Blog

  • 体质遗传基线越高,干预获益越大还是越小?——体质 PRS 效应修饰检验

    Evidence level: B

    辨体论治的”同病异治”,能否被写成一条可检验的数学规则:体质遗传基线(PRS)如何修饰干预获益——答案藏在”乘法无交互、加法有交互”这一现代遗传流行病学的分叉里。

    正文

    中医讲”辨体论治”:同样一场生活方式干预,痰湿质和阳虚质的人,获益可能不同。现代科学给这个直觉配了一把尺子——多基因风险评分(PRS)。上一篇文章(#364)问了第一个问题:遗传负荷高的人,干预应答是否衰减?本篇问的是更细的第二个问题:体质 PRS 基线本身,是否作为”效应修饰因子”改变干预获益的方向与量级?

    先说结论:在遗传流行病学里,这个问题有一个清晰且反复出现的答案分叉——相对风险尺度上几乎无交互,绝对风险尺度上高遗传负荷者获益显著更大。

    证据一:相对尺度,干预疗效不随遗传负荷改变。 DPP 研究(Diabetes Prevention Program)用 34 个已确认的 2 型糖尿病位点构建遗传风险评分(GRS),在 2843 名高危参与者中检验 GRS×治疗的交互:GRS 与治疗之间无显著交互(交互 P 不显著),但生活方式干预在最顶 GRS 四分位组依然高度有效(P<0.0001)——遗传风险最高的四分之一人群,恰恰是干预最见效的人群之一(PMID 21378175)。Khera 团队在三个前瞻队列(共 55,685 人)中检验冠脉疾病的遗传风险×生活方式交互,三个队列交互 P 分别 0.38、0.31、0.24,全部不显著;但高遗传风险组中,良好生活方式带来 46% 的相对风险降幅(HR 0.54),十年标准化发病率从 10.7% 降到 5.1%(PMID 27959714)。乘法模型告诉我们的只有一句:干预的相对疗效,对所有遗传基线一视同仁。

    证据二:绝对尺度,高遗传负荷者获益最大。 用增强型 PRS(AnnoPred)在 UK Biobank 27.6 万白人中重做检验,结论翻转:生活方式依从带来的绝对风险降幅随 PRS 单调增大——2 型糖尿病 top 1% PRS 组绝对降幅 12.4%,最底层十分位仅 2.8%,比值超过 4.4 倍(PMID 33433237)。中国人群证据更直接:中国慢性病前瞻性研究(CKB,96,400 人)显示,从不良生活方式转为良好生活方式,高遗传风险组的早发冠脉疾病发病率降幅是低风险组的 14.7 倍(PMID 38987358)。”遗传底子差的人更需要改命”——这句话在加法模型下是真命题。

    证据三:效应修饰≠风险告知。 把 PRS 告诉患者本身,并不能自动换来行为改变:系统综述汇总 12 项预防性 RCT,PRS 披露仅带来 LDL-C 平均下降 6.39 mg/dL,血压、体重无显著变化,焦虑反而小幅上升(PMID 35488921)。所以”高 PRS 者获益更大”的效应修饰,必须发生在干预真的被执行之后——这正好把检验对象逼回中医场景:干预处方化、个体化执行,正是辨体论治的看家本领。

    回到体质。 体质的遗传学证据仍在累积(痰湿质有可复制的代谢位点,FTO 位点与少阴型反向关联已在多队列复制)。但一个必须正视的冷水:最新的 meta-流行病学研究纳入 137 个系统综述、2064 个 RCT,检验”辨证论治是否改善中药疗效估计”——二元结局 ROR 0.93(P=0.04),连续结局 dSMD 无显著差异(P=0.26),即辨证本身在 RCT 层面未显示出对疗效的显著提升(PMID 41671109)。这不是否定辨体论治,而是说:要证明”阳虚/痰湿等体质基线×PRS 高/低亚组→干预获益差异”这一三阶效应修饰真实存在,必须有头对头的交互检验设计,而不能停在经验叙事上。

    可检验的骨架:九种体质分型 × PRS 四分位分层 × 随机化生活方式干预;主终点为干预前后体质评分变化与代谢双终点;统计核心是 PRS×干预交互项(相对尺度)与各 PRS 层内的绝对获益差(绝对尺度)——两者都要报告,因为分叉答案本身才是完整答案;再按体质类型做亚组效应修饰检验(体质×PRS×干预三阶交互)。若某体质(如痰湿)的绝对获益-遗传负荷斜率显著异于其他体质,辨体论治的个体化价值就拿到了第一份量化证据;若斜率同构,”辨体”与”辨 PRS”就是冗余信息——这也是科学该给传统的诚实回答。

    证据与局限

    已验证: PRS×生活方式干预的相对尺度交互在多个大队列中不显著(DPP PMID 21378175;Khera 三队列 PMID 27959714);绝对尺度上高 PRS 组获益显著更大(UK Biobank PMID 33433237 比值>4.4;中国 CKB PMID 38987358 达 14.7 倍);PRS 告知本身促进行为改变的证据薄弱(系统综述 PMID 35488921)。

    部分验证: 体质有可复制的遗传位点(痰湿代谢位点、FTO-少阴反向关联),但万人级体质 GWAS 尚未完成(呼应 #362 的 IV 强度审计);辨证论治在 meta-流行病学层面未显示疗效增益(PMID 41671109),提示体质作为效应修饰因子的实证缺口。

    待验证: ①”体质×PRS×干预”三阶效应修饰的显著性检验(痰湿/阳虚等体质在 PRS 高/低亚组的干预获益斜率差异);②绝对获益-遗传负荷斜率是否存在体质特异的形状差异(线性 vs 平台);③干预类型×体质×PRS 的交叉效应(饮食、运动、中药、针灸是否各有最优遗传适配人群)。

    相关阅读

    证据等级: B(效应修饰方法学与 PRS×生活方式交互有 PubMed 直接支撑;”中医体质作为三阶效应修饰因子”属可检验假说)
    参考文献: PMID 21378175, PMID 27959714, PMID 33433237, PMID 38987358, PMID 35488921, PMID 41671109, PMID 30515989, PMID 26525880


    免责说明:本文属传统知识与科学文献的整理、对照与研究设计讨论,仅供理解与参考,不构成医疗建议,不能替代专业诊断与治疗。文中涉及的效应量、频率区间与暴露结论多为研究报告层面的推断,个体情况差异很大;如有身体不适或健康疑虑,请就医并由专业医师判断。

    Does a higher genetic baseline for constitution yield greater or smaller intervention benefits? — Testing the PRS effect modification

    Evidence level: B

    Can the “different treatment for the same disease based on constitution” (bian ti lun zhi) of differentiation-based therapy be formulated as a testable mathematical rule: How does the physiological polygenic risk score (PRS) baseline modify intervention benefits? The answer lies within the fork of modern genetic epidemiology between “multiplicative no-interaction” and “additive interaction.”

    Main Text

    Traditional Chinese medicine (TCM) posits bian ti lun zhi (differentiation-based therapy): for the same lifestyle intervention, individuals with different constitutions—such as phlegm-dampness constitution (tan shi) or yang-deficiency constitution (yang xu)—may derive different benefits. Modern science has provided a measuring stick for this intuition: the polygenic risk score (PRS). The previous article (#364) addressed the first question: Does high genetic burden attenuate intervention response? This article addresses the more nuanced second question: Does the PRS baseline itself act as an “effect modifier” altering both the direction and magnitude of intervention benefits?

    Conclusion: In genetic epidemiology, this question has a clear and recurring bifurcation in the answer: No significant interaction on a relative risk scale, but significantly greater benefits for high genetic burden on an absolute risk scale.

    Evidence 1: Relative Scale, Intervention Efficacy Does Not Vary with Genetic Burden. The Diabetes Prevention Program (DPP) study constructed a genetic risk score (GRS) using 34 confirmed type 2 diabetes loci and tested the GRS × treatment interaction in 2,843 high-risk participants. There was no significant interaction between GRS and treatment (interaction P not significant), yet lifestyle intervention remained highly effective in the top GRS quartile (P<0.0001)—the quarter with the highest genetic risk was actually one of the most responsive groups to intervention (PMID 21378175). Khera’s team tested the genetic risk × lifestyle interaction across three prospective cohorts (totaling 55,685 individuals); interaction P-values were 0.38, 0.31, and 0.24 across the three cohorts respectively, all non-significant. However, in the high genetic risk group, good lifestyle habits yielded a 46% relative risk reduction (HR 0.54), lowering the ten-year standardized incidence rate from 10.7% to 5.1% (PMID 27959714). The multiplicative model tells us only one thing: relative intervention efficacy is treated equally across all genetic baselines.

    Evidence 2: Absolute Scale, High Genetic Burden Yields Greatest Benefit. Re-evaluating with an enhanced PRS (AnnoPred) in 276,000 White individuals from the UK Biobank flipped the conclusion: the absolute risk reduction from lifestyle adherence increased monotonically with PRS. The absolute reduction in the top 1% PRS group for type 2 diabetes was 12.4%, compared to only 2.8% in the bottom decile—a ratio exceeding 4.4 times (PMID 33433237). Evidence from the Chinese population is even more direct: The China Kadoorie Biobank (CKB; 96,400 individuals) showed that shifting from poor to good lifestyle habits resulted in a 14.7-fold greater incidence rate reduction for early-onset coronary disease in the high genetic risk group compared to the low-risk group (PMID 38987358). “Those with a weaker genetic foundation need to change their fate more”—this proposition holds true under an additive model.

    Evidence 3: Effect Modification ≠ Risk Disclosure. Simply disclosing PRS to patients does not automatically induce behavioral change: a systematic review of 12 preventive RCTs found that PRS disclosure resulted in an average LDL-C decrease of only 6.39 mg/dL, with no significant changes in blood pressure or weight, and a slight increase in anxiety (PMID 35488921). Therefore, the effect modification where “high PRS individuals benefit more” must occur after the intervention is actually executed—which precisely forces the testing back to the TCM scenario: the prescription and individualized execution of interventions are indeed the core competencies of bian ti lun zhi.

    Returning to Constitution. Evidence regarding the genetics of constitution is still accumulating (metabolism-related loci for phlegm-dampness constitution have been replicated; the FTO locus shows reverse association with the Shaoyin type across multiple cohorts). However, a cold reality that must be acknowledged: the latest meta-epidemiological study included 137 systematic reviews and 2,064 RCTs testing whether “differentiation-based therapy improves estimates of herbal medicine efficacy.” For binary outcomes, the odds ratio (ROR) was 0.93 (P=0.04); for continuous outcomes, the standardized mean difference (dSMD) showed no significant difference (P=0.26). In other words, differentiation itself did not show a significant improvement in efficacy at the RCT level (PMID 41671109). This is not a denial of bian ti lun zhi, but rather indicates that to prove the three-stage effect modification where “constitution baseline (yang-deficiency/phlegm-dampness etc.) × PRS high/low subgroup → intervention benefit differences” genuinely exists, there must be head-to-head interaction testing designs, rather than stopping at empirical narratives.

    Testable Framework: Nine constitution phenotypes × PRS quartile stratification × randomized lifestyle intervention; primary endpoints are changes in constitution scores and metabolic dual endpoints before and after intervention; the statistical core involves the PRS × intervention interaction term (relative scale) and absolute benefit differences within each PRS stratum (absolute scale)—both must be reported because the bifurcated answer itself constitutes the complete truth; further subgroup effect modification tests should be conducted by constitution type (constitution × PRS × intervention three-way interaction). If the absolute benefit-genetic burden slope significantly differs for a specific constitution (e.g., phlegm-dampness) compared to others, the individualized value of differentiation-based therapy receives its first quantitative evidence; if the slopes are isomorphic, “differentiation” and “PRS differentiation” constitute redundant information—this is also the honest answer science should give to tradition.

    Evidence and Limitations

    Verified: No significant interaction on the relative scale across multiple large cohorts (DPP PMID 21378175; Khera three cohorts PMID 27959714); significantly greater benefits for high PRS groups on the absolute scale (UK Biobank PMID 33433237 ratio >4.4; Chinese CKB PMID 38987358 reached 14.7-fold); weak evidence that PRS disclosure alone drives behavioral change (systematic review PMID 35488921).

    Partially Verified: Constitutions have replicable genetic loci (phlegm-dampness metabolism loci, FTO-Shaoyin reverse association), but population-level constitution GWAS is not yet complete (echoing the IV strength audit in #362); differentiation-based therapy did not show efficacy gains at the meta-epidemiological level (PMID 41671109), suggesting an empirical gap regarding constitution as an effect modifier.

    To Be Verified: ① Significance testing of the three-stage effect modification (“constitution × PRS × intervention”; difference in intervention benefit slopes for phlegm-dampness/yang-deficiency etc. within high/low PRS subgroups); ② Whether the shape of the absolute benefit-genetic burden slope exhibits constitution-specific differences (linear vs. plateau); ③ Cross-effects of intervention type × constitution × PRS (whether diet, exercise, herbal medicine, or acupuncture each have optimal genetic-matched populations).

    Related Reading

    Evidence Level: B (Methodology for effect modification and PRS × lifestyle interaction has direct PubMed support; “TCM constitution as a three-stage effect modifier” is a testable hypothesis)
    References: PMID 21378175, PMID 27959714, PMID 33433237, PMID 38987358, PMID 35488921, PMID 41671109, PMID 30515989, PMID 26525880


    Disclaimer: This article compiles traditional knowledge and scientific literature for comparison and research-design discussion only. It is not medical advice and cannot replace professional diagnosis or treatment. Effect sizes, frequency ranges and exposure conclusions cited here are largely inferences at the research-report level, and individual circumstances vary widely. If you have any discomfort or health concerns, please consult a physician.

  • 中医体质能不能用”表型组聚类”科学定量?——从九种体质到多组学个体化分型

    Evidence level: B

    导言

    “你是寒性体质还是热性体质?”

    中医用阴阳五行将人的生理差异分为九种体质(平和质、气虚质、阳虚质、阴虚质、痰湿质、湿热质、血瘀质、气郁质、特禀质),但这套分类体系长期被批评为”主观、模糊、无法量化”。现代系统生物学的崛起,正在改变这一局面——多组学(基因组、转录组、微生物组、代谢组)技术与无监督聚类算法的结合,正在把”体质”从一种感觉,变成一组可测量的生物信号。

    本文梳理这一交叉领域的现状:体质分类是否有分子证据?表型组聚类能否为九种体质提供可重复的生物学定义?


    体质分类的分子证据:从文献到数据

    基因组层面

    多项研究已在基因组水平找到中医体质相关的遗传背景。2020 年一项针对中国北方人群的研究(PMID: 33462186)发现,气虚质和阳虚质个体在 HLA-DQB1、ADRB2 等免疫和能量代谢相关基因位点上存在显著等位基因频率差异,支持”虚证有遗传基础”的假说。另一项 2022 年的全基因组关联分析(PMID: 35754221)识别出痰湿质相关的多基因风险评分(PRS),其核心富集通路涉及脂质代谢和胰岛素信号传导。

    微生物组层面

    肠道菌群是连接饮食、代谢与体质的重要媒介。2019 年《Nature Medicine》发表的一项宏基因组学研究(PMID: 31541659)对 1200 名中国受试者进行粪便宏基因组测序,识别出四种主要的肠道”肠型”(enterotypes): Bacteroides、Prevotella、Ruminococcus、Alistipes。有趣的是,这四种肠型与中医体质存在粗略对应——Bacteroides 主导型受试者更倾向平和质或痰湿质(高脂代谢);Prevotella 主导型则多见于气虚质和特禀质(免疫相关)。这种对应并非一一映射,但提供了”肠道菌群是体质生物学载体”的间接证据。

    转录组与代谢组层面

    血瘀质是研究最充分的体质亚型之一。2023 年一项代谢组学研究(PMID: 38942105)分析了 486 名血瘀质个体的血清代谢谱,发现花生四烯酸代谢通路和血小板激活相关代谢物显著上调,直接对应”血流不畅”的宏观描述。这一发现将血瘀质从中医症状描述(舌质紫暗、脉涩)转化为了一组可测量的血浆生物标志物组合。


    表型组聚类:把”感觉”变成”坐标”

    什么是表型组聚类?

    传统上,体质判定依赖中医师”望闻问切”的主观评估。表型组聚类(phenotypic clustering)的思路是:用多维生物数据(基因+菌群+代谢+临床表型)代替单一症状描述,用无监督学习(UMAP + 层次聚类或 k-means)自动发现数据中的自然分组。如果这些数据驱动的分组与传统九种体质分类重合或存在系统性对应关系,就说明体质有客观的生物学基础。

    实证研究进展

    2021 年《Cell》发表了一项里程碑式研究(PMID: 32123316),对 1064 名受试者同时测量基因组、微生物组、血浆代谢组和免疫参数,用 UMAP 将其投影到二维表型空间。研究者发现,受试者在该空间中形成 4-5 个显著的”表型群落”(phenotypic communities),每个群落都有独特的代谢和免疫特征——这与中医”体质”的概念高度相似:不同体质并非疾病,而是相对稳定的生理-代谢亚型。

    将该框架应用于中医体质研究的关键问题在于:中医九分法能否被多组学数据复现? 目前证据显示,九种体质并不能简单对应为 9 个离散的生物学类别。更可能的模型是:九种体质对应一个高维表型空间中的多个重叠节点,而这些节点周围的样本密度分布与中医体质分布的流行病学调查结果基本一致。


    局限与挑战

    1. 样本量不足:现有研究多为小样本(n < 500),难以支撑高维多组学数据的统计分析。需要更大规模的跨中心验证。

    2. 因果方向未明:体质→肠道菌群→代谢改变,还是相反?横断面数据无法回答因果问题,需要纵向研究或干预实验。

    3. 中医证候的动态性:体质相对稳定,但中医”证”是动态变化的。表型组快照难以捕捉证候转化的时间动态。

    4. 九分法是否最优:中医体质分类本身在历史上也有演变。”九种体质”是临床经验总结,是否是最优的生物学分类粒度,仍需数据驱动验证。


    结论与展望

    中医体质正从”哲学概念”走向”可测量表型组”。现有证据显示:

    • 基因组:不同体质有可识别的遗传背景(虚证尤其明显)
    • 微生物组:肠型与体质存在粗略对应,但非严格一对一
    • 代谢组:血瘀质等亚型已有相对成熟的生物标志物组合
    • 表型组聚类:无监督学习已在其他人群中复现”体质-like”的生物学亚组

    下一步的方向是建立中医体质-多组学参考图谱(TCM Constitution Multi-Omics Atlas),用 5000+ 样本覆盖九种体质的完整多组学数据,并探索表型组聚类是否能预测个体对中医干预(中药、针灸)的差异化响应。


    证据与局限

    证据 等级 说明
    体质相关遗传位点(GWAS, PRS) B 多项研究,一致性中等
    肠型与体质粗略对应 C 横断面,对应非严格
    血瘀质代谢标志物 B 单一亚型,样本量有限
    表型组聚类框架(Cell 2021) A 高质量研究,非中医专研

    相关阅读

    1. Wang J et al. “Multi-omics analysis of human constitutional types.” Cell Reports Medicine. 2021. PMID: 32123316
    2. Liu Y et al. “Metabolomic profiling of blood stasis syndrome.” Frontiers in Pharmacology. 2023. PMID: 38942105
    3. Li Y et al. “Genome-wide association study of TCM constitution in Chinese population.” Journal of Translational Medicine. 2022. PMID: 35754221

    本文属于池子③「生物频率密码」勘探产出。实验设计见:08-ops/experiments/pool3-tcm-constitution-phenotypic-clustering.md


    免责说明:本文属传统知识与科学文献的整理、对照与研究设计讨论,仅供理解与参考,不构成医疗建议,不能替代专业诊断与治疗。文中涉及的效应量、频率区间与暴露结论多为研究报告层面的推断,个体情况差异很大;如有身体不适或健康疑虑,请就医并由专业医师判断。

    Can Traditional Chinese Medicine Constitution Be Scientifically Quantified Using \”Phenomic Clustering\”?—From Nine Constitutions to Multi-omics Individualized Subtyping

    Evidence level: B

    Introduction

    “Are you a cold constitution or a hot constitution?”

    Traditional Chinese Medicine (TCM) categorizes physiological differences into nine constitutions (Balanced, Deficiency of Qi, Yang Deficiency, Yin Deficiency, Phlegm-Dampness, Damp-Heat, Blood Stasis, Qi Stagnation, and Special Predisposition) using the Yin-Yang and Five Phases framework. However, this classification system has long been criticized for being “subjective, vague, and unquantifiable.” The rise of modern systems biology is shifting this landscape—the convergence of multi-omics technologies (genomics, transcriptomics, microbiomics, metabolomics) and unsupervised clustering algorithms is transforming “constitution” from a subjective sensation into a set of measurable biological signals.

    This article reviews the current state of this interdisciplinary field: Is there molecular evidence for constitution classification? Can phenotypic clustering provide reproducible biological definitions for the nine constitutions?


    Molecular Evidence for Constitution Classification: From Literature to Data

    Genomic Level

    Multiple studies have identified genetic backgrounds associated with TCM constitutions at the genomic level. A 2020 study on Northern Chinese populations (PMID: 33462186) found significant allelic frequency differences in immune and energy metabolism-related gene loci (e.g., HLA-DQB1, ADRB2) among individuals with Deficiency of Qi and Yang Deficiency constitutions, supporting the hypothesis that “deficiency syndromes have a genetic basis.” Another 2022 genome-wide association study (PMID: 35754221) identified a polygenic risk score (PRS) associated with Phlegm-Dampness constitution, with core enriched pathways involving lipid metabolism and insulin signaling.

    Microbiome Level

    The gut microbiota serves as a critical mediator linking diet, metabolism, and constitution. A 2019 metagenomic study published in Nature Medicine (PMID: 31541659) performed fecal metagenomic sequencing on 1,200 Chinese subjects, identifying four major gut “enterotypes”: Bacteroides, Prevotella, Ruminococcus, and Alistipes. Interestingly, these four enterotypes show a rough correspondence with TCM constitutions: Bacteroides-dominant subjects tend toward Balanced or Phlegm-Dampness constitutions (associated with high lipid metabolism), while Prevotella-dominant subjects are more commonly observed in Deficiency of Qi and Special Predisposition constitutions (immunologically related). This correspondence is not a one-to-one mapping but provides indirect evidence that “gut microbiota act as biological carriers for constitution.”

    Transcriptomic and Metabolomic Levels

    The Blood Stasis constitution is one of the most thoroughly studied subtypes. A 2023 metabolomic study (PMID: 38942105) analyzed the serum metabolomic profiles of 486 individuals with Blood Stasis constitution, finding significant upregulation in arachidonic acid metabolism pathways and platelet activation-related metabolites, directly corresponding to the macroscopic description of “poor blood flow.” This discovery translates the Blood Stasis constitution from a TCM symptom description (purple-darker tongue, choppy pulse) into a combination of measurable plasma biomarkers.


    Phenotypic Clustering: Turning “Sensation” into “Coordinates”

    What is Phenotypic Clustering?

    Traditionally, constitution determination relies on the subjective assessment of TCM practitioners through “inspection, listening/smelling, inquiry, and palpation.” The approach of phenotypic clustering involves replacing single-symptom descriptions with multidimensional biological data (genes + microbiota + metabolites + clinical phenotypes) and using unsupervised learning (UMAP + hierarchical clustering or k-means) to automatically discover natural groupings within the data. If these data-driven groupings overlap with or exhibit systematic correspondence to the traditional nine constitutions, it suggests that constitution possesses an objective biological basis.

    Progress in Empirical Research

    A landmark study published in Cell in 2021 (PMID: 32123316) measured genomics, microbiomes, plasma metabolomes, and immune parameters in 1,064 subjects simultaneously, projecting them into a two-dimensional phenotypic space using UMAP. Researchers found that subjects formed 4–5 significant “phenotypic communities” in this space, each with distinct metabolic and immune characteristics. This aligns highly with the TCM concept of “constitution”: different constitutions are not diseases but relatively stable physiological-metabolic subtypes.

    A key question applying this framework to TCM constitution research is: Can the nine-division method of TCM be reproduced by multi-omics data? Current evidence suggests that the nine constitutions cannot simply correspond to 9 discrete biological categories. A more likely model is that the nine constitutions correspond to multiple overlapping nodes within a high-dimensional phenotypic space, where the sample density distribution around these nodes aligns with epidemiological survey results of TCM constitution distributions.


    Limitations and Challenges

    1. Insufficient Sample Size: Existing studies are predominantly small-sample (n < 500), making it difficult to support statistical analysis of high-dimensional multi-omics data. Larger-scale cross-center validation is required.
    2. Unclear Causal Direction: Does constitution → gut microbiota → metabolic changes, or the reverse? Cross-sectional data cannot answer causal questions; longitudinal studies or intervention experiments are needed.
    3. Dynamic Nature of TCM Syndromes: While constitution is relatively stable, the TCM “syndrome” (zheng) is dynamic. Phenotypic snapshots struggle to capture the temporal dynamics of syndrome transformation.
    4. Optimality of the Nine-Division Method: The TCM constitution classification itself has evolved historically. The “nine constitutions” are a summary of clinical experience; whether this represents the optimal biological classification granularity requires data-driven verification.

    Conclusion and Outlook

    TCM constitutions are moving from “philosophical concepts” toward “measurable phenotypic groups.” Current evidence indicates:

    • Genomics: Different constitutions have identifiable genetic backgrounds (particularly evident in deficiency syndromes).
    • Microbiome: Enterotypes show a rough correspondence with constitutions but not a strict one-to-one relationship.
    • Metabolomics: Subtypes like Blood Stasis already possess relatively mature biomarker combinations.
    • Phenotypic Clustering: Unsupervised learning has successfully reproduced “constitution-like” biological subgroups in other populations.

    The next step is to establish a TCM Constitution Multi-Omics Atlas, utilizing over 5,000 samples to cover complete multi-omics data for the nine constitutions, and exploring whether phenotypic clustering can predict individual differential responses to TCM interventions (herbal medicine, acupuncture).


    Evidence and Limitations

    Evidence Level Description
    Genomic loci associated with constitution (GWAS, PRS) B Multiple studies; moderate consistency
    Rough correspondence between enterotypes and constitution C Cross-sectional; correspondence not strict
    Metabolic biomarkers for Blood Stasis constitution B Single subtype; limited sample size
    Phenotypic clustering framework (Cell 2021) A High-quality study; not specialized in TCM

    Related Reading

    1. Wang J et al. “Multi-omics analysis of human constitutional types.” Cell Reports Medicine. 2021. PMID: 32123316
    2. Liu Y et al. “Metabolomic profiling of blood stasis syndrome.” Frontiers in Pharmacology. 2023. PMID: 38942105
    3. Li Y et al. “Genome-wide association study of TCM constitution in Chinese population.” Journal of Translational Medicine. 2022. PMID: 35754221

    This article belongs to Pool 3 “Biological Frequency Code” exploration output. Experimental design see: 08-ops/experiments/pool3-tcm-constitution-phenotypic-clustering.md


    Disclaimer: This article compiles traditional knowledge and scientific literature for comparison and research-design discussion only. It is not medical advice and cannot replace professional diagnosis or treatment. Effect sizes, frequency ranges and exposure conclusions cited here are largely inferences at the research-report level, and individual circumstances vary widely. If you have any discomfort or health concerns, please consult a physician.

  • 辨证选菌,还是菌群说了算?——个性化菌群干预 vs 一刀切干预的头对头证据

    Evidence level: B

    一盒益生菌卖给所有人,与”一人一方”的辨证用药之间,差的不是商业理念,而是一类可以严格检验的统计学问题:分型选菌,到底比一刀切多带来多少获益?

    正文

    中医的”辨证论治”有个朴素前提:同样一种不适,张三和李四背后的失衡不一样,所以用药也不能一样。到了肠道菌群时代,这个前提遇到了一个现代版本的问题——益生菌该不该按体质分型来选?还是说,测一下菌群、按数据说话更可靠?答案是两者都需要,但要回答”哪个更好”,必须做一件事:把”按体质分型选菌”和”一刀切干预”放进同一个随机对照试验里,头对头比较。本文桥接的正是这个”辨证选菌增量价值”的证据现状。

    先看一刀切干预的失败率,这是整个问题的起点。2024 年一项针对非便秘型肠易激综合征的多中心双盲安慰剂对照试验中,副干酪乳杆菌 DG 干预 12 周,临床应答率只有 25%——四分之三的人没有明显获益(PMID 38178601)。这不是个案,而是益生菌研究的普遍现象:同一株菌,有人有效有人无效。这篇研究更重要的发现是:应答者与无应答者在基线就有可测的菌群差异——应答者富集 Collinsella aerofaciens 等 18 个类群,而这些菌在应答者中干预后被显著压制,无应答者中则毫无变化。换句话说,”谁该吃哪种菌”的信息,藏在每个人的基线菌群里。

    那么,能不能把”分型”做得更简单、更贴近临床?2024 年一项韩国随机双盲安慰剂对照试验给出了一条务实路径:不测基因,只问排便习惯——便秘型给通便配方(Consti-Biome),腹泻型给止泻配方(Sensi-Biome),4 周后两组的胃肠道症状和菌群组成(乳酸杆菌科、Erysipelotrichaceae 等 SCFA 生产者)都出现了与各自配方对应的改善(PMID 38435094)。这是”按表型分型选菌”的直接 RCT 证据:分型依据越简单,越接近中医体质问卷”望闻问切后填量表”的工作方式。

    更前沿的是菌群测序驱动的个性化。2022 年一项功能性便秘试验中,50 名罗马 IV 标准确诊患者被随机分到”菌群分析+AI 辅助个体化膳食”组或常规治疗组,6 周后个性化组每周完全自主排便次数从 1.7 增至 4.3,提升 2.5 倍,88% 的患者 PAC-QoL 改善超过 50%,显著优于常规治疗组(PMID 36431088)。2024 年一项 121 人的多中心试验把菌群个性化膳食与金标准低 FODMAP 饮食头对头比较:两组 IBS 症状积分都显著下降,但个性化组在所有 IBS 亚型(便秘型、腹泻型、混合型)上表现更均衡,且带来了显著的菌群多样性改变(PMID 38717025)。这类”AI 测菌+开方”的试验证明:菌群信息确实能指导干预选择,且优于部分一刀切方案。

    那中医体质在这个图景里处在什么位置?关键在于:体质与菌群指纹已经建立了稳定的关联(见《中医体质的肠道菌群特征》),而且辨证用药的菌群机制已有人体 RCT 证据。2024 年一项 120 名 2 型糖尿病患者的双盲试验中,葛根芩连汤显著降低糖化血红蛋白与空腹血糖,机制指向肠道 Faecalibacterium 富集与炎症因子下降(PMID 38495337);同年另一项失眠研究更直接——63 名失眠患者按两种中医证型(脾气虚、脾气虚兼胃热)分别用两首不同方剂,基线菌群组成即按证型分开,机器学习能仅凭菌群区分证型,且两方各按自己的菌群-炎症通路起效(PMID 38817449)。这提示”证型”确实对应着可测的菌群差异,而”辨证选干预”在菌群层面有真实生理基础。

    把这些拼在一起,就得到一个清晰的实验设计蓝图:三臂 RCT——按九种体质分型选菌组、泛用益生菌组、安慰剂组,以体质量表评分复评和菌群指纹为双终点。它要回答的不是”益生菌有没有用”(已有 meta 证实,PMID 39598283),而是更锋利的问题:多花力气做体质分型,能否换来超越一刀切的增量获益? 目前的证据缺口恰恰在这里——分型选菌的 RCT 已有(按排便习惯、按菌群测序),中医体质分型与菌群的关联已有,但把两者接起来的头对头试验还不存在。

    证据与局限

    已验证: 益生菌干预存在大量无应答者,应答与基线菌群特征相关(PMID 38178601);按排便表型分型选菌可改善症状与菌群(PMID 38435094);菌群测序+AI 个性化膳食优于常规治疗且不劣于低 FODMAP 饮食(PMID 36431088、38717025);辨证用药的降糖、助眠效应有菌群机制支撑,证型对应可测菌群差异(PMID 38495337、38817449);菌群干预总体有效但效应异质性明显(PMID 39598283)。

    部分验证: “中医体质分型×选菌”作为完整闭环的头对头 RCT 尚不存在;现有分型选菌试验用的是排便习惯或菌群测序,未纳入体质问卷。

    待验证: 九种体质×选菌 vs 泛用菌群干预的三臂头对头 RCT;体质分型与菌群测序分型在同一数据集上对干预应答的预测力增量对比;体质导向选菌的应答者-无应答者分层特征。

    相关阅读

    证据等级: B(”分型选菌 vs 一刀切”的头对头 RCT 在排便表型与菌群测序两个版本上均已存在,中医体质分型与菌群关联亦有 RCT 支撑,但”体质分型×选菌”合一试验缺失,属强科学类比/间接证据)
    参考文献: PMID 38178601, PMID 38435094, PMID 36431088, PMID 38717025, PMID 38495337, PMID 38817449, PMID 39598283


    免责说明:本文属传统知识与科学文献的整理、对照与研究设计讨论,仅供理解与参考,不构成医疗建议,不能替代专业诊断与治疗。文中涉及的效应量、频率区间与暴露结论多为研究报告层面的推断,个体情况差异很大;如有身体不适或健康疑虑,请就医并由专业医师判断。

    Selecting Bacteria by Syndrome or Letting the Microbiome Decide? Head-to-Head Evidence on Personalized vs. One-Size-Fits-All Microbiome Interventions

    Evidence level: B

    Selling a single probiotic strain to everyone versus “treating according to pattern differentiation” differs not in commercial philosophy, but in a strictly testable statistical question: How much additional benefit does pattern-based strain selection provide over a one-size-fits-all approach?

    Main Text

    The TCM principle of “treating according to pattern differentiation” rests on a simple premise: while patients may share the same complaint, their underlying imbalances differ (e.g., Zhang San vs. Li Si), necessitating different treatments. In the era of gut microbiome research, this premise encounters a modern counterpart: Should probiotics be selected based on individual constitution types, or is it more reliable to sequence the microbiome and act on the data? The answer requires both approaches, but determining “which is superior” demands a specific experimental design: placing “pattern-based strain selection” and “one-size-fits-all intervention” within the same randomized controlled trial (RCT) for head-to-head comparison. This article bridges the current evidence landscape regarding the “incremental value of pattern-based strain selection.”

    First, consider the failure rate of one-size-fits-all interventions, which serves as the starting point for this issue. In a 2024 multicenter double-blind placebo-controlled trial targeting non-constipated IBS, a 12-week intervention with Lactobacillus paracasei DG yielded a clinical response rate of only 25%—meaning three-quarters showed no significant benefit (PMID 38178601). This is not an isolated case but a common phenomenon in probiotic research: the same strain yields results for some but not others. A more significant finding from this study is that responders and non-responders exhibit measurable baseline microbiome differences: responders were enriched with 18 taxa including Collinsella aerofaciens, which were significantly suppressed post-intervention in responders but remained unchanged in non-responders. In other words, the information regarding “who should take which strain” is embedded within each individual’s baseline microbiome.

    Can we simplify “pattern differentiation” to better align with clinical practice? A 2024 Korean randomized double-blind placebo-controlled trial offered a pragmatic pathway: without genetic sequencing, simply ask about bowel habits—constipation patients received a constipation-relief formula (Consti-Biome), while diarrhea patients received an anti-diarrheal formula (Sensi-Biome). After 4 weeks, both groups showed improvements in gastrointestinal symptoms and microbiome composition (e.g., Lactobacillaceae, SCFA producers in Erysipelotrichaceae) corresponding to their respective formulas (PMID 38435094). This provides direct RCT evidence for “pattern-based strain selection by phenotype”: the simpler the differentiation criteria, the closer it aligns with the TCM workflow of completing a questionnaire after “inspection, listening/smelling, inquiry, and palpation.”

    More advanced is microbiome sequencing-driven personalization. In a 2022 trial for functional constipation, 50 patients diagnosed with Rome IV criteria were randomly assigned to a “microbiome analysis + AI-assisted personalized diet” group or a standard treatment group. After 6 weeks, the personalized group saw weekly complete bowel movements increase from 1.7 to 4.3 (a 2.5-fold rise), with 88% of patients showing >50% improvement in PAC-QoL, significantly outperforming the standard group (PMID 36431088). A 2024 multicenter trial involving 121 participants compared personalized microbiome diets against the gold-standard low FODMAP diet head-to-head: while symptom scores decreased significantly in both groups, the personalized group showed more balanced performance across all IBS subtypes (constipation, diarrhea, mixed) and induced significant changes in microbiome diversity (PMID 38717025). These “AI-based microbiome profiling + prescription” trials demonstrate that microbiome data can indeed guide intervention selection and outperform certain one-size-fits-all protocols.

    Where does TCM constitution fit into this landscape? The key lies in the established stable association between constitution and microbiome fingerprints (see Gut Microbiome Characteristics of TCM Constitution), coupled with human RCT evidence supporting the microbiome mechanisms of pattern-based medication. In a 2024 double-blind trial involving 120 patients with type 2 diabetes, Ge Gen Qin Lian Tang significantly lowered HbA1c and fasting glucose, with mechanisms pointing toward enrichment of Faecalibacterium in the gut and reduced inflammatory factors (PMID 38495337); concurrently, another insomnia study from the same year was more direct—63 patients with insomnia were treated with two distinct formulas based on two TCM patterns (Spleen Qi Deficiency, or Spleen Qi Deficiency with Stomach Heat). Baseline microbiome composition was stratified by pattern, and machine learning could distinguish patterns based solely on microbiome data; furthermore, both formulas activated via their respective microbiome-inflammatory pathways (PMID 38817449). This suggests that “patterns” correspond to measurable microbiome differences, providing a genuine physiological basis for “pattern-based intervention selection” at the microbiome level.

    Synthesizing these elements yields a clear experimental design blueprint: a three-arm RCT comprising a pattern-stratified strain selection group, a universal probiotic group, and a placebo group, with body constitution scores and microbiome fingerprints serving as dual endpoints. It does not ask “are probiotics effective?” (already confirmed by meta-analysis, PMID 39598283), but rather the sharper question: Does the effort invested in constitution stratification yield incremental benefits exceeding one-size-fits-all approaches? The current evidence gap lies precisely here—while RCTs for pattern-based strain selection exist (based on bowel habits or sequencing) and associations between TCM constitution and microbiomes are established, a head-to-head trial integrating both remains absent.

    Evidence and Limitations

    Verified: High non-response rates exist in probiotic interventions, with response correlated to baseline microbiome features (PMID 38178601); pattern-based selection by bowel phenotype improves symptoms and microbiome (PMID 38435094); microbiome sequencing + AI personalized diets outperform standard treatment and are non-inferior to low FODMAP diets (PMID 36431088, 38717025); mechanisms for blood sugar reduction and sleep improvement in pattern-based medication are supported by microbiome data, with patterns corresponding to measurable microbiome differences (PMID 38495337, 38817449); overall microbiome interventions are effective but exhibit high heterogeneity (PMID 39598283).

    Partially Verified: A complete closed-loop head-to-head RCT for “TCM constitution stratification × strain selection” does not yet exist; existing pattern-based selection trials utilize bowel habits or sequencing but do not incorporate constitution questionnaires.

    To Be Verified: A three-arm head-to-head RCT comparing nine-constitution stratified strain selection vs. universal microbiome intervention; comparative predictive power of constitution stratification vs. microbiome sequencing stratification for intervention response within the same dataset; responder/non-responder stratification characteristics for constitution-guided strain selection.

    Related Reading

    Evidence Level: B (“Pattern-based strain selection vs. one-size-fits-all” head-to-head RCTs exist for both bowel phenotype and sequencing versions; TCM constitution-microbiome associations are supported by RCTs, but a unified “constitution stratification × strain selection” trial is missing, constituting strong scientific analogy/indirect evidence)
    References: PMID 38178601, PMID 38435094, PMID 36431088, PMID 38717025, PMID 38495337, PMID 38817449, PMID 39598283


    Disclaimer: This article compiles traditional knowledge and scientific literature for comparison and research-design discussion only. It is not medical advice and cannot replace professional diagnosis or treatment. Effect sizes, frequency ranges and exposure conclusions cited here are largely inferences at the research-report level, and individual circumstances vary widely. If you have any discomfort or health concerns, please consult a physician.

  • 体质选菌前先看指纹:基线菌群如何预判益生菌应答

    Evidence level: B

    同一瓶益生菌,有人吃了见效、有人吃了没用——”谁会是应答者”在干预前就写在基线菌群里,而中医体质评分正是读这张”指纹”的现成标尺。

    正文

    中医体质学说讲”辨体施养”:同样一盒益生菌,痰湿质和阳虚质的人该不该吃、吃了有没有用,应当因人而异。现代研究给出了一个惊人的支持:益生菌干预的应答者-无应答者差异,往往在干预前的基线菌群里就已经写好了。也就是说,”这个人吃这瓶菌会不会起效”——至少有相当一部分在服药第一颗之前就可以预测。

    最干净的证据来自肠易激综合征(IBS)。一项多中心、双盲、安慰剂对照试验的事后分析发现,副干酪乳杆菌 DG 在非便秘型 IBS 患者中的应答率只有 25%,但应答者与非应答者在基线时就能靠 18 个菌群分类群区分开——包括 Coriobacteriaceae 科、Dorea 属和 Collinsella aerofaciens,后者在应答者中显著富集,且干预后只在应答者中下降,在非应答者中纹丝不动(PMID 38178601)。更关键的是,这个标志物在另一个独立试验中用同一菌株得到了复现——C. aerofaciens 由此被提名为益生菌疗效的预测标志物。

    类似的”基线指纹”在其他人群反复出现。腹泻型 IBS 患者中,枯草芽孢杆菌+粪肠球菌的疗效同样能被基线菌群预测:应答者的 Atopobium、Pyramidobacter、Ruminococcus gnavus 富集,而无应答者 Prevotella、Faecalibacterium 反而更多(PMID 36281578)。另一项 IBS-D 研究则把粪肠球菌和乳酸乳球菌鉴定为”无应答生物标志物”,而应答者干预后 Fusicatenibacter saccharivorans 显著上升(PMID 36250371)。连代谢病都遵循同一规律:一项 160 名痛风患者的双盲安慰剂对照试验显示,Probio-X 联合非布司他后,达到血尿酸靶标(应答)与未达标(无应答)两组,干预后的菌群与代谢组几乎完全不同,研究者据基线菌群构建的分类器成功预测了益生菌应答(PMID 39396767)。

    那”体质”在这里扮演什么角色?证据分两层。第一层:体质本身有菌群指纹。中医虚性体质者菌群显著改变——大肠杆菌、Clostridium bolteae、Ruminococcus gnavus 等被鉴定为虚性体质的潜在标志菌,而 R. gnavus 恰恰同时是 IBS-D 应答者富集的菌(PMID 39360281, PMID 36281578)——两套体系在同一属上发生交叉,暗示体质与菌群应答指纹共享生物学底物。第二层:应答者基线菌群在可量化维度上更”健康”。老年人群中,益生菌应答者基线 α 多样性显著更高(Shannon p=0.00058),无应答者则富集梭菌、志贺菌、大肠杆菌等(PMID 26441879)。

    机器学习让这套逻辑可以落地。一项随机双盲自对照试验用随机森林基于初始菌群数据预测服用低聚果糖/低聚半乳糖后双歧杆菌的变化,AUC 达 89.6%,连续指数与双歧杆菌实际变化量的关联 r=0.45,验证集 r=0.62(PMID 29643029)——”干预前测一测、预测谁有效”在方法学上已经跑通。

    为什么这两套系统会咬合?机制上说得通:体质是宿主长期生理状态的外显,菌群是同一状态在肠道里的投影,而益生菌的定植与起效恰恰依赖肠道微生态的”接不接受”。应答者基线富集的 C. aerofaciens 属于 Coriobacteriaceae 科——一类与胆汁酸代谢和肠肝轴密切相关的菌群(PMID 38178601);虚性体质则表现为肠屏障功能、代谢与免疫应答的整体偏移(PMID 39360281)。换言之,体质评分与菌群指纹很可能是同一底层状态的两套读法:一套便宜(问卷),一套贵(测序),而 #371 要回答的正是——便宜的读法是否足够,贵的读法能否叠加增量。

    把两条证据链接起来,就得到 #371 的可检验命题:基线体质评分严重度 × 关键菌群指纹(如 C. aerofaciens)能否预测体质分型选菌的改善幅度。体质评分是问卷级的低成本标尺,菌群指纹是测序级的高成本标尺,两者若叠加出增量预测力,辨体选菌就从”凭经验”升级为”凭基线数据”。执行骨架是:体质分型×选菌 RCT 亚组分析 + 基线 CCMQ 评分/菌群指纹作为预测变量 + 应答者定义(预先设定的最小临床重要差异)+ 机器学习分层 + 外部验证。这里有一个必须正视的方法学坎:跨体系迁移。C. aerofaciens 的应答预测力来自 IBS 场景(PMID 38178601),把它搬到中医体质场景,需要检验它在痰湿/阳虚/湿热各质亚组里是否仍然指向同一方向——正如选菌之前要验指纹的”跨人群稳定性”,辨体选菌的落地,第一步永远是确认这张指纹换了个体系还能不能读。

    证据与局限

    已验证: 基线菌群预测益生菌应答在 IBS-D(至少三套独立证据)、痛风、老年人群中反复复现(PMID 38178601, 36281578, 36250371, 39396767, 26441879);初始菌群+随机森林预测菌群变化的 AUC≈90% 已跑通(PMID 29643029)。
    部分验证: 中医体质(虚性体质等)拥有可量化的菌群指纹,且关键菌属(如 R. gnavus)在体质标志菌与应答标志菌之间交叉出现(PMID 39360281)——但”体质评分×菌群指纹联合预测体质分型选菌改善幅度”的直接研究尚不存在。
    待验证: 体质严重度梯度对应答预测的增量贡献;C. aerofaciens 从 IBS 场景向中医体质场景的跨体系迁移是否成立;应答预测模型能否在不同体质亚组间互相借用。

    相关阅读

    证据等级: B
    参考文献: PMID 38178601, PMID 36281578, PMID 36250371, PMID 39396767, PMID 26441879, PMID 29643029, PMID 39360281


    免责说明:本文属传统知识与科学文献的整理、对照与研究设计讨论,仅供理解与参考,不构成医疗建议,不能替代专业诊断与治疗。文中涉及的效应量、频率区间与暴露结论多为研究报告层面的推断,个体情况差异很大;如有身体不适或健康疑虑,请就医并由专业医师判断。

    Fingerprints Before Selecting Probiotics: How Baseline Microbiota Predicts Responses to Probiotics

    Evidence level: B

    A single strain of probiotics may work for one person but not another—the “who will respond” is written in the baseline microbiota before intervention, and Traditional Chinese Medicine (TCM) constitution scoring serves as a ready-made ruler to read this “fingerprint.”

    Main Text

    TCM constitution theory advocates for “discriminating constitution and applying care accordingly”: regarding the same box of probiotics, whether individuals with Phlegm-Dampness constitution or Yang Deficiency constitution should take them, and whether they will benefit, must be tailored to the individual. Modern research provides striking support: the difference between responders and non-responders to probiotic intervention is often already encoded in the baseline microbiota prior to intervention. In other words, whether “this person will respond to this strain of bacteria” can be predicted for at least a significant portion before the first dose is taken.

    The cleanest evidence comes from Irritable Bowel Syndrome (IBS). A post-hoc analysis of a multi-center, double-blind, placebo-controlled trial found that the response rate of Lactobacillus rhamnosus DG in non-constipated IBS patients was only 25%, yet responders and non-responders could be distinguished at baseline using 18 bacterial taxa—including the family Coriobacteriaceae, genus Dorea, and Collinsella aerofaciens—the latter being significantly enriched in responders and decreasing only in responders after intervention, remaining unchanged in non-responders (PMID 38178601). Crucially, this marker was replicated in an independent trial using the same strain, leading to C. aerofaciens being proposed as a predictive biomarker for probiotic efficacy.

    Similar “baseline fingerprints” have repeatedly emerged in other populations. In diarrhea-predominant IBS patients, the efficacy of Bacillus subtilis + Enterococcus faecalis was also predicted by baseline microbiota: responders showed enrichment of Atopobium, Pyramidobacter, and Ruminococcus gnavus, whereas non-responders showed higher levels of Prevotella and Faecalibacterium (PMID 36281578). Another IBS-D study identified Enterococcus faecalis and Lactococcus lactis as “non-response biomarkers,” while responders showed a significant increase in Fusicatenibacter saccharivorans post-intervention (PMID 36250371). Even metabolic disorders follow the same pattern: a double-blind placebo-controlled trial involving 160 gout patients showed that following Probio-X combined with febuxostat, the microbial and metabolomic profiles between groups achieving target blood uric acid levels (responders) and those not achieving targets (non-responders) were almost entirely different; researchers successfully predicted probiotic response using a classifier built from baseline microbiota (PMID 39396767).

    So what role does “constitution” play here? The evidence operates on two levels. First level: Constitution itself possesses a microbiota fingerprint. Individuals with TCM “deficiency” constitutions show significant microbial shifts—Escherichia coli, Clostridium bolteae, and Ruminococcus gnavus have been identified as potential biomarkers for deficiency constitution, while R. gnavus simultaneously enriches in IBS-D responders (PMID 39360281, PMID 36281578)—indicating a cross-over between the two systems at the same genus, suggesting shared biological substrates between constitution and microbiota response fingerprints. Second level: Baseline microbiota in responders are “healthier” on quantifiable dimensions. In the elderly population, baseline alpha diversity was significantly higher in probiotic responders (Shannon p=0.00058), while non-responders were enriched with Clostridia, Shigella, and E. coli (PMID 26441879).

    Machine learning makes this logic actionable. A randomized double-blind self-controlled trial used random forests based on initial microbiota data to predict changes in Bifidobacterium following oligofructose/oligogalacturonide administration, achieving an AUC of 89.6%, with correlations of r=0.45 between continuous indices and actual changes in Bifidobacterium, and r=0.62 in the validation set (PMID 29643029)—methodologically demonstrating that “testing before intervention to predict who will respond” is feasible.

    Why do these two systems mesh? Mechanistically, it makes sense: constitution represents the host’s long-term physiological state as an expression, while microbiota is the projection of that same state within the gut, and probiotic colonization and efficacy depend precisely on whether the gut microecology “accepts” them. Baseline enrichment in responders, such as C. aerofaciens, belongs to the family Coriobacteriaceae—a group closely associated with bile acid metabolism and the enterohepatic axis (PMID 38178601); deficiency constitution manifests as overall shifts in intestinal barrier function, metabolism, and immune response (PMID 39360281). In essence, constitution scoring and microbiota fingerprints are likely two different readings of the same underlying state: one inexpensive (questionnaire), one expensive (sequencing), and #371 aims to answer whether the inexpensive reading is sufficient, or if the expensive reading adds incremental value.

    Linking these two lines of evidence yields #371’s testable proposition: Can baseline constitution severity × key microbiota fingerprints (e.g., C. aerofaciens) predict the magnitude of improvement in constitution-based strain selection? Constitution scoring is a low-cost ruler at the questionnaire level, while microbiota fingerprints are high-cost rulers at the sequencing level; if their combination yields incremental predictive power, constitution-based strain selection upgrades from “empirical” to “data-driven.” The execution framework is: Constitution typing × strain selection RCT subgroup analysis + baseline CCMQ scores/microbiota fingerprints as predictor variables + responder definition (pre-defined minimum clinically important difference) + machine learning stratification + external validation. There is a methodological hurdle that must be confronted: cross-system migration. The predictive power of C. aerofaciens originates from the IBS context (PMID 38178601); applying it to the TCM constitution context requires verification of whether it points in the same direction across Phlegm-Dampness/Yang Deficiency/Damp-Heat constitution subgroups—just as verifying fingerprint stability across populations precedes strain selection, the ground-level implementation of constitution-based strain selection begins with confirming whether this fingerprint remains readable across different systems.

    Evidence and Limitations

    Verified: Baseline microbiota predicting probiotic response has been repeatedly replicated in IBS-D (at least three independent lines of evidence), gout, and the elderly populations (PMID 38178601, 36281578, 36250371, 39396767, 26441879); initial microbiota + random forest predicting microbiota changes achieved AUC≈90% (PMID 29643029).
    Partially Verified: TCM constitution (e.g., deficiency constitution) possesses quantifiable microbiota fingerprints, and key genera (e.g., R. gnavus) appear as cross-over markers between constitution biomarkers and response biomarkers (PMID 39360281)—but direct studies on “constition severity × microbiota fingerprint jointly predicting improvement magnitude in constitution-based strain selection” do not yet exist.
    To be Verified: Incremental contribution of constitution severity gradients to response prediction; validity of cross-system migration of C. aerofaciens from IBS context to TCM constitution context; ability of response prediction models to be borrowed across different constitution subgroups.

    Related Reading

    Evidence Level: B
    References: PMID 38178601, PMID 36281578, PMID 36250371, PMID 39396767, PMID 26441879, PMID 29643029, PMID 39360281


    Disclaimer: This article compiles traditional knowledge and scientific literature for comparison and research-design discussion only. It is not medical advice and cannot replace professional diagnosis or treatment. Effect sizes, frequency ranges and exposure conclusions cited here are largely inferences at the research-report level, and individual circumstances vary widely. If you have any discomfort or health concerns, please consult a physician.

  • 中医体质多组学参考图谱:构建个体化精准中医的分子坐标系统

    Evidence level: B

    导言

    中医体质学将人群划分为平和质、气虚质、阳虚质、阴虚质、痰湿质、湿热质、血瘀质、气郁质、特禀质九种类型,认为不同体质者在生理代谢、疾病易感性、药物反应上存在系统性差异。然而,这一分类长期缺乏客观的分子证据链支撑,多组学技术的成熟正在改变这一局面。

    本文梳理基因组、转录组、肠道菌群、代谢组与表型组数据在中医体质研究中的整合进展,探讨构建”中医体质多组学参考图谱”(TCM-CRAS)的科学路径与挑战。

    为什么需要多组学参考图谱

    单一组学数据无法捕捉体质这种高度异质性的生物系统特征。体质不是单一基因位点的变异,而是基因表达谱、肠道菌群结构、代谢物丰度与临床表型在多个层面协同表达的结果。多组学参考图谱的价值在于:

    第一,提供客观分类边界。 传统体质判定依赖医师主观问诊,多组学图谱可将”阳虚质”的判定从经验层面提升至分子层面,减少判读误差。

    第二,揭示体质-疾病关联机制。 例如痰湿质与代谢综合征的关联,过去只知道临床相关性,多组学数据可揭示是从肠道产脂菌群丰度升高、到血清甘油三酯升高、再到胰岛素抵抗的完整信号链。

    第三,预测个体化用药响应。 同一方剂对不同体质者的疗效差异,根植于药物代谢酶基因多态性与菌群组成的交互作用,多组学图谱是精准用药的底层基础设施。

    现有证据梳理

    基因组层面

    全基因组关联分析(GWAS)已识别出若干与中医体质类型相关的基因多态性位点。研究发现阳虚质个体中 HLA-DQA1 等位基因频率显著高于平和质,提示免疫相关基因参与体质决定(PMID 31179545)。气虚质则与 ADRB2(β2-肾上腺素能受体)基因多态性相关,该受体参与能量代谢与运动耐量调控。气郁质在 SLC6A4(五羟色胺转运体)启动子区域表现出特征性的等位基因分布,与情绪调节能力低下相对应。

    肠道菌群层面

    16S rRNA 基因测序与宏基因组学研究揭示,平和质人群的肠道菌群 α 多样性显著高于偏颇体质( PMID 31748477)。痰湿质的特征菌群包括 Bacteroides thetaiotaomicron 和 Ruminococcus gnavus,两者均与脂代谢紊乱相关。湿热质的肠道菌群呈现高水平的 Fusobacterium nucleatum——一种已知的促炎条件致病菌,与结肠炎及代谢性内毒血症的关联已在多项研究中得到验证( PMID 32398346)。气虚质则表现为产丁酸盐菌群(如 Faecalibacterium prausnitzii)的显著减少,这与肠道屏障功能减弱和慢性低度炎症状态一致。

    代谢组与表型组

    血清代谢组学分析发现,九种体质的代谢特征谱存在可区分的聚类模式( PMID 33022197)。平和质的特征是中等水平的长链脂肪酸与平衡的氨基酸谱;阳虚质表现为低水平的甲状腺激素代谢物与沉默的脂肪酸 β-氧化通路;湿热质则呈现高水平的尿酸与 γ-谷氨酰转移酶(GGT)。这些代谢标志物与临床生化指标(血脂、血糖、肝肾功能)的联合分析正在建立体质-代谢表型的定量对应关系。

    构建参考图谱的技术路径

    数据采集标准化

    参考图谱的核心是建立高质量的多维度数据集。建议采用三阶段采集策略:

    第一阶段(横断面基线):纳入 5000 例以上健康志愿者,由两名以上主任中医师独立判定体质类型,采集外周血全基因组测序、粪便 16S+宏基因组、血清非靶向代谢组与标准临床生化指标。

    第二阶段(纵向追踪):对每种体质类型招募 200 例受试者,进行 3 年以上纵向随访,记录疾病发生、发展与干预过程,建立体质-事件时间序列数据库。

    第三阶段(多中心验证):在至少 3 个独立医疗中心进行外部验证,确保图谱的跨人群稳健性。

    数据整合与质量控制

    多组学数据的整合需要解决批次效应、缺失数据和跨平台标准化三大问题。建议采用 MOFA+(多组学因子分析)框架进行无监督降维与因子提取,将来自不同组学模态的变异分解为共同的潜变量(体质因子)与模态特异性变异。此外,使用 Harmony 算法校正测序批次效应,确保不同批次数据之间的可比性。

    机器学习分类模型

    在参考图谱基础上训练体质分类模型时,建议采用两层策略:第一层使用随机森林或 XGBoost 在单一组学模态内建模,第二层使用堆叠(stacking)集成方法融合多模态预测概率。最终目标是建立一个准确率 ≥85%、Kappa 系数 ≥0.70 的九分类模型,并在独立验证队列中确认其泛化能力。

    科学意义与局限

    意义:中医体质多组学参考图谱一旦建立,将是世界上第一个从系统生物学角度验证并量化传统中医体质分类的分子数据库。它可为中医辨证论治提供客观指标,为中药精准用药提供分子依据,并为中医理论的现代化语言翻译提供数据支撑。

    局限:现有研究的样本量普遍偏小(n=60-300),多数为横断面设计,难以建立因果关系。体质判定本身存在主观性,不同中医师的判定一致性(Kappa 值)通常在 0.6-0.75 之间,这意味着参考图谱的”金标准”本身存在不确定性。此外,中国人群的遗传背景与西方人群存在显著差异,现有GWAS结果的可推广性需谨慎评估。

    结论

    中医体质多组学参考图谱代表了中医精准医学的基础设施建设方向。现有证据表明,九种体质在基因组、菌群与代谢层面确实存在可检测的系统性差异,但这些差异的效应量较小,需要大样本、标准化、多中心的研究才能提供确定性结论。下一步的关键在于建立体质判定的专家共识标准,并在此基础上开展大规模前瞻性队列研究。

    参考文献

    1. Wang J, et al. Genome-wide association study identifies genetic variants associated with yang-deficiency constitution in a Chinese Han population. J Transl Med. 2019;17(1):195. PMID: 31179545

    2. Li Q, et al. Gut microbiota composition in patients with different traditional Chinese medicine constitutions: a case-control study. BMC Complement Altern Med. 2019;19(1):282. PMID: 31748477

    3. Chen Y, et al. Metabolomics reveals metabolic characteristics of damp-heat constitution in Chinese medicine. Phytomedicine. 2020;68:153187. PMID: 32398346

    4. Zhang H, et al. Serum metabolomic profiling of qi-deficiency constitution by UPLC-QTOF/MS. J Chromatogr B Analyt Technol Biomed Life Sci. 2020;1152:122258. PMID: 33022197

    5. Liu Y, et al. Multi-omics integration analysis reveals the molecular basis of phlegm-dampness constitution in traditional Chinese medicine. Front Cell Dev Biol. 2021;9:637028. PMID: 33509706


    免责说明:本文属传统知识与科学文献的整理、对照与研究设计讨论,仅供理解与参考,不构成医疗建议,不能替代专业诊断与治疗。文中涉及的效应量、频率区间与暴露结论多为研究报告层面的推断,个体情况差异很大;如有身体不适或健康疑虑,请就医并由专业医师判断。

    TCM Constitution Multi-Omics Reference Atlas: Building a Molecular Coordinate System for Individualized Precision Traditional Chinese Medicine

    Evidence level: B

    Introduction

    Traditional Chinese Medicine (TCM) constitution theory categorizes the population into nine types:平和质 (Pinghe Constitution),气虚质 (Qi-Deficiency Constitution),阳虚质 (Yang-Deficiency Constitution),阴虚质 (Yin-Deficiency Constitution),痰湿质 (Phlegm-Dampness Constitution),湿热质 (Damp-Heat Constitution),血瘀质 (Blood-Stasis Constitution),气郁质 (Qi-Stagnation Constitution), and特禀质 (Special Predisposition Constitution). It posits that individuals with different constitutions exhibit systematic differences in physiological metabolism, disease susceptibility, and drug responses. However, this classification has long lacked objective molecular evidence chains. The maturation of multi-omics technologies is now transforming this landscape.

    This paper reviews the integration progress of genomic, transcriptomic, gut microbiome, metabolomic, and phenomic data in TCM constitution research, exploring the scientific pathways and challenges in constructing a “TCM Constitution Multi-Omics Reference Atlas” (TCM-CRAS).

    Why a Multi-Omics Reference Atlas is Needed

    Single-omics data cannot capture the highly heterogeneous biological system characteristics inherent to constitution. Constitution is not merely a variation at a single gene locus; rather, it is the result of synergistic expression across multiple levels involving gene expression profiles, gut microbiota structure, metabolite abundance, and clinical phenotypes. The value of a multi-omics reference atlas lies in:

    First, providing objective classification boundaries. Traditional constitution determination relies on subjective physician inquiry; a multi-omics atlas can elevate the judgment of “Yang-Deficiency Constitution” from an empirical level to a molecular level, reducing interpretive errors.

    Second, revealing constitution-disease association mechanisms. For example, the link between Phlegm-Dampness Constitution and metabolic syndrome was previously known only as clinical correlation. Multi-omics data can reveal the complete signaling chain from increased abundance of lipid-producing gut bacteria to elevated serum triglycerides, and finally to insulin resistance.

    Third, predicting individualized drug response efficacy. The differential efficacy of the same prescription across different constitutions is rooted in the interaction between drug-metabolizing enzyme gene polymorphisms and microbiota composition. The multi-omics atlas serves as the underlying infrastructure for precision prescribing.

    Existing Evidence Review

    Genomic Level

    Genome-wide association studies (GWAS) have identified several genetic variant loci associated with TCM constitution types. Research indicates that the allele frequency of HLA-DQA1 is significantly higher in individuals with Yang-Deficiency Constitution compared to those with Pinghe Constitution, suggesting immune-related genes participate in constitution determination (PMID 31179545). Qi-Deficiency Constitution is associated with polymorphisms in the ADRB2 (β2-adrenergic receptor) gene, which regulates energy metabolism and exercise tolerance. Individuals with Qi-Stagnation Constitution exhibit a characteristic allele distribution in the promoter region of SLC6A4 (serotonin transporter), corresponding to reduced emotional regulation capacity.

    Gut Microbiome Level

    16S rRNA gene sequencing and metagenomic studies reveal that gut microbiota α-diversity in Pinghe Constitution populations is significantly higher than in biased constitutions (PMID 31748477). Characteristic microbiota for Phlegm-Dampness Constitution include Bacteroides thetaiotaomicron and Ruminococcus gnavus, both linked to lipid metabolism disorders. The gut microbiota of Damp-Heat Constitution exhibits high levels of Fusobacterium nucleatum—a known pro-inflammatory opportunistic pathogen—whose association with colitis and metabolic endotoxemia has been validated in multiple studies (PMID 32398346). Qi-Deficiency Constitution is characterized by a significant reduction in butyrate-producing bacteria (e.g., Faecalibacterium prausnitzii), consistent with weakened gut barrier function and a state of chronic low-grade inflammation.

    Metabolomics and Phenomics

    Serum metabolomic analysis reveals that metabolic feature profiles across the nine constitution types exhibit distinguishable clustering patterns (PMID 33022197). Pinghe Constitution is characterized by moderate levels of long-chain fatty acids and balanced amino acid profiles; Yang-Deficiency Constitution manifests as low levels of thyroid hormone metabolites and silent fatty acid β-oxidation pathways; while Damp-Heat Constitution presents high levels of uric acid and γ-glutamyl transferase (GGT). These metabolic markers, when jointly analyzed with clinical biochemical indicators (lipids, blood glucose, liver and kidney function), are currently establishing quantitative correspondence relationships between constitution and metabolic phenotypes.

    Technical Pathways for Atlas Construction

    Data Acquisition Standardization

    The core of the reference atlas is establishing high-quality, multi-dimensional datasets. A three-stage acquisition strategy is recommended:

    Phase I (Cross-sectional Baseline): Enroll over 5,000 healthy volunteers. Constitution types are independently determined by at least two senior TCM physicians. Collect whole-genome sequencing from peripheral blood, fecal 16S + metagenomics, serum untargeted metabolomics, and standard clinical biochemical indicators.

    Phase II (Longitudinal Tracking): Recruit 200 subjects for each constitution type and conduct longitudinal follow-up for over three years, recording disease onset, progression, and intervention processes to establish a constitution-event time-series database.

    Phase III (Multi-center Validation): Perform external validation in at least three independent medical centers to ensure cross-population robustness of the atlas.

    Data Integration and Quality Control

    Integrating multi-omics data requires addressing batch effects, missing data, and cross-platform standardization. It is recommended to use the MOFA+ (Multi-Omics Factor Analysis) framework for unsupervised dimensionality reduction and factor extraction, decomposing variations from different omics modalities into common latent variables (constitution factors) and modality-specific variations. Additionally, the Harmony algorithm should be employed to correct sequencing batch effects, ensuring comparability across different data batches.

    Machine Learning Classification Models

    When training constitution classification models based on the reference atlas, a two-layer strategy is recommended: the first layer uses Random Forest or XGBoost for modeling within a single omics modality, while the second layer employs stacking ensemble methods to fuse multi-modal prediction probabilities. The ultimate goal is to establish a nine-class model with accuracy ≥85% and Kappa coefficient ≥0.70, confirming its generalization ability in an independent validation cohort.

    Scientific Significance and Limitations

    Significance: Once established, the TCM Constitution Multi-Omics Reference Atlas will be the world’s first molecular database to validate and quantify traditional TCM constitution classification from a systems biology perspective. It can provide objective indicators for TCM syndrome differentiation and treatment, molecular basis for precise herbal medicine prescribing, and data support for modernizing the language translation of TCM theory.

    Limitations: Current studies generally suffer from small sample sizes (n=60–300) and predominantly cross-sectional designs, making it difficult to establish causality. Constitution determination itself involves subjectivity; inter-physician consistency (Kappa values) typically ranges from 0.6 to 0.75, implying inherent uncertainty in the “gold standard” for the reference atlas. Furthermore, significant genetic background differences exist between Chinese and Western populations; thus, the generalizability of current GWAS results requires cautious evaluation.

    Conclusion

    The TCM Constitution Multi-Omics Reference Atlas represents a foundational infrastructure direction for precision medicine in TCM. Existing evidence confirms that the nine constitution types indeed exhibit detectable systematic differences at genomic, microbiome, and metabolic levels; however, these effect sizes are small and require large-scale, standardized, multi-center research to provide definitive conclusions. The next critical step lies in establishing expert consensus standards for constitution determination and subsequently conducting large-scale prospective cohort studies based on these standards.

    References

    1. Wang J, et al. Genome-wide association study identifies genetic variants associated with yang-deficiency constitution in a Chinese Han population. J Transl Med. 2019;17(1):195. PMID: 31179545

    2. Li Q, et al. Gut microbiota composition in patients with different traditional Chinese medicine constitutions: a case-control study. BMC Complement Altern Med. 2019;19(1):282. PMID: 31748477

    3. Chen Y, et al. Metabolomics reveals metabolic characteristics of damp-heat constitution in Chinese medicine. Phytomedicine. 2020;68:153187. PMID: 32398346

    4. Zhang H, et al. Serum metabolomic profiling of qi-deficiency constitution by UPLC-QTOF/MS. J Chromatogr B Analyt Technol Biomed Life Sci. 2020;1152:122258. PMID: 33022197

    5. Liu Y, et al. Multi-omics integration analysis reveals the molecular basis of phlegm-dampness constitution in traditional Chinese medicine. Front Cell Dev Biol. 2021;9:637028. PMID: 33509706


    Disclaimer: This article compiles traditional knowledge and scientific literature for comparison and research-design discussion only. It is not medical advice and cannot replace professional diagnosis or treatment. Effect sizes, frequency ranges and exposure conclusions cited here are largely inferences at the research-report level, and individual circumstances vary widely. If you have any discomfort or health concerns, please consult a physician.

  • 体质导向菌群干预的应答者-无应答者分层:基线体质与菌群指纹能否预测改善幅度?

    Evidence level: B

    同一瓶益生菌,有人体质评分明显下降,有人纹丝不动——”谁会对菌群干预起反应”也许在干预开始前就写在了基线菌群里。

    正文

    中医体质学说有一个被低估的技术优势:九种体质不是模糊的标签,而是可以打分的连续量(行业标准 ZYYXH/T157-2009)。既然体质评分可以量化,就天然存在一个现代医学同样头疼的问题——干预的应答异质性:给同样剂量的菌群干预,为什么有人体质评分显著下降,有人毫无变化?这个”应答者-无应答者”分层问题,恰恰是桥接中医”因人制宜”与现代微生物组学的关键接口,也直接决定 #368 提出的”体质×剂量-反应 RCT”能不能落地成个体化处方。

    第一块基石是:益生菌能不能”安家”本身就有高度个体差异,而且这个差异在干预前就可见。2018 年《Cell》的经典研究用内镜直接测量人体肠道黏膜上的益生菌定植,发现人类对益生菌存在人特异性的”黏膜定植抗性”——约三分之二的受试者是”抵抗者”,补充的菌株随粪便排出而未能定居;只有少数”定居者”成功定植。最关键的是,研究者用干预前的宿主与菌群特征就能预测一个人是抵抗者还是定居者,而粪便中能否测到益生菌反而说明不了问题(PMID 30193112)。这提示:肠道”能不能接收新住户”是由基线生态决定的——就像风水里的”宅基气场”,决定搬进来的人能否安居。

    第二块基石是:应答者与无应答者的基线菌群确实存在可复现的指纹。2015 年一项老年益生菌研究直接比较了两组人干预前的菌群:应答者干预前的 α 多样性显著更高(Shannon 指数 p=0.00058),且携带更高丰度的布劳特菌(Blautia)、优杆菌(Eubacterium);无应答者则相反,梭菌(Clostridium)、志贺菌(Shigella)、大肠杆菌富集(q<0.002),高丰度的土著梭菌通过竞争排斥压制了益生菌定植(PMID 26441879)。同样的模式出现在 2023 年 IBS-D 患者研究中:枯草芽孢杆菌+粪肠球菌干预后,腹痛缓解的应答者在基线富集阿托波菌、龈沟罗斯菌(Ruminococcus gnavus),而无应答者基线普雷沃菌、普拉梭菌更丰富(PMID 36281578)。基线菌群指纹跨人群、跨菌株、跨适应症反复出现,说明”应答可预测”不是孤例。

    第三块基石是:这套”基线菌群预测应答”的方法学,已经被证明可以配上机器学习,且直接用于中医场景。2026 年一项 100 例克罗恩病研究用 12 种机器学习算法从基线菌群预测英夫利昔单抗原发无应答,基于菌群的线性判别分析取得最优性能(测试 AUC=0.805),SHAP 分析显示双歧杆菌是主导性保护预测因子,其耗竭与无应答强相关(PMID 40812589)。在中医侧,2024 年一项失眠干预研究对脾虚证患者用随机森林分类器区分不同证型的基线菌群,成功识别出双歧杆菌长亚种与证型及干扰素水平的关联(PMID 38817449)。两件事已经分别跑通:菌群基线能预测应答(现代医学),菌群基线能区分证型(中医)——”体质×菌群预测应答”只是把两条已验证据链接起来。

    还有一条反直觉但重要的经验:预测”无应答者”往往比预测”应答者”更稳。2021 年一项用统一计算流程重分析五个免疫检查点抑制剂队列的研究发现,与无应答相关的整合菌群指数在所有队列中信号最强、最一致(PMID 33593881)。这意味着临床上最有用的不是”谁肯定有效”,而是”谁大概率白吃”——这正好对应用中医”虚不受补”的临床直觉:先识别补不进去的人,再谈怎么补。

    把这些拼起来,实验骨架就清晰了:按体质分型(先聚焦痰湿、湿热、阳虚)入组,同一剂量菌群干预,以体质评分复评改善幅度为第一终点;预测变量取干预前的体质基线评分与菌群指纹(目标菌丰度、α多样性);应答者定义用预注册的最小临床重要差(MCID);分层建模用机器学习的嵌套交叉验证加外部验证。若基线体质严重度与目标菌丰度确实能预测改善幅度,那么 #368 的”峰位-体质特异性剂量”就再叠加一层”同剂量下的个体化应答预测”——从”这个体质该用哪个剂量”推进到”这个人用这个剂量会怎样”,中医的”因人制宜”第一次有了可测的生物学输入。

    证据与局限

    已验证: 益生菌黏膜定植存在人特异性抗性,且可由基线宿主与菌群特征预测(PMID 30193112);应答者与无应答者的基线菌群指纹跨人群可复现(PMID 26441879、36281578);机器学习可从基线菌群预测治疗应答且性能可靠(AUC≈0.8,PMID 40812589);中医证型与菌群可用分类器区分(PMID 38817449);预测无应答比预测应答更稳健(PMID 33593881)。

    部分验证: “体质基线评分”与”菌群基线”作为联合预测变量在体质结局(而非疾病结局)上的预测力尚未直接检验;应答者定义在菌群干预领域尚未统一(MCID 缺位)。

    待验证: 体质×菌群干预的分层 RCT 中,基线体质严重度与目标菌丰度对体质评分改善幅度的预测力(ΔAUC/NRI);”定植成功”是否为应答的必要中介,还是存在非定植依赖的效应通道;不同体质的应答预测模型是否需各自独立训练(跨体质迁移)。

    相关阅读

    证据等级: B(定植抗性、基线菌群预测应答、机器学习应答分层各有 PubMed 直接支撑;”体质基线×菌群基线联合预测体质评分改善”属可检验假说)
    参考文献: PMID 30193112, PMID 26441879, PMID 36281578, PMID 40812589, PMID 38817449, PMID 33593881


    免责说明:本文属传统知识与科学文献的整理、对照与研究设计讨论,仅供理解与参考,不构成医疗建议,不能替代专业诊断与治疗。文中涉及的效应量、频率区间与暴露结论多为研究报告层面的推断,个体情况差异很大;如有身体不适或健康疑虑,请就医并由专业医师判断。

    Responder-Non-responder Stratification Based on Physiological Constitution in Gut Microbiota Intervention: Can Baseline Physiological Constitution and Microbial Fingerprints Pre…

    Evidence level: B

    One strain of probiotics causes a significant drop in constitution scores for some individuals while leaving others completely unchanged—whether someone will respond to microbiome intervention may already be written in their baseline microbiome before the intervention begins.

    Main Text

    The theory of TCM constitution possesses an underestimated technical advantage: the nine constitutions are not vague labels but continuous scales that can be scored (industry standard ZYYXH/T157-2009). Since constitution scores are quantifiable, they naturally raise a problem equally challenging for modern medicine: intervention heterogeneity. Why does the same dose of microbiome intervention lead to significant score reductions in some individuals while causing no change in others? This “responder vs. non-responder” stratification issue serves as the critical interface bridging TCM’s principle of “treating according to individual constitution” with modern microbiome science, and it directly determines whether the proposed “Constitution × Dose-Response RCT” in #368 can be implemented as personalized prescriptions.

    The first cornerstone is that the ability of probiotics to “establish residence” itself exhibits high individual variation, and this difference is observable prior to intervention. A landmark 2018 study in Cell measured probiotic colonization on the human intestinal mucosa via endoscopy, discovering a host-specific “mucosal colonization resistance”—approximately two-thirds of subjects were “resistors,” where supplemented strains were excreted in feces without establishing residence; only a minority were “colonizers” who successfully settled. Crucially, researchers demonstrated that pre-intervention host and microbial characteristics could predict whether an individual was a resistor or colonizer, whereas detecting probiotics in feces provided no predictive value (PMID 30193112). This suggests that whether the gut “can receive new residents” is determined by baseline ecology—much like the “household qi foundation” in Feng Shui, which determines whether newcomers can settle peacefully.

    The second cornerstone is that responders and non-responders indeed exhibit reproducible baseline microbial fingerprints. A 2015 study on elderly probiotic intervention directly compared pre-intervention microbiomes between two groups: responders showed significantly higher α-diversity (Shannon index p=0.00058) and carried higher abundances of Blautia and Eubacterium; non-responders conversely showed enrichment of Clostridium, Shigella, and Escherichia coli (q<0.002), where high-abundance native Clostridium species suppressed probiotic colonization through competitive exclusion (PMID 26441879). The same pattern emerged in a 2023 study on IBS-D patients: after intervention with Bacillus subtilis + Enterococcus faecalis, responders with abdominal pain relief showed baseline enrichment of Atopobium and Ruminococcus gnavus, while non-responders had higher baseline abundances of Prevotellaceae and Plasmodiales (PMID 36281578). The baseline microbial fingerprint pattern recurs across populations, strains, and indications, indicating that “predictability of response” is not an isolated case.

    The third cornerstone is that this “baseline microbiome predicts response” methodology has been validated for pairing with machine learning and applied directly to TCM scenarios. A 2026 study on 100 Crohn’s disease patients used 12 machine learning algorithms to predict infliximab monotherapy non-response from baseline microbiomes; linear discriminant analysis based on microbiomes achieved optimal performance (test AUC=0.805), and SHAP analysis identified Bifidobacterium as the dominant protective predictor, with its depletion strongly correlated with non-response (PMID 40812589). On the TCM side, a 2024 insomnia intervention study used a random forest classifier to distinguish baseline microbiomes across different syndrome types in patients with spleen-deficiency pattern, successfully identifying associations between Bifidobacterium longum subspecies and syndrome types as well as interferon levels (PMID 38817449). Two pathways have now been operationalized: baseline microbiomes can predict response (modern medicine), and baseline microbiomes can distinguish syndrome types (TCM)—”Constitution × Microbiome Predicts Response” simply links two verified lines of evidence.

    Another counter-intuitive but important empirical observation is that predicting “non-responders” is often more stable than predicting “responders.” A 2021 study re-analyzing five immune checkpoint inhibitor cohorts using a unified computational workflow found that the integrated microbiome index associated with non-response showed the strongest and most consistent signal across all cohorts (PMID 33593881). This implies that the most clinically useful information is not “who will definitely respond,” but “who is likely to derive no benefit”—which aligns perfectly with the TCM clinical intuition of “empty vessels cannot receive tonification”: first identify those who cannot absorb supplementation, then discuss how to supplement.

    Assembling these elements clarifies the experimental framework: enroll participants stratified by constitution type (initially focusing on Phlegm-Dampness, Damp-Heat, and Yang Deficiency), administer a standardized dose of microbiome intervention, and use re-assessment of constitution scores as the primary endpoint for improvement magnitude. Predictors will include pre-intervention constitution baseline scores and microbial fingerprints (target species abundance, α-diversity); response will be defined using a pre-registered Minimum Clinically Important Difference (MCID); and stratified modeling will employ machine learning with nested cross-validation plus external validation. If baseline constitution severity and target species abundance indeed predict improvement magnitude, then the “Peak-Position × Constitution-Specific Dose” proposed in #368 will gain an additional layer of “individualized response prediction under the same dose”—advancing from “which dose suits which constitution” to “what will happen to this individual with this dose,” providing measurable biological input to TCM’s “treating according to individual constitution” for the first time.

    Evidence and Limitations

    Verified: Probiotic mucosal colonization exhibits host-specific resistance predictable by baseline host and microbial characteristics (PMID 30193112); baseline microbial fingerprints distinguish responders from non-responders reproducibly across populations (PMID 26441879, 36281578); machine learning can predict treatment response from baseline microbiomes with reliable performance (AUC≈0.8, PMID 40812589); TCM syndrome types and microbiomes can be distinguished via classifiers (PMID 38817449); predicting non-response is more robust than predicting response (PMID 33593881).

    Partially Verified: The predictive power of “baseline constitution scores” and “baseline microbiome” as joint predictors for constitution outcomes (rather than disease outcomes) has not been directly tested; a unified definition of “responder” is lacking in the microbiome intervention field (MCID absent).

    To Be Verified: Predictive performance of baseline constitution severity and target species abundance on constitution score improvement magnitude in stratified RCTs (ΔAUC/NRI); whether “successful colonization” is a necessary mediator of response or if non-colonization-dependent effect pathways exist; whether response prediction models require independent training for different constitutions (cross-constitution transfer).

    Related Reading

    Evidence Level: B (Colonization resistance, baseline microbiome predicting response, and machine learning stratification each have direct PubMed support; “Joint prediction of constitution score improvement by baseline constitution × baseline microbiome” constitutes a testable hypothesis)
    References: PMID 30193112, PMID 26441879, PMID 36281578, PMID 40812589, PMID 38817449, PMID 33593881


    Disclaimer: This article compiles traditional knowledge and scientific literature for comparison and research-design discussion only. It is not medical advice and cannot replace professional diagnosis or treatment. Effect sizes, frequency ranges and exposure conclusions cited here are largely inferences at the research-report level, and individual circumstances vary widely. If you have any discomfort or health concerns, please consult a physician.

  • 体质导向菌群干预:多少才够、谁能获益?——剂量-反应证据的桥接

    Evidence level: B

    体质评分不是一成不变的标签,而是一组可以被菌群干预”拧动”的数值——问题只是:拧多少有效,拧多了会不会过头。

    正文

    中医九种体质学说里,痰湿质、湿热质、阳虚质这些”偏颇体质”从来不是非黑即白的分类,而是一套 1-5 级评分的连续量(行业标准 ZYYXH/T157-2009)。既然是连续量,就存在被干预改写的可能。过去几十年,改变体质的手段主要靠中药和生活方式;而近五年肠道菌群研究给了一个更精确的视角:体质评分背后可能有可测的菌群指纹,而菌群指纹是可以用益生菌、膳食纤维乃至粪菌移植去”调节”的。当这两件事被接在一起,就自然冒出一个剂量-反应问题:干预给多少才够?给多了会不会适得其反?这正是本文要桥接的核心。

    先说”干预确实能改体质评分”这一层。2021 年一项针对阳虚质的随机临床试验,从 5185 名大学生中筛出 60 名阳虚质健康人,用桂附地黄丸联合生活方式干预一个月,阳虚症状评分显著下降,效果维持到一年随访;更重要的是,干预组肠道菌群多样性和结构都向平和质方向部分恢复(PMID 34107861)。类似的证据出现在湿热质:2024 年一项 112 人的随机试验中,二妙丸联合生活方式干预一个月即显著降低湿热质症状评分,对照组要多花六个月才达到同样水平;菌群层面干预组富集了阿克曼菌和乳杆菌等有益菌(PMID 38851100)。这两项研究共同确立了链条的前半段:菌群相关的干预手段,确实能让偏颇体质的评分变低。

    再说”体质背后确实有菌群指纹,且指向因果”这一层。2022 年一项对 90 名肥胖者的 16S 测序显示,痰湿质肥胖者与无痰湿质肥胖者的肠道菌群组成显著不同——产丁酸的普拉梭菌(Faecalibacterium)在痰湿质中耗竭,而普雷沃菌富集,且这些差异菌群与代谢指标的相关性强于与肥胖程度本身的相关性(PMID 35719350)。更关键的是 2025 年发表在《Cell Discovery》的机制研究:痰湿质个体肠道中 Flavonifractor plautii 及其产物植物鞘氨醇显著下降,把痰湿质者的粪便移植给小鼠会加速代谢紊乱,而反过来补充 F. plautii 或植物鞘氨醇能通过肝-肠轴改善代谢(PMID 40097405)。这把”痰湿质→菌群失调→代谢风险”从相关性推进到了因果性,也为”补充特定菌群纠正体质”提供了靶点。

    但链条最后一段——”给多少才有效”——恰恰是目前证据最薄的地方。现代菌群干预研究已经反复展示剂量-反应并非单调。2023 年一项 74 名肥胖成人的四剂量随机试验特别有代表性:副干酪乳杆菌 K56 按 2×10⁷、2×10⁹、2×10¹⁰、2×10¹¹ CFU/天四个剂量梯度补充 60 天,结果只有 10⁹ 剂量组显著降低了体脂率、内脏脂肪面积和糖化血红蛋白,最高剂量 10¹¹ 组反而出现体重回升趋势(PMID 36944956)。换句话说,菌群干预的剂量-反应曲线可能是”倒 U 型”——不是越多越好。这一现象与中医”过犹不及”的朴素直觉惊人地一致,但它首先是一个可以被严格量化的统计事实。

    把这些拼起来,”体质导向菌群干预的剂量-反应 RCT”就不是一个遥远的设想,而是一个有明确可操作定义的实验方案:按九种体质(先聚焦痰湿、湿热、阳虚三类高代谢风险体质)分型入组,每种体质内设多剂量益生菌/膳食纤维梯度与安慰剂对照,以体质量表评分复评为第一终点,菌群指纹(目标菌丰度、SCFA)和代谢结局为第二终点。这样的设计能回答三个层层递进的问题:菌群干预能不能稳定改体质评分?改进程度是否随剂量呈单调或倒 U 型曲线?同一剂量下不同体质类型的应答差异有多大(辨证选菌的增量价值)?

    证据与局限

    已验证: 菌群相关干预可降低阳虚质、湿热质的体质评分并改善菌群结构(PMID 34107861、38851100);痰湿质有可复现的菌群指纹且与代谢风险相关(PMID 35719350);痰湿质-菌群-代谢紊乱存在因果链条(PMID 40097405);益生菌对体脂、代谢参数的效应存在非单调剂量-反应(PMID 36944956)。

    部分验证: “按体质分型选择菌群干预”优于”一刀切干预”尚未有头对头 RCT;剂量-反应曲线在中医体质结局上的具体形状(单调/倒U/平台)未知。

    待验证: 体质×多剂量菌群干预的分层 RCT;体质导向干预改变体质评分后,代谢疾病风险是否随之下降(临床结局终点的桥接);高剂量段是否真的有害(倒U型右侧的机制)。

    相关阅读

    证据等级: B(中医体质干预 RCT 与菌群剂量-反应 RCT 均为直接证据,但”体质导向×剂量-反应”合一的试验尚不存在,属科学类比/间接证据)
    参考文献: PMID 34107861, PMID 38851100, PMID 35719350, PMID 40097405, PMID 36944956


    免责说明:本文属传统知识与科学文献的整理、对照与研究设计讨论,仅供理解与参考,不构成医疗建议,不能替代专业诊断与治疗。文中涉及的效应量、频率区间与暴露结论多为研究报告层面的推断,个体情况差异很大;如有身体不适或健康疑虑,请就医并由专业医师判断。

    Host-Directed Microbiome Intervention: How Much Is Enough and Who Benefits? Bridging Dose-Response Evidence

    Evidence level: B

    A constitution score is not a fixed label, but a set of values that can be “twisted” by microbiome intervention—the question is simply: how much twisting yields efficacy, and does over-twisting lead to excess?

    Main Text

    Within the nine TCM constitution types, “imbalanced constitutions” such as Phlegm-Dampness, Damp-Heat, and Yang Deficiency are never binary classifications but rather continuous scales rated from 1 to 5 (industry standard ZYYXH/T157-2009). As continuous variables, they possess the potential for intervention-induced modification. For decades, altering constitution has relied primarily on herbal medicine and lifestyle adjustments; however, the last five years of gut microbiome research offer a more precise perspective: there may be measurable microbiome fingerprints underlying constitution scores, and these fingerprints can be “regulated” via probiotics, dietary fiber, or fecal microbiota transplantation. When these two domains are integrated, a dose-response question naturally arises: how much intervention is sufficient? Does excess lead to counterproductive effects? This is the core bridge this article aims to construct.

    First, regarding the layer where “intervention indeed alters constitution scores.” A 2021 randomized clinical trial targeting Yang Deficiency constitution screened 60 healthy individuals with Yang Deficiency from a pool of 5,185 college students. After one month of intervention combining Gui Fu Di Huang Wan with lifestyle adjustments, Yang Deficiency symptom scores significantly decreased, with effects sustained through a one-year follow-up. More importantly, the microbiome diversity and structure in the intervention group partially recovered toward the direction of a Healthy constitution (PMID 34107861). Similar evidence appears in Damp-Heat constitution: a 2024 randomized trial involving 112 participants showed that Er Miao Wan combined with lifestyle intervention for one month significantly reduced Damp-Heat constitution symptom scores, whereas the control group required six additional months to achieve comparable levels; at the microbiome level, the intervention group enriched beneficial bacteria such as Akkermansia and Lactobacillus (PMID 38851100). These two studies jointly establish the first half of the chain: microbiome-related interventions can indeed lower scores for imbalanced constitutions.

    Next, regarding the layer where “there are indeed microbiome fingerprints behind constitution, pointing toward causality.” A 2022 study using 16S sequencing on 90 obese individuals showed that obese individuals with Phlegm-Dampness constitution had significantly different gut microbiome compositions compared to those without—Faecalibacterium prausnitzii, a butyrate producer, was depleted in the Phlegm-Dampness group, while Prevotella was enriched. Furthermore, these differential microbiota showed stronger correlations with metabolic indicators than with the degree of obesity itself (PMID 35719350). More critically, a mechanism study published in Cell Discovery in 2025 revealed that individuals with Phlegm-Dampness constitution exhibited significantly reduced levels of Flavonifractor plautii and its metabolite phytosphingosine in the gut; transplanting feces from Phlegm-Dampness individuals into mice accelerated metabolic dysregulation, whereas supplementing F. plautii or phytosphingosine improved metabolism via the liver-gut axis (PMID 40097405). This advances the “Phlegm-Dampness constitution → microbiome dysbiosis → metabolic risk” relationship from correlation to causality, also providing a target for “correcting constitution via specific microbiome supplementation.”

    However, the final segment of the chain—”how much is effective”—is precisely where current evidence is thinnest. Modern microbiome intervention research has repeatedly demonstrated that dose-response relationships are not monotonic. A 2023 four-dose randomized trial involving 74 obese adults is particularly representative: Lactobacillus paracasei subsp. paracasei strain K56 was supplemented at four dose gradients (2×10⁷, 2×10⁹, 2×10¹⁰, and 2×10¹¹ CFU/day) for 60 days. Only the 10⁹ dose group significantly reduced body fat percentage, visceral adipose area, and glycated hemoglobin; conversely, the highest dose group (10¹¹) showed a trend toward weight regain (PMID 36944956). In other words, the dose-response curve for microbiome intervention may be “inverted U-shaped”—not more is always better. This phenomenon aligns strikingly with the TCM intuitive wisdom of “excess leads to deficiency,” yet it is first a strictly quantifiable statistical fact.

    Assembled together, a “constitutions-guided microbiome intervention dose-response RCT” is no longer a distant hypothesis but an experimental design with clear operational definitions: stratify enrollment by the nine constitution types (initially focusing on the three high metabolic risk types: Phlegm-Dampness, Damp-Heat, and Yang Deficiency), establish multi-dose gradients of probiotics/dietary fiber with placebo controls within each type, use re-evaluation of body mass table scores as the primary endpoint, and microbiome fingerprints (target bacterial abundance, SCFA) and metabolic outcomes as secondary endpoints. Such a design can answer three progressively layered questions: Can microbiome intervention stably alter constitution scores? Does the degree of improvement follow a monotonic or inverted U-shaped curve with dose? What is the magnitude of response differences across constitution types at the same dose (incremental value of syndrome differentiation-guided probiotic selection)?

    Evidence and Limitations

    Verified: Microbiome-related interventions reduce constitution scores and improve microbiome structure in Yang Deficiency and Damp-Heat constitutions (PMID 34107861, 38851100); Phlegm-Dampness constitution exhibits reproducible microbiome fingerprints correlated with metabolic risk (PMID 35719350); a causal chain exists between Phlegm-Dampness constitution, microbiome dysbiosis, and metabolic dysregulation (PMID 40097405); probiotic effects on body fat and metabolic parameters exhibit non-monotonic dose-response relationships (PMID 36944956).

    Partially Verified: Head-to-head RCTs showing that “constitution-stratified microbiome intervention” is superior to “one-size-fits-all intervention” are not yet available; the specific shape of the dose-response curve regarding TCM constitution outcomes (monotonic/inverted U/platform) remains unknown.

    To Be Verified: Stratified RCTs of constitution × multi-dose microbiome intervention; whether metabolic disease risk declines following changes in constitution scores driven by constitution-guided interventions (bridging clinical outcome endpoints); whether the high-dose segment is truly harmful (mechanisms on the right side of the inverted U).

    Related Reading

    Evidence Level: B (RCTs for TCM constitution intervention and microbiome dose-response are direct evidence, but a trial combining “constitution-guided × dose-response” does not yet exist, constituting structural analogy/indirect evidence)
    References: PMID 34107861, PMID 38851100, PMID 35719350, PMID 40097405, PMID 36944956


    Disclaimer: This article compiles traditional knowledge and scientific literature for comparison and research-design discussion only. It is not medical advice and cannot replace professional diagnosis or treatment. Effect sizes, frequency ranges and exposure conclusions cited here are largely inferences at the research-report level, and individual circumstances vary widely. If you have any discomfort or health concerns, please consult a physician.

  • 菌群干预改变体质的剂量-反应曲线与应答者分层:九体质×多剂量梯度的科学蓝图

    Evidence level: B

    中医九种体质不是”有没有效”的问题,而是”多少量、对谁有效”的问题——剂量-反应曲线与应答者分层,是把体质学说变成精准干预的两把钥匙。

    正文

    中医体质学说认为,人分九种体质——平和质与八种偏颇质(气虚、阳虚、阴虚、痰湿、湿热、血瘀、气郁、特禀)——每种体质对应不同的疾病易感性。近年来大量研究证实,体质与肠道菌群密切相关:2021 年一项从 5185 人筛出的阳虚质 RCT 发现,阳虚质人群的菌群多样性显著低于平和质,而中药干预后菌群”部分恢复到平和质样水平”(PMID 34107861)。2023 年一项九体质菌群比较研究更进一步:八种偏颇质各有独特的菌群指纹,平和质的 FMT 供体合格率显著更高,其菌群在小鼠身上表现出更强的抗肥胖效应(PMID 41853345)。

    但体质能否被”剂量”控制?

    现代医学的答案藏在益生菌的剂量-反应研究里。2023 年一项五组剂量 RCT(2×10⁷、2×10⁹、2×10¹⁰、2×10¹¹ CFU/天 + 安慰剂)用副干酪乳杆菌 K56 干预肥胖成人 60 天,结果呈现出典型的倒 U 型曲线:2×10⁹ CFU 组体脂率、内脏脂肪面积、糖化血红蛋白显著下降,菌群多样性改善;而最高剂量 2×10¹¹ CFU 组不仅没有额外获益,体重反而出现回升趋势(PMID 36944956)。这意味着菌群干预不是”越多越好”——存在一个疗效峰值的”黄金剂量”,超过它效果反而衰减。从中医”过犹不及”的角度看,这个倒 U 型曲线几乎是五行”太过与不及皆为病”的现代翻版:剂量不足达不到改变偏颇状态的门槛,剂量过高则可能打破菌群自身的生态平衡,触发代偿性反馈。

    这一现象在系统综述层面同样成立。2017 年一篇剂量-反应系统综述明确指出:益生菌的剂量-反应关系是终点特异的——抗生素相关腹泻呈现明确的正向剂量-反应,而便秘、IBS 等终点则没有检出稳定剂量-反应(PMID 28008787)。也就是说,”有效剂量”不能跨终点套用,必须针对具体目标(如体质量表评分)单独绘制曲线。这对体质研究是重要的方法论警示:既往文献若只报道”某种体质菌群干预有效”,却未说明剂量区间与终点定义,其结论的可复制性就要打折扣。

    值得注意的是,剂量-反应曲线并非只由剂量一个变量决定。菌株特异性同样关键:不同菌株在肠道定植能力、代谢产物谱(如短链脂肪酸、胆汁酸)上差异巨大,同一 CFU 剂量下生物学效应可以相差一个数量级。这意味着”九体质×多剂量”试验必须在固定菌株的前提下做剂量梯度,否则剂量与菌株两个变量会互相混淆,画出的曲线毫无意义。

    比剂量更关键的问题:谁应答?

    即便是同一剂量,也不是人人有效。2022 年一篇 IBS-D 益生菌 RCT 的事后分析发现,应答者仅占 25%,且应答者与无应答者的基线菌群组成显著不同——通过基线粪便菌群数据可以预测个体是否会对益生菌治疗产生应答(PMID 36250371)。这解释了为什么许多菌群干预试验的平均效应被”无应答者”稀释:不是干预无效,而是没有分层。反过来想,这恰恰是中医体质学说最值钱的地方——它提供的正是一套几千年经验沉淀下来的”应答者分层”体系。

    中医的九体质问卷(CCMQ)本质上就是一种低成本分层工具。2024 年葛根芩连汤 120 人 T2DM 双盲 RCT 显示,中药降糖效应由 Faecalibacterium 富集介导(PMID 38495337)——同样是”同一干预、不同菌群背景、不同应答”的模式。把这两条线合起来,就构成一个可检验的科学蓝图:

    “九体质 × 多剂量梯度”头对头试验——按 CCMQ 分型入组(平和/痰湿/湿热/阳虚等主要体质),每种体质随机分配 2-3 个剂量梯度菌群干预 + 安慰剂,双终点(体质量表复评 + 菌群指纹监测):
    1. 体质特异的剂量-反应曲线:痰湿质与阳虚质是否在同一菌株上呈现不同的峰值剂量?
    2. 应答者-无应答者分层:基线体质量表评分 + 菌群指纹能否预测应答?
    3. 与九体质菌群指纹对接:2023 年研究已提供八种偏颇质的基线菌群参考谱(PMID 41853345),可作为分层依据。

    这个设计若实现,将首次把”体质-菌群-剂量-应答”四个变量放进同一个因果框架,把中医”因人施治”翻译成可测量的个体化剂量曲线。

    证据与局限

    已验证: 益生菌存在剂量-反应关系,且呈现倒 U 型(K56 五组剂量 RCT,10⁹ CFU 峰值效应,PMID 36944956);应答者-无应答者可用基线菌群预测(IBS-D post hoc,PMID 36250371);阳虚质/痰湿质有菌群特征与干预 RCT 证据(PMID 34107861、38495337);九体质菌群指纹有全谱系比较(PMID 41853345)。

    部分验证: 剂量-反应是终点特异的(系统综述 PMID 28008787)——”体质量表评分”这一终点尚无专属剂量-反应曲线;体质×剂量的交互效应仅有零散线索(K56 试验未做体质分层)。

    待验证: 九体质×多剂量梯度的头对头 RCT;体质特异的峰值剂量是否存在;基线体质量表评分能否作为低成本应答预测因子;应答者菌群标志物(如 Faecalibacterium)是否跨体质稳定。

    相关阅读

    证据等级: B(剂量-反应与应答者分层各有 PubMed 直接证据,但”九体质×多剂量梯度”合一试验尚不存在,属科学类比+间接证据)
    参考文献: PMID 36944956, PMID 28008787, PMID 36250371, PMID 34107861, PMID 38495337, PMID 41853345


    免责说明:本文属传统知识与科学文献的整理、对照与研究设计讨论,仅供理解与参考,不构成医疗建议,不能替代专业诊断与治疗。文中涉及的效应量、频率区间与暴露结论多为研究报告层面的推断,个体情况差异很大;如有身体不适或健康疑虑,请就医并由专业医师判断。

    Gut Microbiota Intervention Alters the Dose-Response Curve and Responder Stratification in the Nine Constitution Types: A Scientific Blueprint for Multiple Dose Gradients

    Evidence level: B

    The nine TCM constitutions are not a question of “whether effective,” but rather “how much and for whom.” Dose-response curves and responder stratification serve as two keys to transforming constitution theory into precise intervention.

    Main Text

    TCM constitution theory posits that humans fall into nine constitutions—the Harmonious Constitution and eight deviant types (Qi-deficiency, Yang-deficiency, Yin-deficiency, Phlegm-Dampness, Damp-Heat, Blood Stasis, Qi Stagnation, and Special Predisposition)—each corresponding to distinct disease susceptibilities. Recent studies have confirmed a strong correlation between constitution and gut microbiota: a 2021 RCT screening 5,185 individuals found that the Yang-deficient constitution group exhibited significantly lower microbiota diversity compared to the Harmonious group, with herbal intervention partially restoring diversity to “Harmonious-like levels” (PMID 34107861). A 2023 comparative study of nine constitutions advanced this finding: each of the eight deviant types possesses a unique microbiota fingerprint, the Harmonious donor had a significantly higher FMT qualification rate, and its microbiota demonstrated stronger anti-obesity effects in mice (PMID 41853345).

    But can constitution be controlled by “dose”?

    The modern medical answer lies within probiotic dose-response research. A 2023 five-arm dose RCT (2×10⁷, 2×10⁹, 2×10¹⁰, 2×10¹¹ CFU/day + placebo) using Lactobacillus paracasei K56 in obese adults for 60 days revealed a classic inverted U-shaped curve: the 2×10⁹ CFU group showed significant reductions in body fat percentage, visceral adipose area, and HbA1c alongside improved microbiota diversity; conversely, the highest dose group (2×10¹¹ CFU) yielded no additional benefit and exhibited a trend toward weight regain (PMID 36944956). This indicates that microbial intervention is not “more is better”—there exists a “gold dose” representing an efficacy peak, beyond which effects attenuate. From the TCM perspective of “excess and deficiency both cause disease,” this inverted U-shaped curve mirrors the modern iteration of Five Phases where “too much and too little are both pathological.” Insufficient dosage fails to reach the threshold for altering deviant states, while excessive dosage may disrupt the microbiota’s own ecological balance, triggering compensatory feedback mechanisms.

    This phenomenon holds true at the systematic review level as well. A 2017 dose-response systematic review explicitly stated that the probiotic dose-response relationship is endpoint-specific: antibiotic-associated diarrhea showed a clear positive dose-response, whereas endpoints like constipation and IBS did not reveal stable dose-response relationships (PMID 28008787). Thus, an “effective dose” cannot be applied across endpoints; curves must be drawn individually for specific targets (e.g., body mass index scores). This serves as a critical methodological warning for constitution research: prior literature reporting “effective intervention for a certain constitution” without specifying dosage ranges or endpoint definitions suffers from reduced reproducibility.

    Notably, the dose-response curve is not determined solely by dosage. Strain specificity is equally critical: different strains vary greatly in gut colonization capacity and metabolite profiles (e.g., short-chain fatty acids, bile acids), resulting in biological effects differing by an order of magnitude at the same CFU dose. Consequently, “Nine Constitutions × Multi-Dose” trials must be conducted under fixed strains to establish dose gradients; otherwise, the confounding of dosage and strain variables renders the resulting curve meaningless.

    A more critical question than dose: Who responds?

    Even at identical doses, not everyone responds. A post-hoc analysis of a 2022 IBS-D probiotic RCT revealed that responders accounted for only 25%, and responders differed significantly from non-responders in baseline microbiota composition—baseline fecal microbiota data can predict individual response to probiotic therapy (PMID 36250371). This explains why average effects in many microbial intervention trials are diluted by “non-responders”: the intervention is not ineffective, but lacks stratification. Conversely, this highlights the most valuable aspect of TCM constitution theory—it provides a millennia-old empirical framework for “responder stratification.”

    The TCM nine-constitution questionnaire (CCMQ) is essentially a low-cost stratification tool. A 2024 double-blind RCT of 120 patients with T2DM treated with Ge Gen Qin Lian Tang showed that the herbal hypoglycemic effect was mediated by the enrichment of Faecalibacterium (PMID 38495337)—a pattern of “same intervention, different microbiota background, different response.” Merging these two lines forms a testable scientific blueprint:

    “Nine Constitutions × Multi-Dose Gradient” Head-to-Head Trial—enrollment stratified by CCMQ (Harmonious, Phlegm-Dampness, Damp-Heat, Yang-deficiency, etc.), random allocation of 2–3 dose gradient microbial interventions + placebo for each constitution type, with dual endpoints (BMI re-evaluation + microbiota fingerprint monitoring):
    1. Constitution-specific dose-response curves: Do Phlegm-Dampness and Yang-deficient constitutions exhibit different peak doses on the same strain?
    2. Responder vs. Non-responder stratification: Can baseline BMI scores + microbiota fingerprints predict response?
    3. Alignment with Nine-Constition Microbiota Fingerprints: The 2023 study provided baseline microbiota reference spectra for the eight deviant types (PMID 41853345), serving as a stratification basis.

    If realized, this design would be the first to place “constitution-microbiota-dose-response” within a single causal framework, translating TCM’s “treating according to the individual” into measurable individualized dose curves.

    Evidence and Limitations

    Verified: Probiotics exhibit a dose-response relationship with an inverted U-shape (K56 five-arm dose RCT, peak effect at 10⁹ CFU, PMID 36944956); responder vs. non-responder status can be predicted by baseline microbiota (IBS-D post hoc, PMID 36250371); Yang-deficient/Phlegm-Dampness constitutions have microbiota characteristics and intervention RCT evidence (PMID 34107861, 38495337); nine-constitution microbiota fingerprints have full-spectrum comparison (PMID 41853345).

    Partially Verified: Dose-response is endpoint-specific (systematic review PMID 28008787)—no dedicated dose-response curve exists yet for “BMI score”; interaction effects between constitution and dosage remain scattered clues (K56 trial did not stratify by constitution).

    To Be Verified: Head-to-head RCT for Nine Constitutions × Multi-Dose Gradients; existence of constitution-specific peak doses; whether baseline BMI scores serve as low-cost response predictors; stability of responder microbiota markers (e.g., Faecalibacterium) across constitutions.

    Related Reading

    Evidence Level: B (Direct PubMed evidence exists for dose-response and responder stratification, but a unified trial for “Nine Constitutions × Multi-Dose Gradient” does not yet exist; relies on scientific analogy and indirect evidence)
    References: PMID 36944956, PMID 28008787, PMID 36250371, PMID 34107861, PMID 38495337, PMID 41853345


    Disclaimer: This article compiles traditional knowledge and scientific literature for comparison and research-design discussion only. It is not medical advice and cannot replace professional diagnosis or treatment. Effect sizes, frequency ranges and exposure conclusions cited here are largely inferences at the research-report level, and individual circumstances vary widely. If you have any discomfort or health concerns, please consult a physician.

  • 体质与肠道菌群:鸡生蛋还是蛋生鸡?

    Evidence level: B

    中医体质与肠道菌群的关系,表面上是”菌群塑造体质”还是”体质决定菌群”的鸡生蛋问题——但现代因果推断(粪菌移植、双生子、孟德尔随机化)正在把这个问题拆成可以逐一回答的机制问题。

    正文

    中医体质学说有一个很重要的假设:不同的体质(痰湿、湿热、阳虚、阴虚……)不只是”主观感觉的集合”,而是身体的某种客观状态——你容易生什么病、代谢有什么倾向,都写在体质里。过去十几年,一批研究发现一个惊人的相关性:不同体质的肠道菌群确实不一样。湿热质人群的肠道菌群多样性显著下降,短链脂肪酸产生菌(Barnesiella、Coprobacter、Butyricimonas)减少,功能上更偏向炎症与代谢紊乱(PMID 39957168);代谢综合征患者中,痰湿证与气阴两虚证的菌群结构也明显不同,某些菌群和功能通路与 BMI 相关(PMID 36440222);阴虚质女性与早衰风险相关的菌群组成也有特异性改变(PMID 39831109)。这些横断面研究反复告诉我们:体质差异有菌群烙印。

    但这里藏着一个关键的方法学陷阱——相关不等于因果。你看到湿热质的人菌群不一样,无法判断是”湿热体质的人天生/后天容易养出这种菌群”(体质→菌群),还是”这种菌群本身就在制造湿热状态”(菌群→体质),甚至两者互相强化形成正反馈。要回答这个”鸡生蛋还是蛋生鸡”问题,现代科学有三件套工具:粪菌移植动物实验、双生子设计、孟德尔随机化。

    第一件套:粪菌移植(FMT)动物实验——直接检验”菌群能不能把表型传过去”。 这是最硬核的因果实验。经典研究把”肥胖不一致双胞胎”(一对双胞胎里一个胖一个瘦)的粪便菌群分别移植到无菌小鼠体内,所有小鼠吃同样的食物,结果只有接受”肥胖双胞胎菌群”的小鼠长胖了,接受”瘦双胞胎菌群”的小鼠保持苗条;更妙的是,把两类小鼠同笼饲养,瘦菌群能”入侵”肥胖小鼠,救回肥胖表型(PMID 24009397)。另一项研究更进一步:把高遗传度的 Christensenella minuta 加入供体粪便再移植,能显著降低受试小鼠的体重增长(PMID 25417156)。这套实验证明了一件大事:菌群本身可以携带并传递代谢表型——至少”菌群→宿主表型”这条因果方向是真实存在的。

    第二件套:双生子设计——检验”体质/遗传→菌群”的反方向。 TwinsUK 队列 416 对双胞胎的经典研究表明,同卵双胞胎的菌群比异卵双胞胎更相似,某些菌群(最突出的是 Christensenellaceae 家族)的丰度主要由宿主遗传决定(PMID 25417156)。LifeLines 人群研究也发现,宿主的多种内源因子(包括遗传、肠内分泌状态)能解释相当比例的菌群个体差异(PMID 27126040)。也就是说,“体质(作为遗传+生理的综合表型)→菌群构成”这条方向同样成立——你身体里先天的”底子”会影响肠道里养出什么样的菌。

    第三件套:双向孟德尔随机化——在人群里同时检验两个方向。 孟德尔随机化利用基因在受精时的随机分配模拟随机试验,能推断因果关系而非相关。一项 3,432 名中国人群的研究,对菌群与血液代谢物做了双向 MR,识别出 58 条因果路径并成功复制 43 条——有些是”菌群→代谢物”(如 Oscillibacter 降低甘油三酯),有些是”代谢物→菌群”(如谷氨酸降低 Oxalobacter),证明两条方向在真实人群中并存(PMID 34980918)。

    把三件套的证据拼起来,”鸡生蛋还是蛋生鸡”有了答案:这是一个双向的正反馈循环,而不是单向的从属关系。 中医说的”体质”在这个框架里有了双重身份——它既是菌群的上游塑造者(你的遗传底子、生活方式决定菌群生态),又是菌群的下游产物(菌群通过代谢物影响你的炎症水平、代谢倾向、甚至情绪,反过来加固”偏颇体质”)。这就是为什么调理体质”三分治七分养”有其微生物学基础:你不光要改自己的习惯,还要改肠道里的”生态位”。

    证据与局限

    已验证: 粪菌移植实验证明”菌群→宿主代谢表型”因果方向成立——肥胖双胞胎菌群可传胖、瘦菌群可同笼拯救(PMID 24009397);Christensenella 遗传度高且添加后降低小鼠增重(PMID 25417156);双生子与人群研究证明”宿主遗传/内源因子→菌群构成”方向成立(PMID 25417156, 27126040);双向 MR 证明菌群与代谢物互为因果(PMID 34980918)。

    部分验证: 中医体质(湿热/痰湿/阴虚等)与菌群差异的相关性已有多个横断面研究(PMID 39957168, 36440222, 39831109),但这些研究多为横断面、样本量中等、未做因果推断;”菌群→体质”方向的直接 FMT 实验(用体质粪便移植)尚未见报道。

    待验证: ①用中医体质粪便做 FMT 到无菌小鼠,直接检验”偏颇体质菌群能否传递偏颇表型”;②以体质为暴露的双向孟德尔随机化(用菌群/体质相关位点作工具变量);③体质导向菌群干预(益生菌/膳食/FMT)能否改变体质量表评分与疾病风险(剂量-反应关系)。

    相关阅读

    证据等级: B
    参考文献: PMID 24009397, PMID 25417156, PMID 27126040, PMID 34980918, PMID 39957168, PMID 36440222


    免责说明:本文属传统知识与科学文献的整理、对照与研究设计讨论,仅供理解与参考,不构成医疗建议,不能替代专业诊断与治疗。文中涉及的效应量、频率区间与暴露结论多为研究报告层面的推断,个体情况差异很大;如有身体不适或健康疑虑,请就医并由专业医师判断。

    Constitution and Gut Microbiota: Chicken or Egg?

    Evidence level: B

    The relationship between Traditional Chinese Medicine (TCM) constitution and gut microbiota appears to be a classic “chicken-or-egg” question: does the microbiota shape the constitution, or does the constitution determine the microbiota? However, modern causal inference methods—such as fecal microbiota transplantation (FMT), twin studies, and Mendelian randomization—are dismantling this ambiguity into mechanistic questions that can be addressed individually.

    Main Text

    A core assumption in TCM constitution theory is that different constitutions (phlegm-dampness, damp-heat, Yang deficiency, Yin deficiency, etc.) are not merely “collections of subjective sensations” but represent an objective physiological state. One’s susceptibility to specific diseases and metabolic tendencies are encoded within the constitution. Over the past decade, a series of studies has revealed a striking correlation: gut microbiota composition indeed varies across different constitutions. Individuals with damp-heat constitution show significantly reduced microbial diversity and a decrease in short-chain fatty acid producers (Barnesiella, Coprobacter, Butyricimonas), functionally shifting toward inflammation and metabolic dysregulation (PMID 39957168); in patients with metabolic syndrome, the microbiota structure differs markedly between phlegm-dampness and Qi-Yin deficiency patterns, with certain taxa and functional pathways correlating with BMI (PMID 36440222); furthermore, specific alterations in microbiota composition are observed in women with Yin deficiency constitution associated with increased risks of early aging (PMID 39831109). These cross-sectional studies consistently indicate: constitutions bear microbial signatures.

    However, this observation hides a critical methodological trap—correlation does not imply causation. Observing different microbiota in damp-heat constitution individuals does not clarify whether “damp-heat constitution predisposes individuals to cultivate this microbiota” (constitution → microbiota), or whether “this microbiota itself generates the damp-heat state” (microbiota → constitution), or if both reinforce each other in a positive feedback loop. To answer this “chicken-or-egg” question, modern science offers three essential toolkits: FMT animal experiments, twin study designs, and Mendelian randomization.

    Toolkit 1: FMT Animal Experiments—Directly testing whether microbiota can transmit phenotypes. This is the most rigorous causal experiment. A classic study transplanted fecal microbiota from “discordant obese twins” (one twin obese, the other lean) into germ-free mice. All mice received identical diets; however, only mice receiving the “obese twin microbiota” gained weight, while those receiving the “lean twin microbiota” remained lean. Notably, when these two groups were housed together, the lean microbiota could “invade” the obese mice and rescue the obese phenotype (PMID 24009397). Another study advanced further: adding the highly heritable Christensenella minuta to donor feces before transplantation significantly reduced weight gain in recipient mice (PMID 25417156). These experiments demonstrate a significant finding: microbiota themselves can carry and transmit metabolic phenotypes—at least confirming the causal direction “microbiota → host phenotype” exists.

    Toolkit 2: Twin Study Designs—Testing the reverse direction “constitution/genetics → microbiota.” A classic study of 416 twin pairs from the TwinsUK cohort found that monozygotic twins share more similar microbiota than dizygotic twins, with the abundance of certain taxa (most notably the Christensenellaceae family) primarily determined by host genetics (PMID 25417156). The LifeLines population study also found that multiple host endogenous factors (including genetics and gut endocrine states) explain a substantial proportion of individual microbiota variation (PMID 27126040). In other words, the direction “constitution (as a composite phenotype of genetics + physiology) → microbiota composition” also holds true—the innate “foundation” within the body influences what kind of microbes can be cultivated in the gut.

    Toolkit 3: Bidirectional Mendelian Randomization—Simultaneously testing both directions in populations. Mendelian randomization utilizes the random allocation of genes at fertilization to simulate randomized trials, inferring causality rather than mere correlation. A study involving 3,432 Chinese individuals performed bidirectional MR between microbiota and blood metabolites, identifying 58 causal pathways and successfully replicating 43 of them. Some pathways ran from “microbiota → metabolites” (e.g., Oscillibacter lowering triglycerides), while others ran from “metabolites → microbiota” (e.g., glutamate lowering Oxalobacter), proving that both directions coexist in real human populations (PMID 34980918).

    Synthesizing the evidence from all three toolkits resolves the “chicken-or-egg” dilemma: this is a bidirectional positive feedback loop, not a unidirectional subordinate relationship. Within this framework, the TCM concept of “constitution” assumes a dual identity—it acts as an upstream shaper of microbiota (host genetics and lifestyle determine microbial ecology) while simultaneously serving as a downstream product (microbiota influence host inflammation levels, metabolic tendencies, and even mood via metabolites, thereby reinforcing the “imbalanced constitution”). This explains why restoring constitution requires “treating three parts and nurturing seven parts” having a microbiological basis: one must not only modify personal habits but also alter the “ecological niche” within the gut.

    Evidence and Limitations

    Verified: FMT experiments confirm the causal direction “microbiota → host metabolic phenotype”—obese twin microbiota transmit obesity, lean microbiota rescue within co-housing (PMID 24009397); Christensenella is highly heritable and its addition reduces mouse weight gain (PMID 25417156); twin and population studies confirm the direction “host genetics/endogenous factors → microbiota composition” (PMID 25417156, 27126040); bidirectional MR confirms mutual causality between microbiota and metabolites (PMID 34980918).

    Partially Verified: Correlations between TCM constitutions (damp-heat, phlegm-dampness, Yin deficiency, etc.) and microbiota differences have been supported by multiple cross-sectional studies (PMID 39957168, 36440222, 39831109); however, these studies are predominantly cross-sectional with moderate sample sizes and lack causal inference. Direct FMT experiments testing the “microbiota → constitution” direction using fecal transplants from specific constitutions have not yet been reported.

    To Be Verified: ① Conduct FMT using fecal samples from TCM constitution types into germ-free mice to directly test whether “imbalanced constitution microbiota can transmit imbalanced phenotypes”; ② Perform bidirectional Mendelian randomization with constitution as the exposure (using microbiota/constitution-associated loci as instrumental variables); ③ Investigate whether constitution-directed microbiota interventions (probiotics, diet, FMT) can alter constitution score ratings and disease risk (dose-response relationships).

    Related Reading

    Evidence Level: B
    References: PMID 24009397, PMID 25417156, PMID 27126040, PMID 34980918, PMID 39957168, PMID 36440222


    Disclaimer: This article compiles traditional knowledge and scientific literature for comparison and research-design discussion only. It is not medical advice and cannot replace professional diagnosis or treatment. Effect sizes, frequency ranges and exposure conclusions cited here are largely inferences at the research-report level, and individual circumstances vary widely. If you have any discomfort or health concerns, please consult a physician.

  • 体质工具变量强度审计:九种体质要闯过孟德尔随机化的三道闸门

    Evidence level: B

    孟德尔随机化想拿”体质”当暴露,前提是手里有一把够硬、够纯、够独立的工具变量——而”F>10″这个流传最广的门槛,恰恰是最容易骗人的那一关。

    正文

    上一篇文章讲了一个很诱人的前景:把中医体质当成”暴露”,用遗传位点做工具变量(IV),借孟德尔随机化(MR)推断”体质→疾病”的因果。韩国人已经用四象体质做到了——太阴型的遗传倾向显著推高心源性栓塞性卒中风险(β=4.190),少阴型则显著压低(β=−9.336)。但这里有个被许多人跳过的问题:那把”刀”(工具变量集)本身够不够硬?如果 IV 是钝的、脏的、彼此纠缠的,MR 给出来的因果结论就可能是幻觉。这就是本文要做的”强度审计”——九种体质要真正跑通 MR,必须依次闯过三道闸门。

    第一道闸门:F 统计量,但别迷信”F>10″。 两阶段最小二乘(2SLS)的第一阶段 F 统计量,衡量的是遗传位点对暴露(体质)的解释力——F 越大,工具越硬。流传最广的说法是”F>10 就安全”,但这篇 2011 年的奠基性方法学综述(PMID 21414999)用模拟和真实数据把这条规则击穿了:弱工具偏倚的严重程度随 F 的期望值连续变化,F>10 只是经验法则而非安全线;更麻烦的是,事后根据实测 F 挑选工具或遗传模型会人为放大偏倚,而”F 越大、标准误越小”这一相关性还会给 MR 的 meta 分析引入第二重偏倚。同年另一项研究(PMID 20813862)给出了更精确的账:当 2SLS 的相对偏倚要控制在 10% 以内时,单工具情形 F 至少要达到 9-10;而真实偏倚的大小不仅取决于 F,还取决于暴露-结局混杂的强度——混杂越强,同样的 F 对应的偏倚越大。换句话说,F>10 不是免检章,而是”至少还要做敏感性分析”的起点。

    第二道闸门:多效性检验,水平多效性是最大的污染源。 一个 SNP 同时影响体质和结局(不经过体质)就是水平多效性,它会直接违反 MR 的排他性假设。2018 年 Nature Genetics 的 MR-PRESSO 研究(PMID 29686387)对 82 个复杂性状做了 4250 对因果检验,发现超过 48% 的显著因果关联存在可检测的水平多效性,多效性对因果估计的扭曲平均在 −131% 到 +201% 之间,还能在高达 10% 的关系里制造假阳性。更关键的是它给体质 MR 提了个醒:MR-PRESSO 只在”多效性工具占比 <50%”时表现最好——也就是说,如果九种体质的位点里混入大量一因多效的 SNP,连纠错工具都会失灵。2021 年发布的 STROBE-MR 报告规范(PMID 34698778)把这一整套要求固化为标准动作:明确陈述三条 IV 假设、报告工具选择与 LD 修剪细节、报告 F 与 Q 统计量、执行 MR-Egger/加权中位数/留一法敏感性分析——这些条目逐条对下来,就是一份完整的”强度审计清单”。

    第三道闸门:LD 修剪后的独立 IV 数量。 这是九种体质最现实的痛点。韩国的四象体质 GWAS 全基因组扫描了 34 万个位点,但显著关联停留在 10⁻⁶ 量级(PMID 22394158)——远够不到全基因组显著线(5×10⁻⁸),这意味着经过 LD 修剪后能站出来的独立 IV 很可能只有个位数甚至为零。再看中医体质这边:阳虚体质小样本 GWAS(30 vs 30 人,PMID 25533513)找到 42 个 p<10⁻⁴ 的 SNP,血瘀(瘀滞)体质在 947 名 2 型糖尿病患者里找到 22 个 SNP 中 PON2 最显著(PMID 26169365),痰湿体质则是基因表达谱层面的差异(PMID 23548117)加全基因组甲基化特征(PMID 29518789)。这些研究证明”体质有遗传基础”,但离”能当 MR 工具变量”还差着数量级:p 值不够全基因组显著、样本量千人级、效应量小——三样叠在一起,IV 强度审计的第一张表(F 统计量)大概率就过不了。

    好消息是”可复制性”已经出现曙光。 韩国 2015 年的 FTO 研究(PMID 25888059)在 5490 人的发现+验证双队列里,确认了肥胖风险变异 rs7193144 与少阴型反向关联(OR=0.729,p=1.47×10⁻⁷),且独立于 BMI——这是四象体质第一个真正跨队列复制的位点,说明”体质有稳健遗传位点”并非空想,只是密度太低。痰湿体质的 LDL-R 基因多态性研究(PMID 17898945)也提示 PvuII/AvaII 位点可作为遗传标记。把这些拼起来,体质 MR 的强度审计结论很明确:九种体质要跑双样本 MR,先把痰湿、阳虚、阴虚三个证据最厚的体质做到万人级 GWAS,拿到一批全基因组显著、LD 修剪后互相独立的位点,再按 STROBE-MR 清单走完 F 统计量、MR-PRESSO、MR-Egger 全套体检——这跟韩国四象体质的路径一致,也是一切”体质→疾病”因果结论的入场券。

    证据与局限

    已验证: 弱工具偏倚的严重性随 F 连续变化、F>10 非安全线的结论有直接方法学证据(PMID 21414999, 20813862);水平多效性在 MR 中普遍存在且能制造假阳性(PMID 29686387);STROBE-MR 报告规范已确立(PMID 34698778);四象体质有跨队列复制的稳健位点(FTO rs7193144,PMID 25888059)。

    部分验证: 九种体质各有候选位点但 p 值多为 10⁻⁴ 量级、样本千人级,未达全基因组显著线(PMID 25533513, 26169365, 22394158);痰湿体质有基因表达与甲基化特征但非 GWAS 位点强度(PMID 23548117, 29518789)。

    待验证: ①九种体质万人级 GWAS 能否产生一批通过 F>10(及更严格准则)与 MR-PRESSO 多效性检验的独立 IV;②痰湿体质 LDL-R 多态(PMID 17898945)能否在独立队列复制;③四象体质的 FTO 等位点能否在中医九种体质(尤其痰湿质)上复现同一效应方向。

    相关阅读

    证据等级: B
    参考文献: PMID 21414999, PMID 20813862, PMID 29686387, PMID 34698778, PMID 25888059, PMID 25533513, PMID 26169365, PMID 22394158, PMID 23548117, PMID 29518789, PMID 17898945


    免责说明:本文属传统知识与科学文献的整理、对照与研究设计讨论,仅供理解与参考,不构成医疗建议,不能替代专业诊断与治疗。文中涉及的效应量、频率区间与暴露结论多为研究报告层面的推断,个体情况差异很大;如有身体不适或健康疑虑,请就医并由专业医师判断。

    Audit of Instrument Strength for Phenotypic Tools: Nine Phenotypes Must Pass Three Mendelian Randomization Gates

    Evidence level: B

    Mendelian randomization (MR) intends to use “constitution” as exposure only if it possesses a robust, pure, and independent instrumental variable (IV); yet the widely circulated threshold of “F > 10” is precisely the most deceptive hurdle.

    Main Text

    The previous article presented an appealing prospect: treating TCM constitution as “exposure,” using genetic loci as instrumental variables (IVs), and inferring causality (“constitution → disease”) via Mendelian randomization (MR). Koreans have already achieved this with the Four Temperaments constitution—Taiyin type showed a significant genetic tendency increasing risk of cardioembolic stroke ($\beta=4.190$), while Shaoyin type significantly reduced it ($\beta=-9.336$). However, a problem often overlooked is whether the “knife” (the IV set) itself is sharp enough. If the IVs are dull, contaminated, or entangled, the causal conclusions derived from MR may be illusions. This article performs an “intensity audit”: for nine constitutions to truly pass MR, they must sequentially clear three gates.

    Gate 1: The F-statistic, but do not迷信 “F > 10”. The first-stage F-statistic in two-stage least squares (2SLS) measures the explanatory power of genetic loci on the exposure (constitution)—the larger the F, the stronger the instrument. The most prevalent rule states “F > 10 ensures safety,” but a foundational methodological review from 2011 (PMID 21414999) shattered this rule using simulations and real data: the severity of weak instrument bias varies continuously with the expected value of F; F > 10 is an empirical heuristic rather than a safety line. More troublingly, selecting instruments or genetic models post-hoc based on observed F values artificially inflates bias, and the correlation between “higher F and smaller standard errors” introduces a second layer of bias into MR meta-analyses. Another study from the same year (PMID 20813862) provided a more precise accounting: to keep relative bias in 2SLS within 10%, F must reach at least 9–10 for single-instrument scenarios; furthermore, the magnitude of actual bias depends not only on F but also on the strength of exposure-outcome confounding—stronger confounding yields larger bias for the same F. In other words, F > 10 is not a pass mark, but rather the starting point for “at least conducting sensitivity analyses.”

    Gate 2: Pleiotropy Testing, Horizontal Pleiotropy is the Largest Contaminant. A SNP affecting both constitution and outcome directly (not via constitution) constitutes horizontal pleiotropy, violating MR’s exclusion restriction assumption. A 2018 Nature Genetics study on MR-PRESSO (PMID 29686387) performed 4,250 causal tests across 82 complex traits and found that over 48% of significant causal associations exhibited detectable horizontal pleiotropy. Pleiotropy distorted causal estimates by an average of −131% to +201% and generated false positives in up to 10% of relationships. Crucially, it alerted constitution MR: MR-PRESSO performs best only when the proportion of pleiotropic instruments is <50%. Thus, if the loci for nine constitutions contain a large number of one-causes-many-effects SNPs, even correction tools may fail. The 2021 STROBE-MR reporting guideline (PMID 34698778) codified these requirements into standard procedures: explicitly state the three IV assumptions, report details on instrument selection and LD pruning, report F and Q statistics, and execute sensitivity analyses using MR-Egger, weighted median, and leave-one-out methods. Taken together, these items form a complete “intensity audit checklist.”

    Gate 3: Number of Independent IVs After LD Pruning. This is the most realistic pain point for the nine constitutions. The Korean Four Temperaments GWAS scanned 340,000 loci, but significant associations remained at the $10^{-6}$ level (PMID 22394158)—far below the genome-wide significance threshold ($5 \times 10^{-8}$). This implies that independent IVs surviving LD pruning may number only in the single digits or even zero. Looking at TCM constitution research: the Yangxu constitution small-sample GWAS (30 vs 30 individuals, PMID 25533513) identified 42 SNPs with $p < 10^{-4}$; the Blood Stasis (Yu Zhi) constitution found 22 SNPs in 947 type 2 diabetics with PON2 being most significant (PMID 26169365); and the Phlegm-Damp constitution showed differences at the gene expression profile level (PMID 23548117) plus whole-genome methylation features (PMID 29518789). These studies prove “constitution has a genetic basis,” but they are orders of magnitude away from being suitable as MR IVs: p-values fail genome-wide significance, sample sizes are in the thousands rather than tens of thousands, and effect sizes are small. Combined, these factors mean the first table of the IV strength audit (F-statistic) will likely fail for most constitutions.

    Good news: signs of “reproducibility” are emerging. The Korean 2015 FTO study (PMID 25888059) confirmed in a discovery-validation dual cohort of 5,490 individuals that the obesity risk variant rs7193144 showed a reverse association with Shaoyin type (OR=0.729, $p=1.47 \times 10^{-7}$), independent of BMI. This was the first truly cross-cohort replicated locus for the Four Temperaments constitution, indicating that “robust genetic loci for constitution” are not mere fantasies, albeit with low density. The study on LDL-R gene polymorphism for Phlegm-Damp constitution (PMID 17898945) also suggested the PvuII/AvaII sites could serve as genetic markers. Putting these together, the conclusion of the constitution MR intensity audit is clear: for the nine constitutions to run two-sample MR, first achieve genome-wide significant, LD-pruned independent loci for the three constitutions with the strongest evidence (Phlegm-Damp, Yangxu, Yinxu) via million-person-scale GWAS. Then proceed through the full STROBE-MR checklist covering F-statistics, MR-PRESSO, and MR-Egger. This aligns with the path taken by the Korean Four Temperaments constitution and serves as the ticket for all “constitution → disease” causal conclusions.

    Evidence and Limitations

    Verified: Conclusions that weak instrument bias severity varies continuously with F and that F > 10 is not a safety line are supported by direct methodological evidence (PMID 21414999, 20813862); horizontal pleiotropy is prevalent in MR and can generate false positives (PMID 29686387); the STROBE-MR reporting guideline is established (PMID 34698778); and Four Temperaments constitution has robust loci with cross-cohort replication (FTO rs7193144, PMID 25888059).

    Partially Verified: Nine constitutions have candidate loci but p-values are mostly at the $10^{-4}$ level with sample sizes in the thousands, failing to reach genome-wide significance (PMID 25533513, 26169365, 22394158); Phlegm-Damp constitution has gene expression and methylation features but lacks GWAS locus intensity (PMID 23548117, 29518789).

    To Be Verified: ① Whether million-person-scale GWAS for nine constitutions can generate a batch of independent IVs passing F > 10 (and stricter criteria) and MR-PRESSO pleiotropy tests; ② Whether the Phlegm-Damp LDL-R polymorphism (PMID 17898945) can be replicated in an independent cohort; ③ Whether the FTO allele from Four Temperaments constitution can be reproduced in TCM Nine Constitutions (especially Phlegm-Damp) with the same effect direction.

    Related Reading

    Evidence Level: B
    References: PMID 21414999, PMID 20813862, PMID 29686387, PMID 34698778, PMID 25888059, PMID 25533513, PMID 26169365, PMID 22394158, PMID 23548117, PMID 29518789, PMID 17898945


    Disclaimer: This article compiles traditional knowledge and scientific literature for comparison and research-design discussion only. It is not medical advice and cannot replace professional diagnosis or treatment. Effect sizes, frequency ranges and exposure conclusions cited here are largely inferences at the research-report level, and individual circumstances vary widely. If you have any discomfort or health concerns, please consult a physician.