不同饮食对健康的影响 | Dr. Christopher Gardner
摘要
斯坦福大学医学教授、营养学研究主任 Dr. Christopher Gardner 深入探讨了不同饮食方式、食物质量及营养研究方法背后的科学依据。他与主持人 Andrew Huberman 共同探索了为何没有一种饮食适合所有人、超加工食品的危害，以及设计严谨的营养学研究如何拨开笼罩公共卫生领域的重重迷雾。两人的讨论最终聚焦于以全食物为基础、以植物为主的饮食模式，认为这是目前获得最广泛支持的方法——同时也承认个体差异的重要性。
核心要点
- 不存在唯一”最佳饮食” —— 人类具有极强的适应能力，可以在截然不同的饮食模式下保持健康，从几乎全碳水化合物（Tarahumara 族）到几乎全脂肪（因纽特人），只要食物是天然、未经加工的即可。
- 健康问题的共同根源是标准美式饮食 —— 高度加工、热量密集、低纤维，以精制小麦和添加糖为主。
- “全食物植物性饮食”并不等于纯素饮食 —— 它意味着以植物为主，但可包含 10–30% 的动物性食品。这一概念已被过度借用，造成了广泛混淆。
- 低碳水和低脂肪饮食在食物质量同样较高时可产生相似结果 —— 关键变量是食物质量，而非单纯的宏量营养素比例。
- 生酮饮食在降低甘油三酯方面比地中海饮食更有效，但往往会升高低密度脂蛋白胆固醇，且长期坚持难度更大。
- 胰岛素抵抗人群可能更适合低碳水化合物饮食，但基因预测检测目前尚无法可靠判断谁更适合哪种饮食。
- 超加工食品含有数百种”外观添加剂”（色素、乳化剂、香精），这些成分的健康影响可能超出其宏量营养素构成本身——而且许多都可以通过配方改良来去除，其他国家已有实践。
- 生牛奶不能治愈乳糖不耐受 —— Gardner 主持的一项对照研究证实，饮用生牛奶与巴氏杀菌牛奶引起的胃肠道症状完全相同。
- 美国人 90% 的谷物摄入来自小麦，添加糖和精制谷物占总碳水化合物摄入量的 40%——这是代谢功能障碍的主要驱动因素之一。
- 营养学家在共同审阅数据时基本达成共识；公众的困惑源于研究设计不佳、研究者偏见，以及媒体对相互矛盾的研究标题的渲染方式。
详细笔记
个体差异与饮食灵活性
- 人类在不同文化背景下依托截然不同的传统饮食繁衍生息——高脂肪（因纽特人）、高碳水（Tarahumara 族），以及介于两者之间的各种饮食。
- 当传统全食物饮食被包装加工食品取代时，健康问题便会持续出现。
- 具有参考价值的规律性现象：许多人报告称，无论是转向更多动物性食物，还是转向更多植物性食物，健康状况都有所改善——两个方向都存在真实的受益者。
- 遗传性饮食适应最具代表性的例子是乳糖酶持续表达 —— 北欧人进化出成年后持续分泌乳糖酶的能力，从而能够消化乳制品。全球大多数人口存在乳糖不耐受。
- 目前尚未在人群层面发现其他同等明确的基因-饮食交互作用。
标准美式饮食的问题
- 约 40% 的美国人碳水化合物摄入来自添加糖和精制谷物（主要是精制小麦）；仅约 10% 来自健康碳水化合物来源。
- 美国人90% 的谷物摄入为小麦，而全球饮食更为多样，包括大米、燕麦、藜麦、小米等。
- 单一谷物品种的过度消费，加上现代单一作物小麦的遗传特性，可能是麸质敏感和小麦不耐受投诉增加的原因之一。
- 乳糜泻存在漏诊问题——约一半确诊患者并不知情，仍在持续食用小麦。
- 欧洲购买同品牌产品（如 Nutella）的消费者往往获得成分更少的配方，而非美国版本——这表明配方改良在不大幅增加成本的前提下是可行的。
超加工食品与 NOVA 分类体系
- NOVA 分类（由巴西 Carlos Monteiro 开发）按照加工程度对食物进行分类，与营养成分无关。
- 超加工食品含有**“外观添加剂”** —— 色素、乳化剂、上光剂、香精——其用途是增加感官吸引力和延长货架期，而非提供营养。
- NOVA 添加剂清单上约有 150 种化学物质；其中许多也在 FDA 的 **GRAS（通常认为安全）**名单上，该名单收录的物质已从约 800 种增至约 10,000 种。
- 在人体中单独测试各种添加剂的危害在方法学上存在困难：癌症或心脏病等结局需要数十年才能显现；而短期心脏代谢指标（胆固醇、血糖）对单独接触色素并不会有所反应。
- 一个现实问题：立即消除所有超加工食品将使美国约 60% 的超市库存消失，且缺乏可及的替代品——这需要同步推进食品系统改革。
- 关键改革路径：向食品制造商施压，要求其在美国市场采用更清洁的国际版配方，而非一味禁止。
重要营养学研究与发现
A to Z 减重研究（JAMA，2007 年）—— 311 名女性，1 年
- 比较了 Atkins（极低碳水）、Zone、Ornish（极低脂肪）及传统饮食方案。
- 唯一具有统计学意义的差异：Atkins 与 Zone 之间——而非 Atkins 与 Ornish 之间，尽管两者是饮食上的对立面。
- 更值得关注的是：组内变异极大——每组均有参与者减重 30–50 磅，也有人增重 5–10 磅。
- 这种组内变异（而非组间平均值）成为后续研究的核心关注点。
DIETFITS 研究（JAMA，2018 年）—— 600 名成人，1 年，约 800 万美元
- 比较健康低碳水化合物饮食与健康低脂肪饮食。
- 两种饮食均强调全食物；两组均被要求避免添加糖和精制谷物。
- 结果：两组平均减重无显著差异。
- **胰岛素抵抗**状态及所测试的 3-SNP 基因型均无法可靠预测谁在哪种饮食下效果更好。
- 组内变异依然显著：两组减重范围均为 −60 磅至 +20 磅。
- 解读：当两种饮食均以高质量食物为基础时，任何一种方法都可能有效——个体因素比宏量营养素比例更为重要。
生酮饮食与地中海饮食对比研究
- 两种饮食均限制添加糖和精制谷物；两组均包含地上生长的蔬菜。
- 生酮饮食排除豆类、水果和全谷物；地中海饮食则包含这些食物。
- 结果：
- HbA1c（主要结局）无显著差异
- 生酮饮食更有效地降低甘油三酯（通过消除膳食碳水化合物，而这些碳水化合物在肝脏中本会转化为甘油三酯）
- 生酮饮食升高低密度脂蛋白胆固醇（因饱和脂肪摄入增加）
- 生酮饮食的依从性显著更难维持
SWAP-MEAT 研究（Beyond Meat 与红肉对比）
- 红肉来自牧场散养、有机认证供应商（旧金山 Good Eggs），以确保研究的公平性。
- 结果：Beyond Meat 在多项心脏代谢指标上优于红肉。
- 研究结束后，Beyond Meat 对产品进行了重新配方——去除椰子油，加入更温和的成分——显示出行业对研究结果的响应能力。
双胞胎研究（Netflix 纪录片）
- 纯素饮食在心脏代谢结局上优于杂食性饮食。
生牛奶与乳糖不耐受（小型试点研究，n=16）
- 入组要求：参与者需同时未通过客观氢气呼气测试并有自我报告症状。
- 研究组别：生牛奶、巴氏杀菌牛奶、豆浆（对照组）。
- 发现：生牛奶与巴氏杀菌牛奶产生的胃肠道症状完全相同。对乳糖不耐受无任何改善。
- 值得注意：约 50% 自我报告乳糖不耐受的参与者（均为白种人）并未未通过氢气呼气测试——提示可能存在小肠细菌过度生长或其他机制。
蛋白质与”
English Original 英文原文
How Different Diets Impact Your Health | Dr. Christopher Gardner
Summary
Dr. Christopher Gardner, professor of medicine and director of nutrition studies at Stanford, discusses the science behind different dietary approaches, food quality, and nutrition research methodology. He and host Andrew Huberman explore why no single diet works for everyone, the problems with ultra-processed foods, and how well-designed nutrition studies can cut through the confusion dominating public health discourse. Their conversation converges on a whole food, plant-forward dietary pattern as the most broadly supported approach — while acknowledging meaningful individual variation.
Key Takeaways
- There is no single “best diet” — humans are remarkably resilient and can thrive on radically different dietary patterns, from near-total carbohydrate (Tarahumara) to near-total fat (Inuit), as long as food is whole and unprocessed.
- The common thread in poor health outcomes is the standard American diet — highly processed, calorie-dense, low-fiber, and dominated by refined wheat and added sugar.
- “Whole food plant-based” does not mean vegan — it means mostly plants, but can include 10–30% animal products. The term has been co-opted and causes widespread confusion.
- Both low-carb and low-fat diets can produce similar outcomes when both are done with high food quality — the key variable is food quality, not macronutrient ratio alone.
- Ketogenic diet lowers triglycerides more effectively than Mediterranean diet, but tends to raise LDL cholesterol and is harder to sustain long-term.
- Insulin resistance may favor a lower-carbohydrate approach, though genetic predisposition tests have not yet reliably predicted who does better on which diet.
- Ultra-processed foods contain hundreds of “cosmetic additives” (dyes, emulsifiers, flavorants) that may have health effects beyond their macronutrient profiles — and many can be eliminated by reformulation, as already practiced in other countries.
- Raw milk does not cure lactose intolerance — a controlled study by Gardner confirmed identical GI symptoms on raw versus pasteurized cow’s milk.
- Americans eat 90% of their grains as wheat, with 40% of total carbohydrate intake coming from added sugar and refined grains — a primary driver of metabolic dysfunction.
- Nutrition scientists largely agree when reviewing data together; the public confusion stems from poorly designed studies, investigator bias, and media framing of conflicting headlines.
Detailed Notes
Individual Variation and Diet Flexibility
- Humans have thrived on dramatically different traditional diets across cultures — high-fat (Inuit), high-carb (Tarahumara), and everything in between.
- Health problems emerge consistently when traditional whole-food diets are replaced by packaged, processed foods.
- Anecdotal but valid pattern: many individuals report health improvements after switching either toward more animal food or toward more plant food — both directions have real responders.
- The best-established example of genetic dietary adaptation is lactase persistence — northern Europeans evolved to continue producing lactase into adulthood, enabling dairy digestion. Most of the world’s population is lactose intolerant.
- No other comparably clear gene-diet interaction has been identified at population scale.
The Standard American Diet Problem
- ~40% of American carbohydrate intake is from added sugar and refined grains (primarily refined wheat); only ~10% comes from healthy carbohydrate sources.
- Americans consume 90% of their grain intake as wheat, compared to global diversity that includes rice, oats, quinoa, millet, and others.
- Overconsumption of a single grain variety, combined with modern monocrop wheat genetics, may contribute to the rise of gluten sensitivity and wheat intolerance complaints.
- Celiac disease is underdiagnosed — roughly half of people with confirmed celiac are unaware and continue consuming wheat.
- Europeans buying the same branded products (e.g., Nutella) often receive formulations with fewer ingredients than U.S. versions — suggesting reformulation is feasible without major cost increases.
Ultra-Processed Foods and the NOVA Classification
- The NOVA classification (developed by Carlos Monteiro, Brazil) categorizes foods by degree of processing, independent of nutrient content.
- Ultra-processed foods contain “cosmetic additives” — dyes, emulsifiers, glazing agents, flavorants — whose purpose is aesthetic appeal and shelf stability, not nutrition.
- There are approximately 150 chemicals on the NOVA additive list; many are also on the FDA’s GRAS (Generally Recognized as Safe) list, which has grown from ~800 to ~10,000 substances over time.
- Testing individual additives for harm in humans is methodologically difficult: outcomes like cancer or heart disease require decades; short-term cardiometabolic markers (cholesterol, glucose) don’t respond to dye exposure alone.
- A practical concern: eliminating all ultra-processed foods immediately would remove ~60% of U.S. grocery store inventory without accessible replacements — requiring parallel food system reform.
- Key reform pathway: pressure food manufacturers to match their cleaner international formulations in U.S. products, rather than an outright ban.
Key Nutrition Studies and Findings
A to Z Weight Loss Study (JAMA, 2007) — 311 Women, 1 Year
- Compared Atkins (very low carb), Zone, Ornish (very low fat), and a traditional approach.
- Only statistically significant difference: Atkins vs. Zone — not Atkins vs. Ornish, despite being dietary opposites.
- More striking: massive within-group variation — in every group, some participants lost 30–50 lbs while others gained 5–10 lbs.
- This within-group variation, not between-group averages, became the focus of subsequent research.
DIETFITS Study (JAMA, 2018) — 600 Adults, 1 Year, ~$8 Million
- Compared healthy low-carbohydrate diet vs. healthy low-fat diet.
- Both diets emphasized whole foods; both groups were instructed to avoid added sugar and refined grains.
- Result: No significant average difference in weight loss between groups.
- Neither insulin resistance status nor a tested 3-SNP genotype reliably predicted who did better on which diet.
- Huge within-group variation persisted: weight loss ranged from −60 lbs to +20 lbs in both groups.
- Interpretation: when both diets are done with high food quality, either approach can work — individual factors matter more than macronutrient ratio.
Ketogenic vs. Mediterranean Diet Study
- Both diets restricted added sugar and refined grains; both included above-ground vegetables.
- Keto excluded beans, fruit, and whole grains; Mediterranean embraced them.
- Results:
- No significant difference in HbA1c (primary outcome)
- Keto lowered triglycerides more effectively (by eliminating dietary carbohydrates that would otherwise be converted to triglycerides in the liver)
- Keto raised LDL cholesterol (due to higher saturated fat intake)
- Adherence was significantly harder to maintain on keto
SWAP-MEAT Study (Beyond Meat vs. Red Meat)
- High-quality red meat sourced from pasture-raised, organic suppliers (Good Eggs, San Francisco) to ensure equipoise.
- Result: Beyond Meat outperformed red meat on several cardiometabolic markers.
- After the study, Beyond Meat reformulated — removed coconut oil, added more benign ingredients — suggesting industry responsiveness to research.
Twin Study (Netflix-featured)
- Vegan diet outperformed omnivore diet in cardiometabolic outcomes.
Raw Milk and Lactose Intolerance (Small Pilot, n=16)
- Participants required to fail objective hydrogen breath test AND report symptoms to enroll.
- Arms: raw milk, pasteurized milk, soy milk (control).
- Finding: Raw milk produced identical GI symptoms to pasteurized milk. No benefit for lactose intolerance.
- Notable: ~50% of self-reported lactose-intolerant participants (all Caucasian) did not fail the hydrogen breath test — suggesting possible SIBO or other mechanisms.