでは、併用薬物療法が血管造影結果の一部を説明し得るもっともらしい理由となりますが、対照群のないデザインでは、薬物だけで説明がつくのか、あるいはこのLDL-Cが進行停止を保証するのかを確定することはできません [42]。画像診断が行われたのはわずか11名で、38の病変のうち13が除外されており、論文では脱落を2通りの方法で報告しています(図11)。2014年のコホートには脂質データがなく、遵守状況は自己申告、アウトカムは研究者による判定でした(セクション10.2)。それにもかかわらず、ランダム化脂質試験と併せて考えると、2つの観察結果の妥当性が高まります。ホールフードの食事に薬剤を併用して達成された持続的な極低LDL-Cは、画像診断を受けたほとんどの患者における血管造影上の疾患停止と適合しており [20, 21]、ランダム化されたオーニッシュ(Ornish)試験やハイデルベルク(Heidelberg)試験の結果と一致して、ほとんどの患者で狭心症が改善しました [6, 61, 62]。未解決なのは、食事による独立した寄与、油・ナッツ・アボカドを排除することに何らかの意味があるのか、代表的な集団における真の事象率、そして同様のサポートを受けた地中海食や高脂質のホールフードプログラムと結果がどのように比較されるかという点です。
11. ランダム化比較生活習慣試験(主にティア2)
11.1 ライフスタイル・ハート・トライアル(The Lifestyle Heart Trial)
デザイン. 適格となる可能性のある193名の患者のうち、血管造影後に93名が適格として残り、招待デザイン(実験群53名、対照群40名)によってランダム化されました。それぞれ28名と20名が参加に同意し、48名の試験参加者が得られました。35名(実験群20名、対照群15名)が5年間の 冠動脈定量解析法を完了し、割り付けを伏せた状態で読影されました [6]。介入内容は、脂質10%のホールフード菜食と中程度の 有酸素運動の運動、ストレス管理(1年目で1日87分、5年目で49分)、禁煙、およびグループによる心理社会的サポートを組み合わせたものでした。実験群の患者は脂質低下薬を服用しませんでした。対照群の15名中9名(60%)が1年目から5年目の間に服用を開始しました [6].
血管造影:どの報告、どのコホートか。 Two reports describe this trial and are often quoted interchangeably, which makes them look inconsistent. The 1990 Lancet report covers the first year in all 48 participants, analyzing 195 lesions: mean percent diameter stenosis fell from 40.0% (SD 16.9) to 37.8% (16.5) in the experimental group and rose from 42.7% (15.5) to 46.1% (18.5) in controls; among lesions more than 50% stenosed the change was 61.1% to 55.8% versus 61.7% to 64.4%; and 18 of 22 experimental patients (82%) changed in the direction of regression [63]. The 1998 JAMA report covers the 35 participants who completed five-year angiography, a subset with slightly different baseline values (38.92% experimental, 42.50% control), and reports changes rather than start-and-end values [6]. The figures below are from the 1998 report unless stated; a 1990 value and a 1998 value should never be compared directly, because they describe different patients.
Percent diameter stenosis changed by −3.07 percentage points (95% CI −5.91 to −0.24) in the experimental group and +11.77 points (95% CI 3.40 to 20.14) in controls at five years (P = 0.001), corresponding to a 7.9% relative improvement and a 27.7% relative worsening [6]. These are changes in lumen narrowing, not a 7.9% removal of plaque volume. Minimum lumen diameter was unchanged in the experimental group (+0.001 mm) and fell 0.34 mm in controls (P = 0.05). The reference (“normal”) segment diameter decreased slightly in the experimental group (−0.13 mm) and widened slightly in controls (P = 0.01) [6]. Because percent stenosis is calculated relative to that reference segment, a stable minimum diameter with a narrowing reference segment will register as reduced stenosis; the authors interpreted the reference-segment change as flow streamlining [6]. Within the experimental group, stenosis change tracked adherence tertiles (−6.81, −3.02, and −0.37 points; n = 6, 7, and 6), an observational analysis within a randomized trial [6].
Lipids. LDL-C fell from 143.8 to 86.6 mg/dL (−40%) at one year and was 115.4 mg/dL (−20%) at five years; triglycerides rose (227.8 to 258.2 mg/dL at one year), HDL-C fell (40.1 to 34.8 mg/dL at five years), and ApoB returned to baseline by year five as described in Section 5.6. LDL-C did not differ between groups at five years, largely because most controls took lipid-lowering drugs [6].
Events and angina. Over five years there were 25 cardiac events among 28 experimental patients and 45 among 20 controls (rate ratio for controls 2.47; 95% CI 1.48 to 4.20) [6]. These are recurrent-event counts, not numbers of patients with an event, and they are dominated by procedures and hospitalizations: myocardial infarction 2 versus 4, angioplasty 8 versus 14, bypass surgery 2 versus 5, cardiac hospitalizations (which include those events) 23 versus 44, and deaths 2 versus 1 [6]. Hard events were too few for comparison. Reported angina frequency fell 91% within the experimental group at one year (with a 42% fall in duration and a 28% fall in severity) and 72% at five years [6, 63]; between-group differences were no longer significant at five years because the most symptomatic controls had undergone revascularization [6]. The control-group change in angina frequency at one year is given as a 186% increase in the 1998 paper but as 165% in the 1990 report [6, 63]; the difference has not been reconciled and both are stated here.
Attrition and bias. Seven patients lacked one-year angiograms, and four in each group lacked five-year angiograms; fourteen lesions were unavailable overall—four in the experimental group and ten in controls, including four control lesions excluded after revascularization—which would bias toward the null [6]. A correction notice added an omitted author [6].
解釈. ランダム化 supports a causal reading of the comparison between the whole lifestyle package and usual care of the time, though post-allocation consent and attrition limit that confidence. The inability to isolate diet is a limitation of attribution, not an absence of randomization. Participation was selected after allocation: 93 patients were randomized by the invitational design and 48 then consented. Similar consent proportions in the two arms do not eliminate selection bias, and small size and attrition are further limitations.
11.2 PET perfusion (Gould 1995)
Gould and colleagues reported that, in the same 20 experimental and 15 control patients, the size and severity of perfusion abnormalities on rest–dipyridamole PET improved with the lifestyle intervention and worsened in controls, who received mainly antianginal therapy [61]. This is a functional outcome in the same patients, not an independent replication, and it is distinct from ventricular function and angiographic anatomy. Freeman and colleagues’ summary of a “400% increase in 心筋血流” could not be traced to this report and should not be reused [2].
11.3 The Multicenter Lifestyle Demonstration Project
This nonrandomized project followed 333 patients eligible for revascularization: 194 who chose the lifestyle program and 139 controls who underwent revascularization. At three years, 150 of 194 experimental patients (77%) had avoided revascularization, and rates of myocardial infarction, stroke, and death per patient-year were similar between groups [64]. The 77% describes procedure avoidance among self-selected patients; it is not a randomized 77% reduction in procedures or events. Freeman and colleagues’ statement that demonstration projects produced a “greater than 90% reduction in angina within weeks” cites this project and a 24-site program evaluation [2, 65]; the specific study, outcome, and time point could not be traced, and the verified one-year 91% figure from the Lifestyle Heart Trial cannot support a claim about weeks.
11.4 Lower-fat but non-vegan angiographic trials
In STARS, 90 men with coronary disease were randomized to usual care, a lipid-lowering diet (27% of energy from fat), or the diet plus cholestyramine, with angiography at 39 months. Mean absolute width of coronary segments narrowed 0.201 mm with usual care, was unchanged with diet alone (+0.003 mm), and widened 0.103 mm with diet plus cholestyramine; the proportion with progression was 46%, 15%, and 12% respectively [66]. In the Heidelberg trial, 113 men were randomized to usual care or intensive exercise plus a low-fat, low-cholesterol diet without lipid drugs; at one year, lesions progressed in 23% versus 48% and regressed in 32% versus 17% [62]. At six years, 90 patients were re-evaluated, and lipid differences between groups were no longer significant [67]. In ディスコCT, 92 patients with nonobstructive disease on optimal medical therapy were randomized to added intensive dietary counseling with a DASH-type pattern and activity checks, or to medical therapy alone; after about 67 weeks, noncalcified plaque volume fell more with the intervention (−51.3 versus −21.3 mm³; P = 0.045), although the change in total 粉瘤 volume did not differ significantly between groups [68]. At about six years, most of the weight lost during the intervention had been regained in both groups, and one major adverse cardiovascular event had occurred in the DASH group versus four (including one fatal myocardial infarction) among controls—too few events for inference [69]. These trials show that omnivorous or lower-fat regimens well above 10–15% fat also slowed progression.
12. The Strongest Competing Evidence (Tiers 1–2)
12.1 DASH and DASH-Sodium
In controlled feeding of 459 adults for eight weeks, the DASH combination diet lowered systolic and diastolic pressure by 5.5 and 3.0 mm Hg more than a typical American control diet, and by 11.4 and 5.5 mm Hg in the 133 participants with hypertension [11]. Combining DASH with low sodium lowered systolic pressure by 7.1 mm Hg in normotensive participants and 11.5 mm Hg in those with stage 1 hypertension, compared with the high-sodium control diet [16]. These benefits were achieved with low-fat dairy in the diet.
12.2 PREDIMED
The 2013 PREDIMED report was retracted because of randomization irregularities and republished in 2018 with reanalysis. Among 7,447 adults at high cardiovascular risk followed for a median of 4.8 years, major cardiovascular events occurred in 96 of 2,543 participants (3.8%) assigned a Mediterranean diet plus extra-virgin olive oil, 83 of 2,454 (3.4%) assigned a Mediterranean diet plus nuts, and 109 of 2,450 (4.4%) in the control group advised to reduce dietary fat (hazard ratios 0.69, 95% CI 0.53 to 0.91, and 0.72, 95% CI 0.54 to 0.95); results were similar after excluding 1,588 participants with known or suspected protocol departures [12]. The control arm received advice, not a very-low-fat plant-based diet.
12.3 The Lyon Diet Heart Study
In 605 survivors of a first myocardial infarction followed for a mean of 46 months, the Mediterranean-type diet reduced cardiac death plus nonfatal infarction (composite 1: 14 versus 44 events), composite 1 plus 不安定狭心症、脳卒中、 心不全, and embolism (composite 2: 27 versus 90), and composite 2 plus minor events requiring hospitalization (composite 3: 95 versus 180) [70]. The abstract reports adjusted risk ratios ranging from 0.28 to 0.53 across the three composites; the supplementary table of the 2014 Esselstyn article assigns 0.28 (95% CI 0.15 to 0.53) to composite 1 [4, 70]. Summaries of “up to 65%” or “70%” reductions do not match any single endpoint and should be avoided. The comparator was a prudent Western-type diet rather than a healthy active diet [10], and the intervention supplied an alpha-linolenic-acid-rich margarine in place of butter and cream [13]. Lyon shows that a plant-rich but not plant-exclusive diet with substantial unsaturated fat reduced recurrent events.
12.4 CORDIOPREV
CORDIOPREV randomized 1,002 patients with coronary heart disease to a Mediterranean diet (at least 35% fat; 40–60 g/day of extra-virgin olive oil) or a low-fat, high-complex-carbohydrate diet (less than 30% fat, less than 10% saturated fat, lean meat and low-fat dairy permitted), with equally intensive dietitian support and a median seven-year follow-up. Achieved fat was 40.5% of energy in the Mediterranean arm and 32.1% in the low-fat arm. The primary composite occurred in 87 (17.3%) versus 111 (22.2%) patients (unadjusted hazard ratio 0.745; 95% CI 0.563 to 0.986; adjusted models 0.719 to 0.753). Benefit was statistically demonstrable in men (hazard ratio 0.669; 95% CI 0.489 to 0.915), while the estimate in the smaller subgroup of 175 women was inconclusive rather than showing absence of benefit; no single component differed significantly; lipids and glucose did not change differently between diets; and 86.6% took statins at baseline. More low-fat participants abandoned their diet (17.2% versus 9.2%). The trial was funded principally by olive-oil foundations [10].
CORDIOPREV challenges the broad claim that less total fat is always better: within the 30–40% range, among omnivores on statins, the higher-fat Mediterranean diet did better. It does not test a 10–15% fat, animal-free, whole-food diet, which differs from its low-fat arm in fat level, animal-food content, and food processing.
12.5 Direct comparisons of plant-exclusive and Mediterranean diets
No trial has compared these diets on clinical events, and the three risk-factor trials below provide related but distinct comparisons rather than one comparison repeated. In a 16-week crossover trial in 62 overweight adults (52 completers), a low-fat vegan diet reduced body weight by 6.0 kg with no change on a PREDIMED-style Mediterranean diet (treatment effect −6.0 kg; 95% CI −7.5 to −4.5), and lowered LDL-C by 15.3 mg/dL with no significant change on the Mediterranean diet (treatment effect −14.8 mg/dL; 95% CI −23.5 to −6.2), whereas the Mediterranean diet lowered systolic pressure more (treatment effect +6.0 mm Hg favoring Mediterranean; 95% CI 1.0 to 10.9) [23]. The lipid estimate covers the 43 participants with no change in lipid-lowering medication and the blood-pressure estimate the 41 with no change in antihypertensive medication; a significant carryover effect for systolic pressure appeared in the analysis of all participants, including those whose medication changed, but not in the stable-medication subgroup [23]. Reported fat intake at the end of the vegan phase was 17% of energy (95% CI 15 to 19), so the trial tested a low-fat vegan diet rather than a 10–15%-fat one, and one participant’s sequence assignment was changed after randomization so that a mother and daughter followed the same diet [23]. The trial was conducted by an organization that advocates plant-based diets. In CARDIVEG, a three-month crossover in 107 low-risk overweight adults, energy-restricted lacto-ovo-vegetarian and Mediterranean diets produced similar weight and fat loss; LDL-C fell more on the vegetarian diet, triglycerides fell more on the Mediterranean diet, and vitamin B12 fell on the vegetarian diet [71]. EVADE CAD compared a vegan diet with the AHA diet at similar fat intakes (Section 5.3) [9]. Only Barnard compared a vegan with a Mediterranean diet; CARDIVEG compared a lacto-ovo vegetarian diet with a Mediterranean one, and EVADE compared a vegan diet with the AHA diet. Read separately, each found the plant-based arm lowering LDL-C at least as much, with the Mediterranean comparator better on blood pressure (Barnard) or triglycerides (CARDIVEG); these are not three replications of a single comparison.
12.6 The Cochrane review and the event gap
The 2021 Cochrane review of vegan diets (search to February 2020) included 13 randomized trials with at least 12 weeks of follow-up; none reported cardiovascular clinical endpoints. Only one trial, of 63 participants, addressed 二次予防 against another dietary intervention, and showed no clear effect on lipids or blood pressure (low- or very-low-certainty evidence) [72]. Its stricter eligibility—vegan only, minimum 12 weeks, active or minimal-intervention comparators—explains why its conclusions are more cautious than those of the Koch and Wang meta-analyses, which admitted vegetarian diets and different comparators [7, 22]. An updated search for this review, through 10 September 2026, found no published randomized trial of a very-low-fat vegan diet with cardiovascular events as an outcome. The absence of event evidence is a gap, not evidence of no benefit. Pooled analyses also need checking for cohort overlap: the 1990 one-year and 1998 five-year Lifestyle Heart Trial reports describe the same randomized cohort, so an analysis that enters them as separate trials double-counts one cohort and inflates precision.
Direct comparisons are used wherever available. Diets are not ranked by comparing effect sizes across unrelated trials, because populations, background therapy, adherence, and endpoints differ.
Figure 12 displays the randomized trials with clinical outcomes discussed in this section alongside the Lifestyle Heart Trial’s recurrent-event ratio.

Figure 12. Randomized dietary trials with clinical outcomes. Estimates are shown side by side, not pooled; comparators, populations, and endpoints differ. No randomized trial of a very-low-fat vegan diet has reported clinical events.
12.7 比較対照試験への同一基準の適用
本レビューではエセルスティンとオーニッシュの研究を詳細に検討してきたが、地中海食を支持する試験も同様の精査に値する。表2では、各試験に対して同じ質問を投げかけている。その目的は、特定の試験を無価値と決めつけることではなく、各試験が何を立証できるのかを明確にすることである。
PREDIMED。 2018年の再出版文書自体が主な問題を記録している。割り付けを隠蔽するための密封封筒の使用はパイロットフェーズの一部のみであった。425人の世帯員がランダム化なしに登録され、親族と同じ食事が与えられた。ある施設では、467人の参加者が個人単位ではなくクリニック単位で割り当てられた。また別の施設では、ランダム化テーブルが不整合に使用されていた [12]。著者らは統計的な調整を行って試験を再解析したが、影響を受けた1,588人の参加者を除外しても結果は同様であった(残りの5,859人におけるハザード比は0.71および0.68) [12]。対照群は2006年9月まで低脂肪のアドバイスが記載されたリーフレットを年1回受け取るのみで、それ以降になってようやく地中海食群と同じ頻度の接触が行われるようになった。対照群の総脂肪摂取量はほとんど変化しなかったため、PREDIMEDはサプリメントを補った地中海食を、わずかなアドバイスを受けただけの通常の食事と比較したものであり、真の低脂肪食と比較したわけではない [12]。対照群への接触が増えた後に募集された参加者の間でも有益性が認められた。ハザード比は、2006年10月の変更前が0.77(95% CI 0.59~1.00)であったのに対し、変更後は0.49(95% CI 0.26~0.92)であった(不均一性のP = 0.21)。これは、不平等なサポートのみが唯一の説明要因であるという考えに反論するものであるが、募集期間の比較だけでは、その影響を完全に排除することはできない [12]。試験の中止率は対照群で11.3%であったのに対し、地中海食群では4.9%であった。オリーブ油とナッツは生産者から寄付され無料で提供された。参加者は自らの割り当てを知っており、エンドポイント委員会のみが盲検化されていた [12]。試験は第4回中間解析で中止された [12]。有益性のために早期中止された試験は、効果量を過大評価する傾向がある。最後に、複合結果は主に脳卒中(統合ハザード比 0.58; 95% CI 0.42~0.82)によってもたらされたものであり、心筋梗塞(0.80; 0.53~1.21)、心血管死(0.80; 0.51~1.24)、全死因死亡(0.98; 0.77~1.24)には有意な差が認められなかった。著者らはこれを検出力の不足によるものとしている [12](図13)。

図13. PREDIMED 2018:複合結果とその構成要素(合算した地中海食群対対照群)、および2006年10月の対照群へのサポート強化前後の効果。出典:Estruch 2018、表3および本文。
リヨン・ダイエット・ハート・スタディ(Lyon Diet Heart Study)。 リヨン試験では、初回心筋梗塞の生存者605人を単盲検デザインでランダム化し、平均27ヶ月で有益性のために早期中止された。その後、追跡調査は46ヶ月まで延長された [13, 14, 70]。その比較対象は、集中的なサポートを受けた健康的な食事ではなく、慎重な欧米型の食事であった [70]。イベントの大幅な減少は、血清脂質、血圧、BMIが群間で同様のままである中で発生した [14]。これは、食事パターンが脂質低下以外の経路を通じてイベントを減少させうるという真の兆候である(セクション5)が、効果量が誇張されている可能性が高い、早期中止された小規模な試験の結果である。
CORDIOPREV。 CORDIOPREVは、主にオリーブ油財団から資金提供を受けた単一施設試験であった。有益性が統計的に証明されたのは男性のみで、より人数の少ない女性のサブグループでは決定的な推定値が得られなかった。複合項目の構成要素で単独で有意なものはなかった。また、その低脂肪食群の脂肪エネルギー比率は32.1%に達していた [10]。両群が等しく集中的なサポートを受けたため、3つの試験の中で最も強固なデザインであるが、比較対象としているのは中脂肪の雑食であり、超低脂肪の植物性食事ではない。
栄養科学全般に共通する問題。 このレビューのほぼすべての研究には、どの食事法を支持しているかにかかわらず、いくつかの弱点があります。食事法は通常、質問票によって測定されます。アドベンチスト・ヘルス・スタディ2において、繰り返しの24時間思い出し法に対する質問票の妥当性の相関は、...に対して0.76、 赤身肉 しかし、白人の参加者の魚に関しては0.53でした [53]。食事試験の参加者は、自分が何を食べているかを知っています。PREDIMEDにおける対照群の中止、CORDIOPREVにおける食事の断念、Lifestyle Heart Trialにおける遵守スコアの低下に示されるように、遵守率は低下します [6, 10, 12]。比較対象が弱いことが多いです。PREDIMEDの初期の対照群、リヨンの慎重な食事、Lifestyle Heart Trialの通常のケアなどが挙げられます。試験は小規模であったり、早期に終了したりすることが多いです。また、利害関係はあらゆる方向に及びます。オリーブオイルやナッツの生産者、プログラム開発者、啓発団体などが、ここで議論されている研究に資金を提供したり、主導したりしています [4, 10, 12, 23].
批判によって何が変わるか。 これらの限界により、地中海食の利益の大きさや、地中海食が優れていることが証明されたという主張に対する信頼性は低下します。しかし、臨床イベントを伴うランダム化比較試験がまったくないVLF-WFPBの地位を高めるものでもありません。臨床イベントを伴うランダム化比較試験は、たとえ欠陥があったとしても、対照群のないケースシリーズでは対処できない因果関係の問いに答えることができます。ただし、その重みは依然としてバイアス、精度、および当該の食事法をいかに直接的にテストしているかに依存します。両方の陣営に同じ懐疑の目を向けることが、このレビューの結論を信頼できるものにするのです。
表2. 論争の両陣営の主要な試験に同じ質問を適用する。
| 質問 | PREDIMED | コディオプレブ | リヨン式ダイエット心臓病食 | ライフスタイル心臓トライアル | Esselstyn 2014 コホート |
| エビデンスの階層 | 1 | 1 | 1 | 2(少ないイベント数) | 4 |
| ランダム化されているか? | はい、7,447人中1,588人に影響する逸脱あり。封筒法はパイロット試験の一部でのみ使用 [12] | はい [10] | はい [14] | はい、招待制。適格者の半分が辞退 [6] | いいえ [4] |
| 誰がブラインド化されたか? | エンドポイント委員会のみ [12] | エンドポイント 判決 マスク化されたと報告されている。他の食事試験と同様、参加者と栄養士はブラインド化されていない。 | 単盲検デザイン [14] | 単盲検デザイン [14] | なし。研究者がイベントを判定 [4] |
| 比較対象 | 脂肪を減らすようアドバイス。2006年まで年1回のリーフレット配布、その後は同等の接触。総脂肪はほとんど変化せず [12] | 同様にサポートされた低脂肪食。脂肪摂取率32.1%を達成 [10] | 慎重な欧米型食事 [70] | 通常のケア。60%が脂質低下薬を開始 [6] | 遵守していないボランティア [4] |
| 早期終了したか? | はい、第4回中間解析時 [12] | 追跡期間中央値7年 [10] | はい、27ヶ月時点。後に46ヶ月まで延長 [13, 14] | はい、27ヶ月時点。後に46ヶ月まで延長 [13, 14] | 該当なし |
| 何が結果を導いたか | 脳卒中 0.58。心筋梗塞 0.80および心血管死 0.80は有意差なし [12] | 男性で利益が証明可能。女性175名での推定値は決定的ではない。単一の成分で有意なものはなし [10] | 血清脂質、血圧、BMIが同等でも大幅な減少 [14] | 処置および入院(再発回数) [6] | 非遵守群における血行再建術 [4] |
| 利害関係 | 生産者から寄贈されたオリーブオイルとナッツ [12] | 主にオリーブオイル財団 [10] | ここでは評価せず | ここでは評価せず | プログラム開発者。参加者に本を配布 [4] |
| VLF-WFPBと比較された食事法 | 補完された地中海食 vs 軽いアドバイスを受けた通常の食事 | 地中海食(脂肪40.5%) vs 雑食の低脂肪食(脂肪32.1%) | ALA(α-リノレン酸)豊富な地中海型 | ALA豊富なマーガリン+地中海型のアドバイス | オイルフリー、動物性食品なし、ナッツやアボカドなし |
13. 総脂肪が非常に低いこと自体が良いのか?
脂肪を区別する。 飽和脂肪はLDLコレステロールを上昇させ、その削減はイベントを減少させます [26]; 工業用トランス脂肪酸は避けるべきです [73]; 不飽和脂肪は、飽和脂肪の代わりに使用するとLDLコレステロールを低下させ、地中海食の試験ではイベントの減少に関連しています [10, 12]。「総脂肪」は、これらの相反する効果をひとまとめにしています。
脂肪を何に置き換えるかが重要です。 VLF-WFPBでは、置き換えられるのは未加工の穀物、豆類、野菜、果物です。他の低脂肪食では、精製されたでんぷんや糖分である可能性があります。植物性食品のみの食事において、これらの置き換えが臨床アウトカムに及ぼす影響を比較した試験はありません。
ホールフードの植物性食事における唯一の直接的なテスト。 In a randomized crossover trial, 40 adults with at least 5% estimated cardiovascular risk followed a whole-food plant-based vegan diet with either about four tablespoons or less than one teaspoon of extra-virgin olive oil daily, for four weeks each; fat supplied 48% and 32% of energy respectively. Both phases lowered LDL-C, total cholesterol, ApoB, HDL-C, glucose, and hs-CRP from baseline. There was a sequence interaction: moving from high to low oil lowered LDL-C by 12.7 mg/dL (P = 0.04), and moving from low to high raised it by 15.8 mg/dL (P = 0.02); in the first period, LDL-C fell 25.5 versus 16.7 mg/dL (P = 0.162) [8]. Neither phase approached 10–15% fat, the trial was short, and carryover complicates the crossover analysis. It suggests large oil additions can blunt LDL-C lowering; it does not show that 10–15% fat is better than a nut- and seed-based plant diet at around 30%.
Nuts, seeds, and olive oil. Mediterranean diets rich in extra-virgin olive oil or supplemented with nuts reduced events against comparators [10, 12]. No trial shows that excluding nuts, seeds, or olive oil from an otherwise low-saturated-fat diet improves clinical outcomes. The acute impairment of flow-mediated dilation after a single high-fat meal [30], cited in support of oil exclusion [4], cannot establish that oils cause atherosclerosis.
Triglycerides, HDL, and ApoB–LDL-C 不一致. Very-low-fat, high-carbohydrate diets can raise triglycerides and lower HDL-C, as in the Lifestyle Heart Trial’s first year [6]; across plant-based trials overall, triglycerides did not change [22], and Mediterranean diets lowered triglycerides more than a vegetarian diet in CARDIVEG [71]. When triglyceride-rich lipoproteins rise, LDL-C can understate 動脈硬化誘発性粒子数. The Lifestyle Heart Trial illustrates this: at five years LDL-C was 20% below baseline while ApoB was not [6]. Future trials should measure ApoB directly.
Attribution. Where VLF-WFPB outperforms a comparator on LDL-C, the advantage may reflect fat quantity, fat quality (almost no saturated fat), food quality (fiber, plant protein), weight loss, or adherence. The Barnard crossover, for example, produced both larger LDL-C reductions and larger weight loss on the vegan diet [23]. No available trial separates these contributions. On current evidence, very low total fat is not established as an independent requirement; very low saturated fat within a whole-food diet is the component with the strongest causal support.
14. Prevention, Treatment, and Reversal
14.1 Outcomes that must not be conflated
Fewer cardiovascular events, slower progression, プラーク安定化, improved angina or perfusion, reduced angiographic stenosis, and quantitatively measured regression of plaque volume are different outcomes. Primary and secondary prevention are also different settings.
14.2 What imaging can and cannot show
Quantitative coronary angiography measures lumen, not plaque. Percent stenosis depends on the reference segment and on 血管運動神経緊張, and outward (positive) remodeling can hide substantial plaque behind a normal lumen. Myocardial perfusion and stress testing measure function; improvement can occur without anatomic regression, as the Heidelberg investigators noted [62]. Intravascular ultrasound そして 冠動脈CTアンギオグラフィー measure plaque volume and composition; CT additionally characterizes noncalcified and 低吸収プラーク. Coronary calcium scores can rise when plaque stabilizes, because statins increase calcified and fibrous plaque while reducing fibrofatty and necrotic-core components, so calcium-score change alone indicates neither success nor failure [74]. Plaque can enlarge substantially before angiography detects narrowing [74]; conversely, an apparently wider lumen can reflect changes in vasomotor tone rather than smaller plaque. In a meta-regression summarized by Dawson and colleagues, each 1% reduction in プラーク体積百分率 was associated with about 20% lower odds of major adverse events, but there is no direct evidence that regression itself reduces events [74]. Table 3 summarizes what each finding can and cannot establish.
Table 3. What symptom and imaging findings can and cannot establish.
| 所見 | What it establishes | What it does not establish |
| Less angina; better ストレステスト or PET perfusion [6, 61, 62] | Better symptoms or myocardial blood supply | A measured loss of plaque |
| Reduced angiographic percent stenosis | A less narrowed lumen relative to a reference segment | The percentage of plaque removed; reference-segment changes can alter the result [6] |
| Change in minimum lumen diameter | Absolute lumen change at the lesion | Plaque volume, which 外向きの再構築 can hide [74] |
| CCTA or IVUS plaque volume or composition change | Change in the measured plaque compartment, within acquisition and segmentation limits | Eradication of disease or a guaranteed event reduction |
| Rising coronary calcium score | A higher アガトンスコア, which reflects calcified area, density, or both | Greater total plaque burden or treatment failure; statins increase 石灰化プラーク while reducing fibrofatty and 壊死核 [74] |
14.3 The primate and pharmacologic benchmarks
Two bodies of evidence set the standard against which dietary regression claims should be read. Neither is a dietary trial in humans, and both are cited here as benchmarks rather than as support for any diet.
Controlled primate experiments (Tier 5). In rhesus monkeys given an atherogenic diet and then switched to a regression diet, atheromatous 冠動脈 lost lipid [75], while arterial fibrous proteins behaved differently from the lipid compartment [76]. The achieved cholesterol level mattered: in animals whose plasma cholesterol was held near 300 mg/dL there was little or no increase in lumen area attributable to regression of intimal plaque, whereas at about 200 mg/dL a considerable share of the lumen gain was attributable to plaque regression, and the contribution depended on both plaque size and time [77]. This is the closest thing in the literature to a controlled demonstration that lowering atherogenic lipoproteins by dietary means depletes the lipid-rich compartment while denser tissue persists. It is also animal evidence, in a different species, with induced rather than spontaneous disease, and it cannot establish what a human diet achieves.
Pharmacologic imaging (Tiers 1–2). イン グラゴフ, 968 statin-treated patients with angiographic coronary disease were randomized to monthly エボロクマブ または プラセボ for 76 weeks, and 846 had evaluable serial intravascular ultrasound. Time-weighted LDL-C was 36.6 mg/dL versus 93.0 mg/dL, percent atheroma volume changed by −0.95% versus +0.05% (difference −1.0 percentage point; P < 0.001), and plaque regression occurred in 64.3% versus 47.3% of patients [78]. Notably, the trial’s virtual-histology substudy did not show a difference in dense calcium volume [78], so GLAGOV demonstrates volumetric regression rather than a compositional shift; trials designed to measure plaque composition are outside the scope of this review and no compositional figures are quoted. Two comparisons follow. No dietary trial has produced coronary regression of this magnitude under this level of rigor, and no dietary trial identified here used serial intravascular ultrasound at all, which is the modality that measures atheroma volume most precisely.

Figure 14. Benchmarks for coronary plaque change. Pharmacologic regression measured by serial intravascular ultrasound (GLAGOV) alongside the dietary imaging evidence available in this review, which uses different modalities and cannot be placed on the same axis. No dietary trial identified here used serial intravascular ultrasound. Sources: Nicholls 2016; Henzel 2021; Ornish 1998.
Exercise as a co-lever. Physical activity belongs alongside the three dietary levers rather than inside them. The populations with the least coronary calcium are also the most active: Tsimane men average six to seven hours and women four to six hours of physical activity a day [35]. In the two randomized imaging trials that showed benefit, exercise was part of the intervention rather than a controlled covariate—the Lifestyle Heart Trial bundled aerobic exercise with the diet [6], and DISCO-CT bundled activity counseling with DASH [68]—which is why neither can apportion credit. No quantitative estimate of the ApoB reduction attributable to aerobic training is given here; the point stands qualitatively that diet, activity, and lipid-lowering therapy act on overlapping pathways and are complementary rather than competing.
14.4 What the evidence supports in each domain
一次予防. Evidence consists of cohort associations [40], population data (Sections 6–8), and randomized lipid trials [22]. No randomized event trial of VLF-WFPB exists.
Secondary prevention (treatment). The Lifestyle Heart Trial reduced angina and cardiac procedures and hospitalizations for a package that included a very-low-fat vegetarian diet [6]. Randomized event reductions with diet alone in secondary prevention come from Mediterranean-type trials [10, 70].
Slowed progression and partial regression. Intensive lifestyle programs incorporating very-low-fat vegetarian diets have demonstrated modest 血管造影上の退縮 in selected patients; the independent contribution of diet, and of total-fat restriction, remains uncertain [6, 42]. Lower-fat but non-vegan diets also slowed angiographic progression [62, 66], and a DASH-type program added to medical therapy reduced noncalcified plaque on CT [68]. Partial regression appears most likely when large, sustained reductions in atherogenic lipoproteins are achieved, often with drugs, and more regression was seen at five years than at one year in the Ornish program [6]. Regression does not mean eradication of disease or zero future risk.
The Dawson review. As summarized in the ACC’s published key points, the 2022 JACC state-of-the-art review judged the demonstrated effect of diet, exercise, and smoking cessation on coronary plaque volume and composition to be limited [74]. Claims sometimes attributed to that review in secondary summaries, including the assertion that most statin-treated dietary arms progressed, are not used here. Limited diet-only regression evidence does not mean dietary treatment has little clinical value: stabilizing plaque, improving perfusion, and preventing progression can be clinically meaningful even when anatomic change is small, although existing diet trials do not establish the mechanism of any event reduction.
15. Feasibility, Adequacy, and Safety
Adherence. Reported adherence has been high in selected volunteers: 89% by self-report in the 2014 Esselstyn cohort [4], and 94% versus 70% for vegan versus AHA diets at eight weeks in EVADE CAD, whose participants represented 14% of those meeting initial criteria [9]. In the Lifestyle Heart Trial, half of eligible invited patients declined, and the adherence score fell from 1.29 at one year to 1.06 at five years [6]. Over seven years in CORDIOPREV, more patients abandoned a conventional low-fat diet than a Mediterranean diet [10].
Nutrients. Vitamin B12 supplementation is essential; median intake in the EVADE vegan arm fell to 1.2 µg/day, and B12 levels fell during the vegetarian phase of CARDIVEG [9, 71]. Zinc and omega-3 fatty acid intakes were also lower on the vegan diet in EVADE [9]. Calcium, vitamin D, iodine, iron, and long-chain omega-3 status require planning, and fortified foods or supplements are appropriate where intake is inadequate [79]; “whole food” should not become a reason to reject them. Unfortified terrestrial plant foods supply alpha-linolenic acid but little or no preformed EPA or DHA, which algae-derived supplements and fortified products can provide within a vegan diet [79]; Esselstyn’s protocol added flaxseed as an alpha-linolenic acid source [4]. Aggressive fat restriction should not displace necessary energy or essential fats.
Energy and protein. EVADEヴィーガン食におけるタンパク質摂取量の中央値は1日あたり50g(エネルギーの12.9%)であった [9]。エネルギー制限のない菜食は、通常の食事と比較してエネルギー摂取量を1日あたり約276 kcal減少させた [7]。また、2014年のエセルスティン・コホートでは、体重データのある135名の患者が平均18.7ポンド(約8.5kg)減量した [4]。低エネルギー密度は減量には有利であるが、食欲不振や サルコペニア を抱える高齢者、および必要量の多い活動的な人々にとっては潜在的な危険性となる。タンパク質とエネルギーの充足度は、個別に計画されるべきである。
生活の質(QOL)。 EVADE CAD試験において、QOLスコアは両方の食事で同様に改善した [9].
食事と薬剤。 統合されたランダム化エビデンスによると、食事によるApoBの減少は約14%である [22]。強力なスタチンや PCSK9 療法下でのApoB低下の対応する推定値はここでは示されていないため、数値的な比較は行わない。臨床的な論点は他の根拠に基づいている。確立されたASCVDにおいて、食事はガイドラインに基づいた療法の補完であり、代替ではない。特定の食事に従うために、指示された療法を保留したり中止したりすべきではない。
このレビューは教育目的のものである。食事の変更は、処方された心血管療法に代わるものではなく、それと共に行われるべきであり、薬剤の調整は担当医が行うべきものである。
16. 臨床的位置付け
2023年のAHA/ACC慢性冠疾患ガイドラインは、心血管イベントを減少させるために、野菜、果物、豆類、ナッツ、全粒穀物、および低脂肪タンパク質を強調した食事を推奨しており(クラス1、レベルB-R)、飽和脂肪をエネルギーの6%未満に抑え、それを一価不飽和脂肪、多価不飽和脂肪、複合炭水化物、および食物繊維に置き換えることが有益である(クラス2a)と述べている [73]。その栄養に関する図では、 一価不飽和脂肪 オリーブオイルや多価不飽和脂肪などを、選択すべき食品として挙げている [73]。 2026年 ACC/AHA 多学会合同脂質異常症ガイドライン(2018年のコレステロールガイドラインに代わるもの)は、好ましい食事パターンを主に植物ベースのもの、すなわち地中海食、DASH食、ヴィーガン食、または菜食であると説明している [27].
AHAの2026年版食事ガイダンスには、健康的なタンパク質源の選択、飽和脂肪の代わりとなる不飽和脂肪源の選択、最小限に加工された食品の選択、 添加糖の最小化、およびナトリウムの削減を含む、心臓に良いパターンの特徴が挙げられている [25].
これらのガイドラインは、植物中心の食事療法を支持し、ヴィーガンパターンにも対応している。それらは、唯一有効な超低脂肪ヴィーガン逆転プロトコルを承認しているわけではなく、ナッツや不飽和油を明示的に含んでいる。専門家によるパターンの推奨は、特定の食事におけるすべての構成要素や除外が必要であるという証拠ではない。
17. 決定的な次なる研究
主要なデザイン。 CTで確認された冠状動脈性動脈硬化症を有する、治療中の成人を対象とした、盲検化されたエンドポイント評価を伴うランダム化並行群間比較試験。意図的に女性、高齢者、および専門的なライフスタイルプログラム以外の患者を募集する。群:(A) VLF-WFPB食(脂肪からのエネルギー10〜15%、動物性食品なし、油の添加なし、ナッツとアボカドを制限)、(B) より不飽和脂肪の多いホールフード植物ベース食(ナッツ、種子、アボカド、およびエクストラバージンオリーブオイルからの脂肪約30〜40%、動物性食品なし)、(C) 地中海食(PREDIMED/CORDIOPREV形式)またはDASH食。A対Bは、動物性食品を含まない食事内での総脂肪制限を検証し、B対Cは動物性食品の除外の効果を推定する(ただし代替食品は異なる)。すべての群において、接触時間、調理指導、食事サポート、および運動のアドバイスは同一とする。
メカニズム解明のためのサブスタディ。 3群構成では動物性食品の除外と総脂肪を完全に分離することができないため、制御された食事による2×2要因サブスタディ(動物性食品なし対特定の限定的な動物性食品、および脂肪からのエネルギー約10〜15%対30〜35%を交差させ、飽和脂肪、ナトリウム、食物繊維、タンパク質、およびエネルギーを可能な限り一致させる)を行うことで、安定した体重条件下で各要因を分離する。第3の要因として、主要栄養素を一致させたホールフード対超加工植物性食品を用い、入院患者を対象とした加工試験のデザインに従って、加工の程度による影響を直接検証する [19]. 代替食品における不可避的な差異を測定し、報告しなければならない。
2つの推定母数(Estimands)。 体重や遵守状況への影響を含む、自由摂取(ad libitum)パターンを割り当てることの実用的(pragmatic)な効果は、安定した体重における食事組成の生物学的効果とは異なる。主要解析において体重減少の影響を調整して排除すると、割り当てられた食事の効果の一部が失われることになる。給食サブスタディでは組成について検討する。
薬物療法。 すべての群が同じ臨床治療アルゴリズムに従い、食事の対照を際立たせるために適応となる治療が控えらえることは決してない。脂質目標値に向けた用量調節(タイトレーション)により、ApoBにおける食事の差異が一部相殺されるため、必要とされる脂質低下療法の強度が事前に規定された副次評価項目となり、ApoBは薬剤の強度とともに解析される。本試験は、共通の治療アルゴリズム下での比較有効性を推定するものであり、それ自体が累積的なApoB曝露量を均等にするわけではない。また、残留するベネフィットをApoBに依存しない経路に帰属させるには、媒介因子の反復的な縦断的測定と明確な因果関係の仮定が必要となる。
アウトカム(評価項目)。 第1段階の主要評価項目:冠動脈CT血管造影におけるアテローマ容積率(または非石灰化プラーク容積)の24ヶ月間の変化。これは症状に基づかず、プロトコルのスケジュールに従い、盲検化されたコアラボラトリーで読影される。副次評価項目:時間加重平均ApoB、非HDLコレステロール、トリグリセリド、自由行動下血圧、体重および体組成、HbA1c、高感度CRP、TMAO、PET心筋血流、狭心症(シアトル狭心症質問票)、機能、QOL、コスト、継続率、および栄養状態(B12、鉄、ビタミンD、ヨウ素、オメガ3指数)。遵守状況は、繰り返しの食事記録に加え、客観的指標(血漿脂肪酸プロファイル、カロテノイド、尿中窒素およびカリウム)によって測定される。第2段階:心血管死、心筋梗塞、虚血性脳卒中のハードな複合エンドポイントを用いたイベント試験。手続きは盲検下で判定され、初回イベント発生までの時間(time-to-first-event)と再発イベントの両方の解析とともに、別途報告される。
解析。 欠測データおよび投薬変更の事前規定された処理を伴う、ITT(Intention-to-treat)主要解析。副次解析としてCACE(Complier-Average Causal Effect)解析およびPP(Per-protocol)解析。遵守状況の解析では、遵守者を新たにランダム化された群として扱うことはしない。ApoB、体重、血圧を用いた媒介分析は探索的として分類される。
例示的なサンプルサイズ(前提条件を記載)。 第1段階:アテローマ容積率の24ヶ月間の変化について、標準偏差を3.0パーセントポイント、群間の最小重要差を1.0ポイントと想定し、2つの主要な二群間比較(A対B、B対C)をそれぞれ両側α = 0.025、検出力90%で検定する場合、各群約224名、脱落率20%を考慮すると各群280名(計840名)が必要となる。第2段階:両側α = 0.025、検出力80%で、各二群間比較において0.80のハザード比を検出するには、比較ごとに約764のイベントが必要である。対照群の4年間の初回イベントリスクを15%と想定すると、各群約2,800名(計約8,400名)が必要になる。これらの数値は例示であり、プロトコルには遵守状況、クロスオーバー、薬物増減に関する仮定が必要となる。また、これらはなぜイベントに関するエビデンスがまだ存在しないのかを説明している。
18. ドメイン別の結論
表4(本文後)はエビデンスベースをまとめており、表5は提案された各利点が対照ダイエットと共有されているか、あるいは追加のベネフィットのエビデンスがあるかを示している。
ラベル: 確立されている—記載されたアウトカムについて一貫したランダム化比較試験のエビデンスがある。 支持されているが不確実—小規模、多成分、または中間アウトカムに関するランダム化比較試験のエビデンス、あるいは一貫した観察研究のエビデンスがある。 生物学的に妥当—十分なアウトカムのエビデンスはないが、メカニズム的または間接的な支持がある。 支持されていない—十分なエビデンスがない、あるいは否定的なエビデンスがある。
予防。 植物ベースの食事が雑食の食事と比較してLDL-CおよびApoBを低下させることは、 確立されている (pooled LDL-C difference about 11.6 mg/dL) [22]. That VLF-WFPB lowers ASCVD incidence relative to a typical Western diet is supported but uncertain: cohort associations (vegetarian IHD relative risk 0.79) are consistent with lipid-mediated plausibility, but there is no randomized event evidence [40]. That it prevents ASCVD better than a Mediterranean or DASH diet is unsupported. Coronary and stroke outcomes should be reported separately: in EPIC-Oxford, vegetarians had lower IHD but higher hemorrhagic and total stroke rates [41].
治療. That an intensive lifestyle package including a very-low-fat vegetarian diet reduces angina and cardiac procedures or hospitalizations compared with usual care is supported but uncertain, resting on one small randomized trial from 1986–1992 [6]. That the diet alone reduces events in established disease is biologically plausible. That it reduces events more than a Mediterranean diet, which has randomized secondary-prevention evidence [10, 70], is unsupported.
Regression. That intensive lifestyle programs incorporating very-low-fat vegetarian diets produce modest angiographic regression in selected patients is supported but uncertain [6]. That VLF-WFPB alone produces quantitative plaque-volume regression by intravascular ultrasound or CT is unsupported because it has not been tested. That regression eliminates future risk is unsupported.
Superiority and necessity. That very low total fat is independently necessary for benefit is unsupported; evidence within plant diets is short-term and mixed, and randomized event evidence favors including unsaturated fats in omnivorous diets [8, 10, 12]. That complete exclusion of animal foods adds benefit beyond a low-saturated-fat diet is biologically plausible (zero dietary cholesterol, very low saturated fat, possibly lower TMAO) but untested on outcomes.
Whole foods and processing. That minimally processed plant foods are associated with lower cardiovascular risk, and ultra-processed plant foods with higher risk, is supported but uncertain: large cohorts are consistent, and randomized evidence exists for energy intake but not for atherosclerosis [17–19]. That diet quality and processing matter as much as animal-food exclusion is untested: no study has compared the two levers directly.
Beyond ApoB. That whole-food plant diets protect arteries through pathways beyond ApoB is biologically plausible, with partial observational support: mechanisms are demonstrated in animals, leafy-green nitrate is associated with lower risk in a large cohort after adjustment for hypercholesterolemia, and the Lifestyle Heart Trial showed continued angiographic improvement after ApoB had returned to baseline [6, 28, 29]. It is not established, because no trial has compared diets at matched ApoB.
Vegans compared with other vegetarians. That vegans have lower heart disease risk than lacto-ovo or pesco-vegetarians is unsupported: across Adventist, British, and pooled cohorts, vegan estimates are favorable in direction but inconclusive, and no cohort provides a direct comparison between vegetarian subgroups [41, 53, 54, 56]. The signal among Adventist vegan men is hypothesis-generating.
Role relative to drug therapy. Using diet in place of indicated lipid-lowering or other guideline-directed therapy in established ASCVD is unsupported: no trial reviewed here tested withdrawal of indicated therapy, and every intervention trial showing benefit was delivered alongside it [10, 12, 22].

Figure 15. Totality of evidence by claim and evidence stream. Author synthesis of Sections 3–17; “against” marks evidence that contradicts the claim as stated.
19. The Strongest Claim the Evidence Justifies Today
ホールフード・プラントベース(WFPB)食は、一般的な西欧型食生活と比較して動脈硬化性リポタンパク質を確実に低下させ、動物性食品の摂取が生涯を通じて少ない集団では、コレステロール値が低く、冠動脈疾患の死亡率も低いことが記録されています。超低脂肪のベジタリアン食を中心とした集中的なライフスタイルプログラムは、選択された患者を対象とした小規模な研究において、通常のケアと比較して血管造影上のわずかな退縮、心筋血流の改善、狭心症の減少、および心臓処置の減少をもたらしました。これらの証拠を総合すると、このような食事療法が、主に累積的なApoB暴露の低減を通じて、冠動脈動脈硬化の予防、進行の遅延、および一部の患者における部分的な退縮に役立つ可能性があることは信頼に値します。その有益性は、動物性食品を排除すること単独よりも、加工を最小限に抑え、食物繊維が豊富で、野菜をふんだんに摂るという食事のホールフード的性質と最も一貫して関連しており、実証はされていませんが、その一部はApoB以外の経路を介している可能性を示唆する説得力のある証拠もあります。一方で、すべての動物性食品とほぼすべての添加脂肪を排除することが、適切に実施された地中海食、DASH食、または不飽和脂肪酸の多いプラントベース食よりも優れた保護効果を持つことは、証拠によって確立されていません。
より強力な主張を行うには、同等のサポート強度と薬物療法の下でこれらの食事をランダム化比較し、期間中を通じてApoBを測定し、盲検化された定量的プラーク画像診断を行い、最終的に判定された臨床イベントを評価する必要があります。そのようなデータが存在するまでは、超低脂肪(VLF)-WFPBは、唯一無二の有効な退縮プロトコルとしてではなく、LDL-C低下能に優れた、植物主体の心臓に健康な食事群の中の、エビデンスに裏付けられた選択肢の一つとして提示するのが最善です。
表4. エビデンス表。階層:1 = 臨床イベントを伴うランダム化試験、2 = 中間アウトカムのランダム化試験、3 = 前向きコホート、4 = 集団比較、横断的研究、または症例シリーズ(セクション1.2)。CI = 95%信頼区間。
| 勉強 | 階層 | デザイン、n、期間 | 提供された食事 / 達成された食事 | 背景療法 | アウトカムおよび効果推定値 | 主な限界 |
| Koch 2023 メタ解析 [22] | 2 | 30のRCT;6つの試験からのApoB | ベジタリアンまたはヴィーガン 対 雑食;脂肪含有量は様々 | 様々;一部は脂質低下薬を服用 | TC −0.34 mmol/L (CI −0.44 ~ −0.23);LDL-C −0.30 mmol/L (−0.40 ~ −0.19) ≈ −11.6 mg/dL;ApoB −12.92 mg/dL (−22.63 ~ −3.20;I² = 71.7%、6つの試験);TGに差なし | 食事内容と対照群の不均一性;短期間の試験;イベントなし;脂肪レベルを分離していない;ApoBの推定値は、ここでは独立性が確認できなかった実質的な異質性を持つ6つの試験に基づいている |
| Wang 2023 メタ解析 [7] | 2 | 20のRCT、1,878人;平均25.4週間 | ベジタリアン(オーニッシュ食および低脂肪ヴィーガンを含む) | 大部分が心臓代謝薬を服用 | LDL-C −6.6 mg/dL (−10.1 ~ −3.1);HbA1c −0.24% (−0.40 ~ −0.07);体重 −3.4 kg (−4.9 ~ −2.0);SBP −0.1 mm Hg (−2.8 ~ 2.6) | LDL-Cはアクティブな対照群と比較して有意差なし;大部分でバイアスのリスクが高い;「要点」のテキストが結果と異なっている |
| Rees 2021 Cochrane [72] | 2 | 13のRCT;12週間以上;2020年2月までの検索 | ヴィーガンのみ | — | 臨床イベントを報告した試験はなし;1つの二次予防試験(n = 63):明確な脂質または血圧への影響なし | 小規模な試験;古い検索データ;確実性が低い |
| Dybvik 2023 コホートメタ解析 [40] | 3 | 13のコホート、844,175人 | ベジタリアンまたはヴィーガン 対 非ベジタリアン(自己申告) | — | IHD RR 0.79 (0.71–0.88; 8つのコホート); CVD 0.85 (0.79–0.92); 脳卒中 0.90 (0.77–1.05); ヴィーガン IHD 0.82 (0.68–1.00; 6) | 残差交絡 (E値 1.86);約1/5はBMIが媒介;初期の追跡期間を除外すると関連が弱まる;脂肪は評価されていない |
| EPIC-Oxford [41] | 3 | コホート;48,188人;18.1年 | ヴィーガンを含むベジタリアン 対 肉食者;ヴィーガンは約28%の脂肪摂取 | — | IHD HR 0.78 (0.70–0.87)、リスク因子調整後は0.90 (0.81–1.00);全脳卒中 HR 1.20 (1.02–1.40)、大部分が出血性;ヴィーガン IHD 0.82 (0.64–1.05)、ラクト・オボ 0.77 (0.69–0.86)、魚食者 0.87 (0.77–0.99) | 観察研究;自己申告による食事およびリスク因子 |
| AHS-2 [53] | 3 | コホート;73,308人のアドベンチスト;5.79年 | ヴィーガン、ラクト・オボ、ペスコ、セミベジタリアン 対 非ベジタリアン | — | Vegetarians: IHD death 0.81 (0.64–1.02); men 0.71 (0.51–1.00). By group: vegan 0.90 (0.60–1.33), lacto-ovo 0.82 (0.62–1.06), pesco 0.65 (0.43–0.97); vegan men 0.45 (0.21–0.94), vegan women 1.39 (0.87–2.24) | Short follow-up; healthy-user profile (Figure 9); few vegan deaths |
| Plant diet indices [17] | 3 | 3 cohorts; ~209,000; 8,631 CHD | Healthful vs unhealthful plant-based indices | — | CHD HR (extreme deciles): PDI 0.92 (0.83–1.01); hPDI 0.75 (0.68–0.83); uPDI 1.32 (1.20–1.46) | FFQ-based; health professionals; hPDI rewards oils and nuts |
| UK Biobank processing [18] | 3 | Cohort; 126,842; median 9 y | Plant-sourced foods split by ultra-processing | — | Per 10% energy: plant non-UPF CVD 0.93 (0.91–0.95), CVD death 0.87 (0.80–0.94); plant UPF CVD 1.05 (1.03–1.07), death 1.12 (1.05–1.20) | Observational; 24-h recalls; NOVA classification |
| Danish nitrate cohort [29] | 3 | Cohort; 53,150; up to 23 y; 14,088 CVD | Vegetable nitrate quintiles (median 23 vs 59 mg/day) | Adjusted for hypercholesterolemia | CVD HR 0.85 (0.82–0.89); IHD 0.88 (0.82–0.94); ischemic stroke 0.83 (0.76–0.91); PAD 0.74 (0.67–0.83); 21.9% mediated by SBP | Observational; plateau ~60 mg/day; nitrate mainly lettuce and potato |
| Key 1999 pooled cohorts [54] | 3 | 5 cohorts; 76,172; mean 10.6 y | Vegan, lacto-ovo, fish eaters, occasional meat vs regular meat eaters | — | IHD mortality: vegetarians 0.76 (0.62–0.94); vegans 26% lower, lacto-ovo and fish eaters 34% lower; benefit limited to diet >5 y and larger at younger ages | Mortality only; older cohorts; adjusted for age, sex, smoking only |
| AHS-2 2024 [55] | 3 | Cohort; 88,400; ~11 y | Five diet groups | — | Vegetarians lower IHD mortality; vegan diet not associated with all-cause mortality overall; vegan men lower mortality at younger ages | Hazard ratios vary with age; stroke and dementia higher in older vegetarians |
| AHS-2 protein [57] | 3 | Cohort; 81,337; 9.4 y; 2,276 CVD deaths | Protein-source factors | — | CVD mortality: meat factor 1.61 (98.75% CI 1.12–2.32); nuts and seeds factor 0.60 (0.42–0.86) | Factor analysis; FFQ; observational |
| Vegan diets review [56] | 3 | Systematic review; 7 studies; ≥7,661 vegans | Vegan vs non-vegan | — | No cohort showed significantly higher or lower primary CVD risk in vegans | Few vegans; low power |
| Tzu Chi cohorts [58] | 3 | 2 cohorts; 13,352; Taiwan | Buddhist vegetarians (mostly lacto-ovo) vs non-vegetarians | — | Ischemic stroke 0.26 (0.08–0.88) and 0.41 (0.19–0.88); hemorrhagic 0.34 (0.12–1.00) | Stroke only; healthy-user setting; few events |
| Xiamen lacto-vegetarians [59] | 4 | Cross-sectional; 169 vs 126 men | Chinese lacto-vegetarian vs omnivore | — | Lower BP, LDL-C, ApoB, TG, glucose; thinner carotid IMT | Single measurement; surrogate markers; dairy-eating |
| Hong Kong vegans (review) [60] | 4 (review) | Review of cross-sectional and supplementation studies | Vegans and vegetarians with low B12 | — | ~80% B12 deficiency in Hong Kong vegans; impaired FMD and thicker IMT with deficiency; improved with B12 | Surrogate outcomes; caution for unsupplemented vegan diets |
| Tsimane [35] | 4 | Cross-sectional; 705 adults 40–94; CT calcium | 14% fat, 14% protein, 72% carbohydrate; game and fish; unprocessed | なし | CAC 0 in 85%, 1–100 in 13%, >100 in 3%; >75 y: 65% zero; LDL-C 91 mg/dL; ApoB 97 mg/dL | CAC misses noncalcified plaque; cross-sectional; 6–7 h/day activity (men) |
| China Study I [43] | 4 | Ecological; 65 counties, 130 villages; diet 1983–84; mortality 1973–75 | 14% fat; animal protein ~1% of energy; not vegan | なし | Mean TC 127 vs 203 mg/dL (US); CAD mortality ages 0–64: 4.0 (men) and 3.4 (women) vs 66.8 and 18.9 per 100,000 (US) | Ecological; decade gap; ages truncated; ascertainment; activity and competing mortality |
| Okinawa (Willcox) [45, 47] | 4 | Ecological; 1949 survey; vital statistics | ~1,785 kcal/day; sweet potato ~69% of energy; fat ~6%; some pork and fish | なし | Older cohorts reported ~80% lower CHD mortality than US | Post-war scarcity; energy restriction; body size; transition; no individual linkage |
| Esselstyn 1995/1999 [5, 42] | 4 | Case series; 22–24 enrolled; 11 imaged at ~5 y | ≤10% fat; skim milk, nonfat yogurt allowed | Cholestyramine + lovastatin most often | TC 246 → 132.4 mg/dL (LDL-C 71.6); % stenosis 53.4 → 46.2% (−7 points, CI 3.3–10.7); MLD +0.08 mm (−0.06 to 0.22), NS; regression in 8/11 patients by % stenosis | No control; 13 of 38 lesions excluded (4 graft-proximal lesions that progressed); attrition reported as 5 and as 11; drugs; regression to the mean |
| Esselstyn 2014 [4] | 4 | Cohort; 198 (177 adherent, 21 nonadherent); mean 44.2 mo | No animal foods, oil, nuts, avocado; fat not measured | Usual medications, not recorded | Adherent: 1 progression-related stroke per investigators (0.6%); 18/177 “worse” (10%); 0 cardiac deaths. Nonadherent: 13/21 (62%) with ≥1 event, 7 of 13 events revascularizations | Self-selected; self-reported adherence; no lipids; investigator adjudication; not randomized; 2.2% in supplementary table |
| Lifestyle Heart Trial [6, 63] | 2 (events: counts) | RCT (invitational), 48; 35 with 5-y QCA; 5 y | 10% fat vegetarian (nonfat dairy, egg white); achieved 6.2% (1 y), 8.5% (5 y); plus exercise, stress management, support | No lipid drugs (exp.); 60% of controls started them | 5-y (1998 report, n = 35): % diameter stenosis −3.07 (−5.91 to −0.24) vs +11.77 (3.40 to 20.14), P = .001; minimum lumen diameter essentially unchanged (+0.001 mm) vs −0.34 mm (P = .05); 1-y (1990 report, n = 48): 40.0% → 37.8% vs 42.7% → 46.1%, 82% of experimental patients toward regression; events 25 vs 45, rate ratio 2.47 (1.48–4.20); angina −91% at 1 y (within group) | Multicomponent; small; half of eligible declined; attrition; events = recurrent counts, mostly procedures; stenosis and lumen diameter are not plaque volume |
| Gould 1995 PET [61] | 2 | Same trial; 20 vs 15 | As above | As above | Perfusion abnormalities smaller/less severe vs worsening in controls | Not independent; functional endpoint |
| Multicenter Lifestyle Demonstration [64] | 4 | Nonrandomized; 194 vs 139 revascularized; 3 y | オーニッシュ・プログラム | Not reported | 150/194 (77%) avoided revascularization; similar MI, stroke, death rates per patient-year | Self-selection; comparator is revascularized patients |
| STARS [66] | 2 | RCT; 90 men; 39 mo | Lipid-lowering diet, 27% fat (not vegan) ± cholestyramine | Cholestyramine in one arm | MAWS −0.201 (usual care), +0.003 (diet), +0.103 mm (diet + resin); progression 46%, 15%, 12% | Small; men only; not plant-exclusive |
| Heidelberg [62] | 2 | RCT; 113 men; 12 mo | Low-fat, low-cholesterol diet + intensive exercise | No lipid drugs | Progression 23% vs 48%; regression 32% vs 17% | Exercise co-intervention; lipid differences gone at 6 y |
| DISCO-CT [68] | 2 | RCT; 92; ~67 wk; CCTA | DASH-based counseling + activity + OMT vs OMT | OMT in both arms | Noncalcified plaque −51.3 vs −21.3 mm³ (P = .045); total atheroma change not significantly different; ~6-y follow-up: MACE 1 vs 4 [69] | Single center; not plant-exclusive; weight largely regained; too few events |
| EVADE CAD [9] | 2 | RCT; 100; 8 wk | Vegan vs AHA diet; median fat 29.9% vs 30.2% of energy at 8 wk; unsaturated oils encouraged and olive oil in recipes for both; groceries provided | 94–96% statins | hs-CRP 32% lower (β 0.68, 0.49–0.94); LDL-C 13% lower (β 0.87, 0.78–0.97; NS at Bonferroni α) | Short; biomarker only; not very-low-fat; 14% of those meeting initial criteria enrolled |
| Barnard 2021 crossover [23] | 2 | Crossover RCT; 62 randomized, 52 completers; 16 wk per diet | Low-fat vegan (17% of energy from fat achieved; 95% CI 15–19) vs PREDIMED-style Mediterranean | Lipid estimate in 43 without lipid-drug changes; BP estimate in 41 without antihypertensive changes | Weight −6.0 kg (−7.5 to −4.5) and LDL-C −14.8 mg/dL (−23.5 to −6.2) favoring vegan; SBP +6.0 mm Hg (+1.0 to +10.9) favoring Mediterranean | Short; diet-induced weight loss may mediate the effects, so the direct effect of dietary composition is not isolated; advocacy sponsor; not a 10–15%-fat diet; significant SBP carryover in the all-participant analysis but not in the stable-medication subgroup; 1 participant reassigned after randomization to match a household member |
| CARDIVEG [71] | 2 | クロスオーバーRCT、低リスク107名、1食事につき3ヶ月間 | 低カロリーのラクト・オボ・ベジタリアン対低カロリーの地中海食 | 低リスクの成人 | 体重減少は同等、LDL-Cはベジタリアンでより低下、TGは地中海食でより低下、B12はベジタリアンで減少 | 低リスク、ヴィーガンまたは極低脂肪ではない |
| 心臓の健康のためのレシピ [8] | 2 | クロスオーバーRCT、40名、各フェーズ4週間 | 高EVOO(脂肪48%)のWFPBヴィーガン対低EVOO(脂肪32%) | 特定なし | 両群ともLDL-CとApoBが低下、第1期のLDL-Cは−25.5対−16.7 mg/dL (P = .162)、順序交互作用あり | 短期間、キャリーオーバーあり、どちらのフェーズも極低脂肪ではない |
| ポートフォリオ・メタ解析 [15] | 2 | 対照試験、参加者439名 | NCEP Step II食にナッツ、植物性タンパク質、粘性繊維、ステロールを追加、脂肪含有量は試験により異なる | 多様 | LDL-Cが約17%低下、ApoBと非HDL-Cも低下 | 脂質エンドポイントのみ、ナッツを含む(低脂肪ではない) |
| 加工に関するRCT [19] | 2 | 入院クロスオーバー、20名、各2週間 | 超加工食品対未加工食品、提示された栄養素は一致 | — | 超加工食品で+508 ± 106 kcal/日、+0.9対−0.9 kg | 短期間、エネルギーバランス、動脈硬化ではない |
| DASH食 [11] | 2 | 給食RCT、459名、8週間 | 低脂肪乳製品を含むDASHコンビネーション | — | 血圧−5.5/−3.0 mm Hg、高血圧患者で−11.4/−5.5 | 血圧エンドポイントのみ |
| DASH-食塩 [16] | 2 | 給食RCT | DASH+低ナトリウム対高ナトリウム対照 | — | SBP −7.1(正常血圧)、−11.5 mm Hg(ステージ1高血圧) | 血圧エンドポイントのみ |
| PREDIMED 2018 [12] | 1 | RCT、高リスク7,447名、中央値4.8年 | 地中海食+EVOOまたはナッツ対脂肪摂取抑制のアドバイス | — | 96/2,543対83/2,454対109/2,450、HR 0.69 (0.53–0.91) EVOO、0.72 (0.54–0.95) ナッツ | 2013年の報告はランダム化の不備により撤回、対照群はアドバイスベース |
| リヨン・ダイエット・ハート [70] | 1 | RCT、心筋梗塞後605名、平均46ヶ月 | バターやクリームの代わりにALA豊富なマーガリンを供給、さらに地中海食のアドバイス、対慎重な西洋食 | — | 心臓死/心筋梗塞 14対44、より広範な複合項目 27対90および95対180、複合項目全体での調整後RR 0.28–0.53 | 対照群が弱い、治療の時代が古い |
| CORDIOPREV [10] | 1 | RCT、CHD患者1,002名、中央値7年 | 地中海食(脂肪40.5%)対低脂肪の雑食(脂肪32.1%) | スタチン服用率 86.6% | 87 (17.3%) 対 111 (22.2%)、HR 0.745 (0.563–0.986)、男性 HR 0.669 (0.489–0.915)、女性175名における推定値は結論に至らず | 単一施設、オリーブオイル財団からの資金提供、低脂肪群はVLF-WFPBではない |
| Hooper 2020 Cochrane [26] | 1 | 主要イベント解析における12のRCT、53,758名 | 飽和脂肪酸の削減 | — | 複合心血管イベント RR 0.83 (0.70–0.98) | 飽和脂肪を検証したもので、総脂肪量やVLF-WFPBを検証したものではない |
表5. VLF-WFPBの提唱されている利点:対照となる食事と共通か、それともさらなる有益性のエビデンスがあるか?
| 提唱されている利点 | 主なエビデンス | 地中海食/DASH食と共通か? | VLF-WFPBのさらなる有益性のエビデンス | 判定 |
| LDL-CおよびApoBの低下(極低飽和脂肪、食事性コレステロールなし、食物繊維、植物性タンパク質) | 統合RCT対雑食 [22]; ヴィーガン対地中海食クロスオーバー [23]; CARDIVEG [71]; Portfolio [15] | 部分的に:すべて飽和脂肪を低下させる。Portfolioはナッツにより約17%のLDL-C低下を達成。直接比較では地中海食のLDL-C低下はより小さかった。 | いくつかのプラントベースの介入は、数週間から数ヶ月にわたり、選択された対照群よりもLDL-CおよびApoBを低下させる。特定の脂肪10–15%プロトコルの優位性は確立されていない。 | 支持される(バイオマーカー)、特定のプロトコルについては未確立 |
| 体重減少/低エネルギー密度 | Wang [7]; Barnard [23]; Ornish [6] | 両方の食事がエネルギー制限されている場合は同等 [71] | 短期間の試験で自由摂取の場合はより大きい | 状況に依存 |
| 血圧低下 | Wang(有意差なし) [7]; DASH [11, 16]; Barnard [23] | はい、DASH食と地中海食は同等またはそれ以上 | 示されていない | さらなる有益性は示されていない |
| より良好 血糖コントロール | Wang、2型糖尿病サブグループ [7] | 部分的に | 従来の糖尿病食との比較であり、地中海食との比較ではない | 支持されているが不確実 |
| トリグリセリドおよびHDL-C | Ornish [6]; Koch [22]; CARDIVEG [71] | 地中海食の方がTGをより低下させる | 極低脂肪食はトリグリセリドを上昇させ、HDL-Cを低下させる可能性がある。HDL-Cの低下のみでは心血管系への害は証明されない [80]、そしてトリグリセリドはApoB、非HDL-C、およびアウトカムと併せて読み解くべきである | 不確実、デメリットの可能性あり |
| 炎症の軽減 (hs-CRP) | EVADE CAD [9] | 地中海食はCARDIVEGにおいても炎症マーカーを低下させた [71] | 脂質約30%のAHA食との比較。超低脂肪では試験されていない | 生物学的に妥当 |
| 血管内皮機能の改善 | 単回食事研究 [30, 81]; オリーブオイルのメタ解析 [32]; CORDIOPREVサブスタディ [33]; EVADE EndoPAT [9] | はい:持続的なオリーブオイル摂取および地中海食はFMDを改善した | VLF-WFPBにおいて持続的なものはなし | 支持されていない |
| 脳卒中リスクの低下 | EPIC-Oxford [41]; Dybvik [40] | — | いいえ:EPIC-Oxfordにおける高い脳卒中発生率はベジタリアンのグループに関するものであり、脳卒中について試験されていないVLF-WFPBのことではない | いずれの方向性も実証されていない |
| 遵守を助ける明確な食事ルール | 遵守データ [4, 6, 9] | 構造化されたあらゆるプログラム | 行動学的仮説。厳格さは一部の人には役立ち、他の人には妨げになる可能性がある | 仮説 |
| ホールフード、最小限の加工食品 | Plant diet indices [17]; UKバイオバンク [18]; 加工に関するRCT [19] | 部分的に:DASH食や地中海食も最小限の加工パターンの食事である | 非脂質要因として最強の手段。植物性の超加工食品は高リスクと関連している | 支持されているが不確実 |
| 葉物野菜の硝酸塩 → 一酸化窒素 | デンマークのコホート [29] | はい:野菜が豊富なあらゆる食事パターン | 脂質の調整後も関連が持続。ヴィーガン特有ではない | もっともらしく、部分的に支持されている |
| 食物繊維 → 酪酸 | マウスモデル [28] | はい:高食物繊維のあらゆる食事パターン | マウスではコレステロール非依存的。ヒトのアウトカムデータはなし | 生物学的に妥当 |
| TMAOの低下/良好なマイクロバイオーム | コホートの関連 [37]; メンデルランダム化解析では有意差なし [38] | 部分的に(食物繊維が豊富なパターン) | アウトカムの証拠なし。TMAOの因果効果を否定する遺伝学的証拠 | 弱い仮説 |
| 添加油の排除 | 心臓の健康のためのレシピ [8]; PREDIMED [12]; CORDIOPREV [10] | いいえ:EVOO(エクストラバージンオリーブオイル)が豊富な食事は、対照群と比較してイベントを減少させた | 大量の油の添加はLDL-Cの低下を鈍らせる可能性がある。イベントのデータはなし | 必要条件としては支持されていない |
| ナッツと種子類の排除 | PREDIMEDナッツ群 [12] | いいえ:ナッツを補充した食事はイベントを減少させた | なし | 支持されていない |
| 血管造影上の進行遅延または退縮 | Lifestyle Heart研究 [6]; STARS [66]; ハイデルベルク研究 [62]; DISCO-CT [68] | はい:低脂肪の非ヴィーガン食やDASHベースのプログラムも進行を遅らせた | パッケージ(包括的介入)効果のみ。食事の寄与分は不明 | 支持されているが不確実 |
| 動物性食品の完全な排除 | AHS-2 [53]; EPIC-Oxford [41]; プールされたコホート [54]; ヴィーガンのレビュー [56]; AHS-2タンパク質 [57] | 該当なし | なし:ヴィーガンはラクト・オボ・ベジタリアンやペスコ・ベジタリアンよりも良好な結果を示さなかった。ナッツや種子類のタンパク質はCVD死亡率の低下と関連していた | 支持されていない |
略語:ApoB、アポリポタンパク質B;BP、血圧;CCTA、冠動脈CT血管造影;CHD、冠動脈疾患;EVOO、エクストラバージンオリーブオイル;IHD、虚血性心疾患;MAWS、冠動脈セグメントの平均絶対幅;MLD、最小管腔径;NS、有意差なし;OMT、最適薬物療法;QCA、定量的冠動脈造影;RCT、 無作為化比較試験;TC、総コレステロール;TG、中性脂肪;VLF-WFPB、超低脂肪ホールフード植物ベース食。
参考文献
- Wang T, Masedunskas A, Willett WC, Fontana L. Vegetarian and vegan diets: benefits and drawbacks. Eur Heart J. 2023;44(36):3423–3439. doi:10.1093/eurheartj/ehad436.
- Freeman AM, Morris PB, Barnard N, et al. Trending cardiovascular nutrition controversies. J Am Coll Cardiol. 2017;69(9):1172–1187. doi:10.1016/j.jacc.2016.10.086. PMID 28254181.
- Kahleova H, Levin S, Barnard ND. Vegetarian dietary patterns and cardiovascular disease. Prog Cardiovasc Dis. 2018;61(1):54–61. doi:10.1016/j.pcad.2018.05.002. PMID 29800598.
- Esselstyn CB Jr, Gendy G, Doyle J, Golubic M, Roizen MF. A way to reverse CAD? J Fam Pract. 2014;63(7):356–364b. PMID 25198208.
- Esselstyn CB Jr. Updating a 12-year experience with arrest and reversal therapy for coronary heart disease (an overdue requiem for palliative cardiology). Am J Cardiol. 1999;84(3):339–341, A8. doi:10.1016/S0002-9149(99)00290-8. PMID 10496449.
- Ornish D, Scherwitz LW, Billings JH, et al. Intensive lifestyle changes for reversal of coronary heart disease. JAMA. 1998;280(23):2001–2007. doi:10.1001/jama.280.23.2001. PMID 9863851. Correction (author list): JAMA. 1999;281(15):1380.
- Wang T, Kroeger CM, Cassidy S, et al. Vegetarian dietary patterns and cardiometabolic risk in people with or at high risk of cardiovascular disease: a systematic review and meta-analysis. JAMA Netw Open. 2023;6(7):e2325658. doi:10.1001/jamanetworkopen.2023.25658. PMID 37490288.
- Krenek AM, Mathews A, Guo J, et al. Recipe for Heart Health: a randomized crossover trial on cardiometabolic effects of extra virgin olive oil within a whole-food plant-based vegan diet. J Am Heart Assoc. 2024;13(15):e035034. doi:10.1161/JAHA.124.035034. PMID 39045758.
- Shah B, Newman JD, Woolf K, et al. Anti-inflammatory effects of a vegan diet versus the American Heart Association–recommended diet in coronary artery disease trial. J Am Heart Assoc. 2018;7(23):e011367. doi:10.1161/JAHA.118.011367. PMID 30571591.
- Delgado-Lista J, Alcala-Diaz JF, Torres-Peña JD, et al. Long-term secondary prevention of cardiovascular disease with a Mediterranean diet and a low-fat diet (CORDIOPREV): a randomised controlled trial. Lancet. 2022;399(10338):1876–1885. doi:10.1016/S0140-6736(22)00122-2. PMID 35525255.
- Appel LJ, Moore TJ, Obarzanek E, et al. A clinical trial of the effects of dietary patterns on blood pressure. DASH Collaborative Research Group. N Engl J Med. 1997;336(16):1117–1124. doi:10.1056/NEJM199704173361601. PMID 9099655.
- Estruch R, Ros E, Salas-Salvadó J, et al. Primary prevention of cardiovascular disease with a Mediterranean diet supplemented with extra-virgin olive oil or nuts. N Engl J Med. 2018;378(25):e34. doi:10.1056/NEJMoa1800389. PMID 29897866. (Replaces the retracted 2013 report.)
- Kris-Etherton P, Eckel RH, Howard BV, St Jeor S, Bazzarre TL. AHA science advisory: Lyon Diet Heart Study. Benefits of a Mediterranean-style, National Cholesterol Education Program/American Heart Association Step I dietary pattern on cardiovascular disease. Circulation. 2001;103(13):1823–1825. doi:10.1161/01.CIR.103.13.1823.
- de Lorgeril M, Renaud S, Mamelle N, et al. Mediterranean alpha-linolenic acid-rich diet in secondary prevention of coronary heart disease. Lancet. 1994;343(8911):1454–1459. doi:10.1016/S0140-6736(94)92580-1.
- Chiavaroli L, Nishi SK, Khan TA, et al. Portfolio dietary pattern and cardiovascular disease: a systematic review and meta-analysis of controlled trials. Prog Cardiovasc Dis. 2018;61(1):43–53. doi:10.1016/j.pcad.2018.05.004. PMID 29807048.
- Sacks FM, Svetkey LP, Vollmer WM, et al. Effects on blood pressure of reduced dietary sodium and the Dietary Approaches to Stop Hypertension (DASH) diet. N Engl J Med. 2001;344(1):3–10. doi:10.1056/NEJM200101043440101. PMID 11136953.
- Satija A, Bhupathiraju SN, Spiegelman D, et al. Healthful and unhealthful plant-based diets and the risk of coronary heart disease in U.S. adults. J Am Coll Cardiol. 2017;70(4):411–422. doi:10.1016/j.jacc.2017.05.047. PMID 28728684.
- Rauber F, Louzada MLC, Chang K, et al. Implications of food ultra-processing on cardiovascular risk considering plant origin foods: an analysis of the UK Biobank cohort. Lancet Reg Health Eur. 2024;43:100948. doi:10.1016/j.lanepe.2024.100948. PMID 39210945.
- Hall KD, Ayuketah A, Brychta R, et al. Ultra-processed diets cause excess calorie intake and weight gain: an inpatient randomized controlled trial of ad libitum food intake. Cell Metab. 2019;30(1):67–77.e3. doi:10.1016/j.cmet.2019.05.008.
- Ference BA, Ginsberg HN, Graham I, et al. Low-density lipoproteins cause atherosclerotic cardiovascular disease. 1. Evidence from genetic, epidemiologic, and clinical studies. A consensus statement from the European Atherosclerosis Society Consensus Panel. Eur Heart J. 2017;38(32):2459–2472. doi:10.1093/eurheartj/ehx144. PMID 28444290.
- Cholesterol Treatment Trialists’ (CTT) Collaboration. Efficacy and safety of more intensive lowering of LDL cholesterol: a meta-analysis of data from 170 000 participants in 26 randomised trials. Lancet. 2010;376(9753):1670–1681. doi:10.1016/S0140-6736(10)61350-5. PMID 21067804.
- Koch CA, Kjeldsen EW, Frikke-Schmidt R. Vegetarian or vegan diets and blood lipids: a meta-analysis of randomized trials. Eur Heart J. 2023;44(28):2609–2622. doi:10.1093/eurheartj/ehad211. PMID 37226630.
- Barnard ND, Alwarith J, Rembert E, et al. A Mediterranean diet and low-fat vegan diet to improve body weight and cardiometabolic risk factors: a randomized, cross-over trial. J Am Nutr Assoc. 2022;41(2):127–139. doi:10.1080/07315724.2020.1869625. PMID 33544066.
- McMurry MP, Connor WE, Cerqueira MT. Dietary cholesterol and the plasma lipids and lipoproteins in the Tarahumara Indians: a people habituated to a low cholesterol diet after weaning. Am J Clin Nutr. 1982;35(4):741–744. doi:10.1093/ajcn/35.4.741.
- Lichtenstein AH, Khera A, Anderson CAM, et al. 2026 dietary guidance to improve cardiovascular health: a scientific statement from the American Heart Association. Circulation. Published online 31 March 2026. doi:10.1161/CIR.0000000000001435.
- Hooper L, Martin N, Jimoh OF, Kirk C, Foster E, Abdelhamid AS. Reduction in saturated fat intake for cardiovascular disease. Cochrane Database Syst Rev. 2020;(8):CD011737. doi:10.1002/14651858.CD011737.pub3. (Version of record superseding the May 2020 issue, PMID 32428300.)
- Blumenthal RS, Morris PB, Gaudino M, et al. 2026 ACC/AHA/AACVPR/ABC/ACPM/ADA/AGS/APhA/ASPC/NLA/PCNA guideline on the management of dyslipidemia: a report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. Circulation. 2026. doi:10.1161/CIR.0000000000001423. Also J Am Coll Cardiol. doi:10.1016/j.jacc.2025.11.016.
- Kasahara K, Krautkramer KA, Org E, et al. Interactions between Roseburia intestinalis and diet modulate atherogenesis in a murine model. Nat Microbiol. 2018;3(12):1461–1471. doi:10.1038/s41564-018-0272-x.
- Bondonno CP, Dalgaard F, Blekkenhorst LC, et al. Vegetable nitrate intake, blood pressure and incident cardiovascular disease: Danish Diet, Cancer, and Health Study. Eur J Epidemiol. 2021;36(8):813–825. doi:10.1007/s10654-021-00747-3.
- Vogel RA, Corretti MC, Plotnick GD. Effect of a single high-fat meal on endothelial function in healthy subjects. Am J Cardiol. 1997;79(3):350–354. doi:10.1016/S0002-9149(96)00760-6. PMID 9036757.
- Vogel RA, Corretti MC, Plotnick GD. The postprandial effect of components of the Mediterranean diet on endothelial function. J Am Coll Cardiol. 2000;36(5):1455–1460. doi:10.1016/S0735-1097(00)00896-2. PMID 11079642.
- Schwingshackl L, Christoph M, Hoffmann G. Effects of olive oil on markers of inflammation and endothelial function—a systematic review and meta-analysis. Nutrients. 2015;7(9):7651–7675. doi:10.3390/nu7095356. PMID 26378571.
- Yubero-Serrano EM, Fernandez-Gandara C, Garcia-Rios A, et al. Mediterranean diet and endothelial function in patients with coronary heart disease: an analysis of the CORDIOPREV randomized controlled trial. PLoS Med. 2020;17(9):e1003282. doi:10.1371/journal.pmed.1003282. PMID 32903262.
- Christ A, Günther P, Lauterbach MAR, et al. Western diet triggers NLRP3-dependent innate immune reprogramming. Cell. 2018;172(1-2):162–175.e14. doi:10.1016/j.cell.2017.12.013. PMID 29328911.
- Kaplan H, Thompson RC, Trumble BC, et al. Coronary atherosclerosis in indigenous South American Tsimane: a cross-sectional cohort study. Lancet. 2017;389(10080):1730–1739. doi:10.1016/S0140-6736(17)30752-3. PMID 28320601.
- Koeth RA, Wang Z, Levison BS, et al. Intestinal microbiota metabolism of L-carnitine, a nutrient in red meat, promotes atherosclerosis. Nat Med. 2013;19(5):576–585. doi:10.1038/nm.3145. PMID 23563705.
- Tang WHW, Wang Z, Levison BS, et al. Intestinal microbial metabolism of phosphatidylcholine and cardiovascular risk. N Engl J Med. 2013;368(17):1575–1584. doi:10.1056/NEJMoa1109400. PMID 23614584.
- Jia J, Dou P, Gao M, et al. Assessment of causal direction between gut microbiota-dependent metabolites and cardiometabolic health: a bidirectional Mendelian randomization analysis. Diabetes. 2019;68(9):1747–1755. doi:10.2337/db19-0153.
- Watts GF, Jackson P, Mandalia S, et al. Nutrient intake and progression of coronary artery disease. Am J Cardiol. 1994;73(5):328–332. PMID 8109545.
- Dybvik JS, Svendsen M, Aune D. Vegetarian and vegan diets and the risk of cardiovascular disease, ischemic heart disease and stroke: a systematic review and meta-analysis of prospective cohort studies. Eur J Nutr. 2023;62(1):51–69. doi:10.1007/s00394-022-02942-8. PMID 36030329.
- Tong TYN, Appleby PN, Bradbury KE, et al. Risks of ischaemic heart disease and stroke in meat eaters, fish eaters, and vegetarians over 18 years of follow-up: results from the prospective EPIC-Oxford study. BMJ. 2019;366:l4897. doi:10.1136/bmj.l4897. PMID 31484644.
- Esselstyn CB Jr, Ellis SG, Medendorp SV, Crowe TD. A strategy to arrest and reverse coronary artery disease: a 5-year longitudinal study of a single physician’s practice. J Fam Pract. 1995;41(6):560–568. PMID 7500065.
- Campbell TC, Parpia B, Chen J. Diet, lifestyle, and the etiology of coronary artery disease: the Cornell China study. Am J Cardiol. 1998;82(10B):18T–21T. doi:10.1016/S0002-9149(98)00718-8. PMID 9860369.
- Connor WE, Cerqueira MT, Connor RW, Wallace RB, Malinow MR, Casdorph HR. The plasma lipids, lipoproteins, and diet of the Tarahumara Indians of Mexico. Am J Clin Nutr. 1978;31(7):1131–1142. doi:10.1093/ajcn/31.7.1131. PMID 665563.
- Willcox BJ, Willcox DC, Todoriki H, et al. Caloric restriction, the traditional Okinawan diet, and healthy aging: the diet of the world’s longest-lived people and its potential impact on morbidity and life span. Ann N Y Acad Sci. 2007;1114:434–455. doi:10.1196/annals.1396.037. PMID 17986602.
- Willcox DC, Willcox BJ, Todoriki H, Suzuki M. The Okinawan diet: health implications of a low-calorie, nutrient-dense, antioxidant-rich dietary pattern low in glycemic load. J Am Coll Nutr. 2009;28(Suppl):500S–516S. doi:10.1080/07315724.2009.10718117. PMID 20234038.
- Willcox BJ, Willcox DC. Caloric restriction, caloric restriction mimetics, and healthy aging in Okinawa: controversies and clinical implications. Curr Opin Clin Nutr Metab Care. 2014;17(1):51–58. doi:10.1097/MCO.0000000000000019. PMID 24316687.
- Gavrilova NS, Gavrilov LA. Comments on dietary restriction, Okinawa diet and longevity. Gerontology. 2012;58(3):221–223. doi:10.1159/000329894. PMID 21893946.
- Lindeberg S, Lundh B. Apparent absence of stroke and ischaemic heart disease in a traditional Melanesian island: a clinical study in Kitava. J Intern Med. 1993;233(3):269–275. doi:10.1111/j.1365-2796.1993.tb00986.x. PMID 8450295.
- Lindeberg S, Nilsson-Ehle P, Terént A, Vessby B, Scherstén B. Cardiovascular risk factors in a Melanesian population apparently free from stroke and ischaemic heart disease: the Kitava study. J Intern Med. 1994;236(3):331–340.
- Lindeberg S, Nilsson-Ehle P, Vessby B. Lipoprotein composition and serum cholesterol ester fatty acids in nonwesternized Melanesians. Lipids. 1996;31(2):153–158. doi:10.1007/BF02522614. PMID 8835402.
- Kwok CS, Umar S, Myint PK, Mamas MA, Loke YK. Vegetarian diet, Seventh Day Adventists and risk of cardiovascular mortality: a systematic review and meta-analysis. Int J Cardiol. 2014;176(3):680–686. doi:10.1016/j.ijcard.2014.07.080.
- Orlich MJ, Singh PN, Sabaté J, et al. Vegetarian dietary patterns and mortality in Adventist Health Study 2. JAMA Intern Med. 2013;173(13):1230–1238. doi:10.1001/jamainternmed.2013.6473. PMID 23836264.
- Key TJ, Fraser GE, Thorogood M, et al. Mortality in vegetarians and nonvegetarians: detailed findings from a collaborative analysis of 5 prospective studies. Am J Clin Nutr. 1999;70(3 Suppl):516S–524S. doi:10.1093/ajcn/70.3.516s. PMID 10479225.
- Abris GP, Shavlik DJ, Mathew RO, et al. Cause-specific and all-cause mortalities in vegetarian compared with those in nonvegetarian participants from the Adventist Health Study-2 cohort. Am J Clin Nutr. 2024;120(4):907–917. doi:10.1016/j.ajcnut.2024.07.028. PMID 39098708.
- Kaiser J, van Daalen KR, Thayyil A, Cocco MTARR, Caputo D, Oliver-Williams C. A systematic review of the association between vegan diets and risk of cardiovascular disease. J Nutr. 2021;151(6):1539–1552. doi:10.1093/jn/nxab037. PMID 33831953.
- Tharrey M, Mariotti F, Mashchak A, Barbillon P, Delattre M, Fraser GE. Patterns of plant and animal protein intake are strongly associated with cardiovascular mortality: the Adventist Health Study-2 cohort. Int J Epidemiol. 2018;47(5):1603–1612. doi:10.1093/ije/dyy030. PMID 29618018.
- Chiu THT, Chang HR, Wang LY, Chang CC, Lin MN, Lin CL. Vegetarian diet and incidence of total, ischemic, and hemorrhagic stroke in 2 cohorts in Taiwan. Neurology. 2020;94(11):e1112–e1121. doi:10.1212/WNL.0000000000009093. PMID 32102976.
- Yang SY, Li XJ, Zhang W, et al. Chinese lacto-vegetarian diet exerts favorable effects on metabolic parameters, intima-media thickness, and cardiovascular risks in healthy men. Nutr Clin Pract. 2012;27(3):392–398. doi:10.1177/0884533611436173. PMID 22412169.
- Woo KS, Kwok TCY, Celermajer DS. Vegan diet, subnormal vitamin B-12 status and cardiovascular health. Nutrients. 2014;6(8):3259–3273. doi:10.3390/nu6083259. PMID 25195560.
- Gould KL, Ornish D, Scherwitz L, et al. Changes in myocardial perfusion abnormalities by positron emission tomography after long-term, intense risk factor modification. JAMA. 1995;274(11):894–901. doi:10.1001/jama.1995.03530110056036. PMID 7674504.
- Schuler G, Hambrecht R, Schlierf G, et al. Regular physical exercise and low-fat diet: effects on progression of coronary artery disease. Circulation. 1992;86(1):1–11. doi:10.1161/01.cir.86.1.1. PMID 1617762.
- Ornish D, Brown SE, Scherwitz LW, et al. Can lifestyle changes reverse coronary heart disease? The Lifestyle Heart Trial. Lancet. 1990;336(8708):129–133. doi:10.1016/0140-6736(90)91656-U. PMID 1973470.
- Ornish D. Avoiding revascularization with lifestyle changes: the Multicenter Lifestyle Demonstration Project. Am J Cardiol. 1998;82(10B):72T–76T. doi:10.1016/S0002-9149(98)00744-9. PMID 9860380.
- Silberman A, Banthia R, Estay IS, et al. The effectiveness and efficacy of an intensive cardiac rehabilitation program in 24 sites. Am J Health Promot. 2010;24(4):260–266. (Cited only as the second source of an untraced statement in ref. 2.)
- Watts GF, Lewis B, Brunt JN, et al. Effects on coronary artery disease of lipid-lowering diet, or diet plus cholestyramine, in the St Thomas’ Atherosclerosis Regression Study (STARS). Lancet. 1992;339(8793):563–569. doi:10.1016/0140-6736(92)90863-X. PMID 1347091.
- Niebauer J, Hambrecht R, Velich T, et al. Attenuated progression of coronary artery disease after 6 years of multifactorial risk intervention: role of physical exercise. Circulation. 1997;96(8):2534–2541. PMID 9355890.
- Henzel J, Kępka C, Kruk M, et al. High-risk coronary plaque regression after intensive lifestyle intervention in nonobstructive coronary disease: a randomized study. JACC Cardiovasc Imaging. 2021;14(6):1192–1202. doi:10.1016/j.jcmg.2020.10.019. PMID 33341413.
- Makarewicz-Wujec M, Henzel J, Kępka C, et al. Long-term outcomes of the Dietary Approaches to Stop Hypertension (DASH) intervention in nonobstructive coronary artery disease: follow-up of the DISCO-CT study. Nutrients. 2025;17(15):2565. doi:10.3390/nu17152565.
- de Lorgeril M, Salen P, Martin JL, Monjaud I, Delaye J, Mamelle N. Mediterranean diet, traditional risk factors, and the rate of cardiovascular complications after myocardial infarction: final report of the Lyon Diet Heart Study. Circulation. 1999;99(6):779–785. doi:10.1161/01.CIR.99.6.779. PMID 9989963.
- Sofi F, Dinu M, Pagliai G, et al. Low-calorie vegetarian versus Mediterranean diets for reducing body weight and improving cardiovascular risk profile: CARDIVEG Study (Cardiovascular Prevention With Vegetarian Diet). Circulation. 2018;137(11):1103–1113. doi:10.1161/CIRCULATIONAHA.117.030088. PMID 29483085.
- Rees K, Al-Khudairy L, Takeda A, Stranges S. Vegan dietary pattern for the primary and secondary prevention of cardiovascular diseases. Cochrane Database Syst Rev. 2021;2(2):CD013501. doi:10.1002/14651858.CD013501.pub2. PMID 33629376.
- Virani SS, Newby LK, Arnold SV, et al. 2023 AHA/ACC/ACCP/ASPC/NLA/PCNA guideline for the management of patients with chronic coronary disease. Circulation. 2023;148(9):e9–e119. doi:10.1161/CIR.0000000000001168.
- Dawson LP, Lum M, Nerleker N, Nicholls SJ, Layland J. Coronary atherosclerotic plaque regression: JACC state-of-the-art review. J Am Coll Cardiol. 2022;79(1):66–82. doi:10.1016/j.jacc.2021.10.035. PMID 34991791.
- Armstrong ML, Megan MB. Lipid depletion in atheromatous coronary arteries in rhesus monkeys after regression diets. Circ Res. 1972;30(6):675–680. doi:10.1161/01.res.30.6.675.
- Armstrong ML, Megan MB. Arterial fibrous proteins in cynomolgus monkeys after atherogenic and regression diets. Circ Res. 1975;36(2):256–261. doi:10.1161/01.res.36.2.256.
- Clarkson TB, Bond MG, Bullock BC, Marzetta CA. A study of atherosclerosis regression in Macaca mulatta. IV. Changes in coronary arteries from animals with atherosclerosis induced for 19 months and then regressed for 24 or 48 months at plasma cholesterol concentrations of 300 or 200 mg/dl. Exp Mol Pathol. 1981;34(3):345–368. doi:10.1016/0014-4800(81)90052-6.
- Nicholls SJ, Puri R, Anderson T, et al. Effect of evolocumab on progression of coronary disease in statin-treated patients: the GLAGOV randomized clinical trial. JAMA. 2016;316(22):2373–2384. doi:10.1001/jama.2016.16951. PMID 27846344.
- Melina V, Craig W, Levin S. Position of the Academy of Nutrition and Dietetics: vegetarian diets. J Acad Nutr Diet. 2016;116(12):1970–1980. doi:10.1016/j.jand.2016.09.025. PMID 27886704.
- Voight BF, Peloso GM, Orho-Melander M, et al. Plasma HDL cholesterol and risk of myocardial infarction: a Mendelian randomisation study. Lancet. 2012;380(9841):572–580. doi:10.1016/S0140-6736(12)60312-2. PMID 22607825.
- Vogel RA. The Mediterranean diet and endothelial function: why some dietary fats may be healthy. Cleve Clin J Med. 2000;67(4):232–236. (Reports the Vogel group’s postprandial olive-oil, vitamin, and vinegar-salad experiments.)