Tag: 中医体质

  • 体质作暴露:孟德尔随机化如何把”体质→疾病”从相关变成因果

    Evidence level: A

    体质研究最大的软肋是”只能证明相关、不能证明因果”——而孟德尔随机化给出了一条出路:把体质当成”暴露”,用遗传位点作工具变量,借”基因彩票”的自然随机分配来推断因果。这条路,韩国已经走通了第一步。

    正文

    中医体质学说有个反复被质疑的点:痰湿体质和代谢病、阳虚体质和抑郁症之间”有关联”,但这个关联到底是”体质导致疾病”,还是”疾病让人显得像某种体质”?过去所有的横断面研究都答不了这个问题——因为观察性研究永远无法排除反向因果和混杂因素。体质研究真正缺的不是数据,而是一把能切进因果关系的刀。

    这把刀就是孟德尔随机化(MR)。 它的原理听起来像魔术,其实是个天然实验:人的基因变异在受精那一刻随机分配,不受后天生活习惯、社会经济地位、甚至疾病本身的影响。如果你能用”决定体质的遗传位点”当工具变量(IV),那么”基因→体质”这一段是确定无疑的,于是顺着”基因→体质→疾病”这条链推断因果,就绕开了所有后天混杂。这就是为什么 #194 的多中心 GWAS 那么关键——体质必须有可重复的遗传位点,MR 才有工具可用;而 #195 的 PRS 证明了”先天禀赋”可以量化,MR 则是把这种量化推进到因果层面的下一步。

    一个合格的体质工具变量必须同时满足三个条件:相关性(位点确实关联体质)、独立性(位点不受”体质→疾病”通路之外的混杂影响)、排他性(位点只能通过体质影响疾病,不能直接作用于结局)。第三条最难,因为基因常有”一因多效”——所以现代 MR 的标配是一整套敏感性分析:MR-Egger 检验水平多效性、加权中位数对异常 SNP 稳健、留一法确认结果不是被单个位点带偏。换句话说,体质 MR 不是”跑一个回归就完事”,而是一场围绕因果推断的证伪演练,每一个假设都要有对应的检验来保护。

    韩国已经做出了直接示范。 2024 年,韩国韩医学研究院(KIOM)发表了一项真正意义上的”体质作暴露”研究(PMID 38927518):他们用四象体质(太阴/少阴/少阳/太阳)相关的遗传位点作工具变量,对卒中的四个亚型做孟德尔随机化。结果很有冲击力——太阴型(TE,代谢病风险最高的类型)的遗传倾向显著增加心源性栓塞性卒中(CES)风险(β=4.190,p=0.035);而少阴型(SE)的遗传倾向则显著降低 CES 风险(β=−9.336,p=3.87×10⁻⁵)。两个体质类型对同一个疾病结局给出了方向相反的因果效应,而且效应方向与四象体质已知的代谢特征完全吻合。这已经不是”体质和病有关系”,而是”你是什么体质,遗传上就在多大程度上推向某种病”——因果推断意义上的结论。

    中介链证据也在同步补上。 台湾生物库 11,030 人的研究(PMID 39257026)发现,抑郁症的多基因风险对抑郁发病的效应,有 10.21% 经由阳虚体质传导、8.41% 经由血瘀体质传导——”基因→体质→疾病”的中介路径首次在中医体质上被量化。而两步 MR 中介分析的技术模板也已经成熟:2024 年两项研究分别证明 LDL-C 介导 GLP1 受体激动剂对心衰的保护效应 4.23%(PMID 39198854)、炎症因子 LIFR 介导脂质-心衰通路 23.5%–25.2%(PMID 39145416)。把这些拼起来,九种体质做”体质→中间生物标志物→疾病”两步 MR 的全部工具都已就位。

    剩下的是三道硬门槛。 第一,工具变量强度:体质 GWAS 目前样本量小(韩国 1222 人,PMID 22394158),能通过 F>10 检验的稳健 IV 数量不足,这是最现实的瓶颈;第二,水平多效性:一个 SNP 可能同时影响多个表型,必须用 MR-Egger、加权中位数、MR-PRESSO 做敏感性分析排除;第三,体质定义的一致性:韩国四象与中医九种体质并不等同,跨体系复现是必答题。这三道门槛,恰好就是”体质作暴露”从韩国示范走向中医九种体质的施工图。

    值得注意的还有一条正在打开的新路:2023 年《中医杂志》的系统评述(DOI 10.13288/j.11-2166/r.2023.05.002)明确提出,MR 在中医领域最大的应用空间不是验证某个方药,而是”简化证型、理清不同病证间的关系,实现异病同治”——也就是把体质/证候本身当作可因果推断的暴露变量来研究。这说明”体质作暴露”不是一个孤立的脑洞,而是整个中医现代化因果研究的方向共识。对九种体质而言,最现实的第一步,是回到 #194 提到的多中心 GWAS:先把痰湿、阳虚、阴虚这三个证据最厚的体质做到万人级样本,拿到一批稳健位点,然后跑第一轮”体质→代谢指标→疾病”的两步 MR。韩国用四象体质证明了这条路走得通,中医九种体质能不能复现,就是下一步最值得盯的问题。

    证据与局限

    已验证: 以体质相关 SNP 作工具变量的 MR 设计已被真实执行(四象体质→卒中亚型,PMID 38927518);体质作为”基因→疾病”中介节点的因果证据存在(阳虚 10.21%、血瘀 8.41%,PMID 39257026);两步 MR 中介方法学成熟(PMID 39198854, 39145416);体质遗传位点已有多项 GWAS 基础(PMID 19954339, 22394158, 22889377)。

    部分验证: 中医九种体质(区别于韩国四象体质)尚无直接以体质为暴露的双样本 MR 研究;现有体质 GWAS 样本量小(千人级),IV 强度不足是普遍问题。

    待验证: ①九种体质 GWAS 位点能否构建通过 F>10 与多效性检验的稳健 IV 集;②”体质→代谢标志物→疾病”两步 MR 的中介比例与稳定性;③四象体质 MR 结果能否在中医九种体质上跨体系复现。

    相关阅读

    证据等级: A
    参考文献: PMID 38927518, PMID 39257026, PMID 39198854, PMID 39145416, PMID 22394158, PMID 19954339, PMID 22889377


    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。

    Phenotype as Exposure: How Mendelian Randomization Transforms the Association from \”Constitution to Disease\” into Causality

    Evidence level: A

    The greatest vulnerability in constitution research is the inability to prove causality despite demonstrating association—yet Mendelian Randomization (MR) offers a pathway: treating constitution as “exposure” and using genetic loci as instrumental variables (IVs) to infer causality through the natural random allocation of a “genetic lottery.” South Korea has already navigated this path successfully in its initial phase.

    Main Text

    A recurring point of scrutiny regarding Traditional Chinese Medicine (TCM) constitution theory is the observed “association” between phlegm-dampness constitution and metabolic disorders, or yang-deficiency constitution and depression. However, does the constitution cause the disease, or does the disease manifest as if the individual possesses a specific constitution? Cross-sectional studies have historically failed to answer this question because observational designs cannot rule out reverse causality or confounding factors. The true deficit in constitution research is not data, but a scalpel capable of cutting into causal inference.

    This scalpel is Mendelian Randomization (MR). Its mechanism may appear symbolic, but it functions as a natural experiment: genetic variants are randomly allocated at fertilization, unaffected by lifestyle habits, socioeconomic status, or the disease itself. If one uses “genetic loci determining constitution” as an IV, the link from “gene → constitution” is deterministic; thus, tracing the chain “gene → constitution → disease” infers causality while bypassing all postnatal confounders. This explains why the multi-center GWAS in #194 is critical—constitution must possess reproducible genetic loci for MR to have an instrument available; meanwhile, #195 demonstrated that “innate endowment” can be quantified via Polygenic Risk Scores (PRS), and MR represents the next step advancing this quantification to the causal level.

    A valid constitution IV must satisfy three conditions: relevance (the locus genuinely associates with constitution), independence (the locus is unaffected by confounders outside the “constitution → disease” pathway), and exclusion restriction (the locus affects the outcome only through constitution, not directly). The third condition is most difficult due to “pleiotropy” (one gene affecting multiple traits); therefore, modern MR standards include a suite of sensitivity analyses: MR-Egger tests for horizontal pleiotropy, weighted median robustness against outlier SNPs, and leave-one-out methods to confirm results are not driven by a single locus. In short, constitution MR is not merely “running one regression,” but a falsification exercise centered on causal inference where every hypothesis requires corresponding tests for protection.

    South Korea has provided a direct demonstration. In 2024, the Korean Institute of Oriental Medicine (KIOM) published a genuine “constitution as exposure” study (PMID 38927518): they used genetic loci associated with the Four Temperaments (Taiyin/Shaoyin/Shaoyang/Taiyang) as IVs to perform MR on four subtypes of stroke. The results were impactful—the genetic propensity for the Taiyin type (associated with the highest metabolic disease risk) significantly increased the risk of cardioembolic stroke (CES) ($\beta=4.190$, $p=0.035$); conversely, the genetic propensity for the Shaoyin type (SE) significantly decreased CES risk ($\beta=-9.336$, $p=3.87\times10^{-5}$). Two constitution types yielded causally opposite effects on the same disease outcome, with effect directions fully consistent with known metabolic characteristics of the Four Temperaments. This transcends mere “association between constitution and disease”; it establishes that “what constitution you are genetically predisposes you to a specific disease”—a conclusion in the sense of causal inference.

    Mediating chain evidence is also being filled simultaneously. A study of 11,030 individuals from the Taiwan BioBank (PMID 39257026) found that 10.21% of the effect of depression’s polygenic risk on disease onset is mediated via yang-deficiency constitution, and 8.41% via blood stasis constitution—quantifying the “gene → constitution → disease” mediation pathway for the first time in TCM constitution. Furthermore, technical templates for two-step MR mediation analysis are now mature: two studies in 2024 proved that LDL-C mediates 4.23% of the protective effect of GLP1 receptor agonists on heart failure (PMID 39198854) and that the inflammatory factor LIFR mediates 23.5%–25.2% of the lipid-heart failure pathway (PMID 39145416). Combining these, all tools for conducting “constitution → intermediate biomarker → disease” two-step MR across nine constitutions are now in place.

    Remaining are three hard thresholds. First, IV strength: current constitution GWAS sample sizes are small (Korea: 1,222 individuals; PMID 22394158), resulting in insufficient robust IVs passing the $F>10$ test, representing the most realistic bottleneck. Second, horizontal pleiotropy: a single SNP may influence multiple phenotypes, requiring sensitivity analyses via MR-Egger, weighted median, and MR-PRESSO to exclude such effects. Third, consistency in constitution definition: Korea’s Four Temperaments are not equivalent to the nine TCM constitutions; cross-system replication is a mandatory question. These three thresholds precisely map the blueprint for moving from Korea’s demonstration to the nine TCM constitutions as “exposure.”

    Notably, a new path is opening: a 2023 systematic review in the Journal of Traditional Chinese Medicine (DOI 10.13288/j.11-2166/r.2023.05.002) explicitly states that MR’s largest application space in TCM is not validating specific formulas, but “simplifying syndromic patterns, clarifying relationships between different disease-syndromes, and achieving ‘same treatment for different diseases’”—treating constitution/syndrome itself as a causally inferable exposure variable. This indicates that “constitution as exposure” is not an isolated idea but a consensus direction for modernizing TCM causal research. For the nine constitutions, the most practical first step is returning to the multi-center GWAS mentioned in #194: first, build a population-level sample (ten thousand+) for the three constitutions with the strongest evidence (phlegm-dampness, yang-deficiency, yin-deficiency) to obtain a set of robust loci, then run the first round of two-step MR for “constitution → metabolic marker → disease.” Korea has proven this path is viable; whether the nine TCM constitutions can replicate these results is the most critical question to watch next.

    Evidence and Limitations

    Verified: MR designs using constitution-associated SNPs as IVs have been executed in reality (Four Temperaments → Stroke subtypes, PMID 38927518); causal evidence exists for constitution as a “gene → disease” mediator node (yang-deficiency 10.21%, blood stasis 8.41%, PMID 39257026); two-step MR mediation methodology is mature (PMID 39198854, 39145416); multiple GWAS foundations exist for constitution genetic loci (PMID 19954339, 22394158, 22889377).

    Partially Verified: There are no direct two-sample MR studies using the nine TCM constitutions (distinct from Korea’s Four Temperaments) as exposure; existing constitution GWAS sample sizes are small (thousands), with insufficient IV strength being a common issue.

    To Be Verified: ① Can GWAS loci for the nine constitutions construct a robust IV set passing $F>10$ and pleiotropy tests? ② What are the mediation proportions and stability of “constitution → metabolic marker → disease” two-step MR? ③ Can MR results from the Four Temperaments be replicated across the nine TCM constitutions in a cross-system manner?

    Related Reading

    Evidence Level: A
    References: PMID 38927518, PMID 39257026, PMID 39198854, PMID 39145416, PMID 22394158, PMID 19954339, PMID 22889377


    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.

  • 先天禀赋的可改变幅度有多大?体质PRS与生活方式干预的随机对照试验

    Evidence level: A

    中医说”体质禀赋于先天”,那么先天的遗传底色,到底能在多大程度上被后天的生活方式干预改写?随机对照试验给出了一个比”宿命论”和”万能论”都更精确的答案。

    正文

    “江山易改,禀性难移”——这句话在中医语境里,指向的是体质中那个似乎改不动的部分:先天禀赋。阳虚的人怕冷,痰湿的人易胖,血瘀的人易堵,中医把这部分归因于”先天之精”的底子。现代遗传学给了这个古老的直觉一个可量化的名字:多基因风险评分(Polygenic Risk Score, PRS)——把一个人身上成千上万个与某表型相关的基因位点按效应加权加总,得到一个代表”遗传易感性”的数字。那么问题来了:这个数字是”判决书”吗?生活方式干预能把它改写到什么程度?这恰恰是”先天禀赋可修正幅度”这个边界问题的核心。

    第一个关键结论来自糖尿病预防计划(DPP)系列研究——一个随访数千人、随机分到”强化生活方式 / 二甲双胍 / 安慰剂”三组的里程碑式 RCT。DPP 的遗传子研究反复追问同一个问题:携带高风险基因的人,干预效果是不是更差?答案出人意料地”否”。一项用 17 个胰岛素抵抗位点构建遗传风险评分(GRS)的研究发现,高 GRS 者基线胰岛素敏感性确实更低(β=−0.754/单位,P=0.001),但在生活方式和二甲双胍干预下,各组胰岛素敏感性的改善幅度与遗传负担无关——治疗×GRS 的交互项不显著(P=0.54),也就是说,遗传负担高的人,通过生活方式干预同样能获得胰岛素敏感性的改善(PMID 26525880)。这是”先天禀赋不是宿命”最直接的一条 RCT 证据。

    但”修正幅度”确实存在边界,边界恰恰藏在交互作用里。DPP 的另一项研究用 32 个脂质相关位点构建 GRS,考察其对干预后血脂的影响,结果出现了耐人寻味的”反应衰减”:在生活方式干预组,高 GRS 者 1 年后的 LDL-C 仍显著偏高(β=+0.87 mg/dl/allele,交互 P=0.02),小而密 LDL 颗粒数也更多(交互 P=0.01)——即遗传负荷越高,生活方式干预对脂质的改善效果越打折扣(PMID 22951888)。换言之:遗传不是”改不了”,而是”改的斜率”可能因人而异。这正是中医”体质调理””同病异治”的现代翻版——同样一碗补药、一套功法,落在不同遗传底子上,边际收益并不相同。

    第三条证据指向一个更反直觉的边界:光”知道”没用。 2026 年刚发表的一项 1020 人双盲 RCT 测试了”把高冠心病 PRS 告知年轻超重成人”能否驱动行为改变,结果 12 个月后 BMI、血脂、血压、药物依从性均无改善(PMID 41513891)。单纯的风险告知,不足以产生行为改变。这提示:PRS 的价值不在”宣判”,而在”分层”——它必须先转化为差异化的干预处方,才可能兑现为健康收益。

    中医自己的 RCT 也站在同一边。 一项纳入 5185 人筛出 60 名阳虚体质健康者的随机对照研究,用中药(桂附地黄丸)+生活方式干预一个月,阳虚症状评分显著下降,肠道菌群多样性与代谢组部分恢复至平和质水平,且疗效在一年随访中保持(PMID 34107861)。另一项针对吸毒戒断青年男性、按中医辨证(阳虚/阴虚/气虚/血虚)指导膳食的 RCT 同样显示,辨证饮食能显著改善相应证候要素评分(PMID 31871480)。中医”辨体论治”与”PRS 分层干预”在方法论上是同构的——都是先测出个体的”底”,再据此开不同的”方”。

    把这四条证据链拼起来,”先天禀赋可修正幅度”的答案就清晰了:遗传底色普遍可以被生活方式干预改善(DPP 铁证),但边际收益随遗传负荷与目标表型而异(脂质衰减),而且干预必须落到”处方化”而不是”告知”(PRS 沟通 RCT 阴性)。 中医体质学说的”先天禀赋”,在现代语言里就是一条条带斜率的修正曲线——不是宿命,但也不是白纸;它决定的是”改起来快慢”,而不是”能不能改”。

    证据与局限

    已验证: 高胰岛素抵抗遗传负担者在生活方式/二甲双胍干预下,胰岛素敏感性改善幅度与遗传负担无关(PMID 26525880)——遗传高风险不等于干预无效,这是 A 级 RCT 直接支撑。

    部分验证: 脂质相关遗传负荷越高,生活方式干预对 LDL-C 的改善越衰减(PMID 22951888),说明”修正幅度”存在表型特异的边界;中医阳虚体质中药+生活方式干预 RCT 有效(PMID 34107861)。

    待验证: 中国人群”九种体质 PRS × 生活方式干预”的剂量-反应 RCT(高/低 PRS 分层设计)尚未见报道;辨体 vs 一刀切干预的增量价值缺乏头对头随机证据;体质 PRS 基线水平对干预应答的效应修饰仍需跨中心验证。

    相关阅读

    证据等级: A
    参考文献: PMID 26525880, PMID 22951888, PMID 34107861, PMID 41513891, PMID 31636172


    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。

    How Far Can Innate Endowment Be Shifted? Randomized Controlled Trials of Constitution PRS and Lifestyle Interventions

    Evidence level: A

    Traditional Chinese Medicine (TCM) posits that “constitution is innate.” To what extent can this genetic baseline be rewritten by postnatal lifestyle interventions? Randomized controlled trials (RCTs) provide an answer more precise than both “fatalism” and “universalism.”

    Main Text

    “江山易改,禀性难移” (The landscape changes easily, but innate nature is hard to alter) — in the context of TCM, this points to the seemingly unchangeable component within constitution: innate constitution. Individuals with Yang deficiency fear cold; those with phlegm-dampness tend to gain weight; those with blood stasis are prone to blockages. TCM attributes this portion to the foundation of “Essence of the Heavens and Earth” (先天之精). Modern genetics has given this ancient intuition a quantifiable name: Polygenic Risk Score (PRS) — summing weighted effects across thousands of genetic loci associated with a specific phenotype to generate a number representing “genetic susceptibility.” The question arises: Is this number a “verdict”? To what extent can lifestyle interventions rewrite it? This is precisely the core of the boundary problem regarding “the modifiability of innate constitution.”

    The first key conclusion comes from the Diabetes Prevention Program (DPP) series of studies — a landmark RCT following thousands of participants randomly assigned to three groups: “intensive lifestyle / metformin / placebo.” The genetic sub-studies of DPP repeatedly asked the same question: Do individuals carrying high-risk genes show worse intervention outcomes? The answer is surprisingly “no.” A study constructing a genetic risk score (GRS) using 17 insulin resistance loci found that while high GRS individuals indeed had lower baseline insulin sensitivity ($\beta=-0.754$ per unit, $P=0.001$), the magnitude of improvement in insulin sensitivity across all groups was unrelated to genetic burden under lifestyle and metformin interventions — the interaction term (Treatment $\times$ GRS) was not significant ($P=0.54$). In other words, individuals with high genetic burden can still achieve improvements in insulin sensitivity through lifestyle interventions (PMID 26525880). This is the most direct RCT evidence supporting “innate constitution is not fate.”

    However, a “modification ceiling” does exist, hidden precisely within interaction effects. Another DPP study constructed a GRS using 32 lipid-related loci to examine its impact on intervention-induced lipid changes, revealing a perplexing “response attenuation”: in the lifestyle intervention group, high GRS individuals still had significantly elevated LDL-C one year later ($\beta=+0.87$ mg/dl/allele, interaction $P=0.02$) and higher numbers of small dense LDL particles (interaction $P=0.01$) — meaning higher genetic load results in a discounting of lifestyle intervention efficacy for lipids (PMID 22951888). In short: genetics is not “unchangeable,” but the “slope of modification” may vary by individual. This mirrors the TCM concepts of “constitution regulation” and “treating different diseases differently with the same condition” — the same tonic or set of exercises yields different marginal returns when applied to different genetic baselines.

    A third line of evidence points to a counter-intuitive boundary: mere “knowledge” is insufficient. A 2026 double-blind RCT involving 1,020 participants tested whether informing young overweight adults of their high coronary heart disease PRS could drive behavioral change. After 12 months, BMI, lipids, blood pressure, and medication adherence showed no improvement (PMID 41513891). Simple risk notification is insufficient to produce behavioral change. This suggests: The value of PRS lies not in “sentencing,” but in “stratification” — it must first be translated into differentiated intervention prescriptions before health benefits can be realized.

    RCTs within TCM itself stand on the same side. A randomized controlled study screening 5,185 individuals for 60 healthy subjects with Yang deficiency constitution found that a one-month intervention combining Chinese herbal medicine (Gui Fu Di Huang Wan) and lifestyle changes significantly reduced Yang deficiency symptom scores, restored gut microbiome diversity and parts of the metabolome to healthy constitution levels, and maintained efficacy over a one-year follow-up (PMID 34107861). Another RCT targeting young male drug addicts, guiding diets according to TCM syndromic differentiation (Yang/Yin/Qi/Blood deficiency), similarly showed that syndromic dieting significantly improved corresponding syndrome element scores (PMID 31871480). TCM’s “syndrome-based treatment” and “PRS stratified intervention” are methodologically isomorphic — both involve first measuring the individual’s “baseline,” then prescribing a different “formula” accordingly.

    Assembling these four evidence chains clarifies the answer to “the modifiability of innate constitution”: The genetic baseline is generally improvable via lifestyle interventions (DPP ironclad evidence), but marginal returns vary with genetic load and target phenotype (lipid attenuation), and interventions must be “prescription-based” rather than merely “informative” (PRS communication RCT negative results). In modern language, TCM’s theory of “innate constitution” is a series of correction curves with slopes — not fate, nor a blank slate; it determines the “speed of modification,” not whether modification is possible.

    Evidence and Limitations

    Verified: Individuals with high genetic burden for insulin resistance show no difference in the magnitude of insulin sensitivity improvement under lifestyle/metformin intervention (PMID 26525880) — high genetic risk does not equal ineffective intervention, directly supported by Level A RCT.

    Partially Verified: Higher lipid-related genetic load leads to greater attenuation of lifestyle intervention efficacy for LDL-C (PMID 22951888), indicating a phenotype-specific boundary for “modification ceiling”; TCM Yang deficiency constitution herbal medicine + lifestyle intervention RCT was effective (PMID 34107861).

    To be Verified: No reported dose-response RCT for “Nine Constitutions PRS $\times$ Lifestyle Intervention” in Chinese populations (high/low PRS stratified design); lack of head-to-head randomized evidence on the incremental value of syndrome-based vs. one-size-fits-all interventions; need for multi-center validation on effect modification by baseline constitution PRS levels on intervention response.

    Related Reading

    Evidence Level: A
    References: PMID 26525880, PMID 22951888, PMID 34107861, PMID 41513891, PMID 31636172


    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.

  • 体质调理能改菌群,菌群干预能改体质吗

    Evidence level: A

    中医体质不是”天生定死”的标签,而是可被饮食、中药、益生菌与粪菌移植(FMT)推动的动态表型——问题只在于:按体质分型选干预,到底比”一刀切”好多少?

    正文

    上一篇文章我们论证了体质与肠道菌群互为因果、构成双向正反馈循环。一个自然的推论随之而来:既然循环是双向的,那么从”菌群”这一端推一把,是不是也能改变体质、进而改变疾病风险?这正是中医”调理偏颇体质”的古老直觉——只是过去没有仪器能回答”调理到底改变了什么”。

    第一块拼图:体质评分确实可被干预推动

    2021 年南方医科大学团队的一项随机临床试验给出了教科书级的证据。研究从 5185 人的前瞻队列中筛出 60 名健康的阳虚质大学生,随机分组后干预组接受桂附地黄丸加生活方式调理一个月。结果不仅阳虚症状积分显著下降,且干预后肠道菌群多样性与结构明显恢复——原本阳虚质偏高的 Bacteroidetes 与 Bacteroides 丰度回落,菌群与尿代谢组”部分恢复到平和质样水平”,一年随访疗效仍保持(PMID 34107861)。这条证据的含金量在于:干预同时改动了体质量表评分和菌群指纹,两者方向一致。

    2025 年北京中医药大学深圳医院的 RCT 补齐了另一块:78 名受试者(气虚质 63 例、平和质 15 例)接受气虚调体方干预后,气虚质得分显著下降,血清 IgA、IgG 上升,转录组锁定 8 个免疫相关治疗靶基因,分子对接确认木犀草素、槲皮素等成分可稳定结合靶点(PMID 41246353)。也就是说,”调理改变体质”不再只是主观量表的事,它有免疫指标和转录组的客观依据。

    第二块拼图:菌群干预确实可改变代谢结局

    反过来,从菌群端直接下手也有硬数据。2021 年《自然·医学》发表的重度肥胖合并代谢综合征 2 期双盲 RCT 显示,口服粪菌移植联合低发酵纤维,6 周即显著改善胰岛素敏感性(HOMA2-IR 3.77→3.16),效果独立于纤维本身(PMID 34226737)。更关键的是长期维度:青少年肥胖 FMT 试验的 4 年随访中,FMT 组腰围缩小 10.0 cm、全身脂肪减少 4.8%、代谢综合征严重度评分下降 0.58、炎症标志物 hs-CRP 下降 68%,供体来源菌株在受者肠道持续定植(PMID 40877311)。菌群干预不是”换了个菌群就完事”,而是把代谢轨道整体扳动了一下。

    中医自身也有对号入座的实践:2025 年泽泻汤加减治疗痰湿质 1 期高血压的 30 例临床研究显示,42 天干预后血压下降、菌群多样性上升、SCFA 显著升高,机制指向”肠-免疫-肾轴”(PMID 40606611)。痰湿质是代谢病的高危体质,这个方子恰好从菌群通道起效——中医辨证选方,客观上就是在做”按体质定向的菌群干预”。

    第三块拼图:把两件事接起来

    至此逻辑闭环浮现:体质评分可被干预推动(中药+生活方式),菌群干预可改变代谢结局(FMT/膳食纤维),且同一套干预同时作用于体质量表和菌群指纹(阳虚质 RCT)。体质、菌群、疾病三者构成一个可被外部干预旋动的环。但必须诚实指出缺口:至今没有一项头对头 RCT 直接比较”按体质分型选菌群干预”与”不按体质分型的一刀切干预”,因此”辨证选菌”优于”泛用菌”的增量价值,仍是待证假设而非事实。

    证据与局限

    已验证: 中药+生活方式干预可降低偏颇体质评分并同步恢复菌群多样性(PMID 34107861);体质干预存在免疫学与转录组客观标志(PMID 41246353);FMT 联合纤维可改善胰岛素敏感性(PMID 34226737),长期随访显示代谢与炎症结局改善(PMID 40877311);痰湿质方剂可从肠-免疫-肾轴调菌群降压(PMID 40606611)。

    部分验证: 干预后菌群”恢复到平和质样水平”是近似表述,尚无研究把九种体质的菌群指纹做成金标准对照库;FMT 研究结果存在不一致(部分试验临床指标无差异),效应量受供体、剂量、随访时长调节。

    待验证: 按体质分型选菌群干预 vs 一刀切干预的头对头 RCT;菌群干预改变体质量表评分的剂量-反应曲线;不同体质对同一菌群干预的应答差异(应答者/无应答者分层);干预停止后体质与菌群的回落动力学。

    相关阅读

    证据等级: A
    参考文献: PMID 34107861, PMID 41246353, PMID 34226737, PMID 40877311, PMID 40606611


    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 Microbiome Intervention Modify Physiological Constitution

    Evidence level: A

    Traditional Chinese Medicine (TCM) constitution is not a “fixed-at-birth” label, but a dynamic phenotype that can be modulated by diet, herbal medicine, probiotics, and fecal microbiota transplantation (FMT). The question remains: to what extent does selecting interventions based on constitution typing outperform a “one-size-fits-all” approach?

    Main Text

    In the previous article, we argued that constitution and gut microbiota are causally linked, forming a bidirectional positive feedback loop. A natural inference follows: if the loop is bidirectional, does pushing from the “microbiome” end also alter constitution and subsequently disease risk? This aligns with the ancient intuition of TCM regarding “adjusting imbalanced constitutions”—the only gap was the lack of instrumentation to answer what exactly constitutes an “adjustment.”

    First Piece of the Puzzle: Constitution Scores Can Be Driven by Intervention

    A 2021 randomized controlled trial (RCT) by a team at Southern Medical University provided textbook-level evidence. Screening 60 healthy students with Yang Deficiency constitution from a prospective cohort of 5,185 individuals, the intervention group received Gui Fu Di Huang Wan combined with lifestyle adjustments for one month. Results showed a significant reduction in Yang Deficiency symptom scores, alongside a clear restoration of gut microbiota diversity and structure. Specifically, abundances of Bacteroidetes and Bacteroides, which were typically elevated in Yang Deficiency, decreased, and the microbiome and urinary metabolome partially recovered to levels resembling “Peaceful Constitution” (Ping He Zhi), with efficacy maintained at one-year follow-up (PMID 34107861). The evidentiary value lies in the fact that intervention simultaneously altered constitution score metrics and microbiome fingerprints, moving in the same direction.

    A 2025 RCT by Shenzhen Hospital of Beijing University of Chinese Medicine completed another piece of the puzzle: 78 participants (63 with Qi Deficiency, 15 with Peaceful Constitution) showed a significant decrease in Qi Deficiency scores after intervention with a Qi-tonifying formula. Serum IgA and IgG levels increased, transcriptomics identified 8 immune-related therapeutic target genes, and molecular docking confirmed that compounds such as luteolin and quercetin could stably bind to these targets (PMID 41246353). Thus, “adjustment altering constitution” is no longer merely a matter of subjective scales; it now has objective bases via immune markers and transcriptomics.

    Second Piece of the Puzzle: Microbiome Intervention Can Indeed Alter Metabolic Outcomes

    Conversely, direct intervention from the microbiome end also yields hard data. A 2021 Phase II double-blind RCT published in Nature Medicine regarding severe obesity with metabolic syndrome showed that oral FMT combined with low-fermentable fiber significantly improved insulin sensitivity (HOMA2-IR 3.77→3.16) within six weeks, an effect independent of the fiber itself (PMID 34226737). More critically regarding the long-term dimension, a 4-year follow-up in an adolescent obesity FMT trial showed that the FMT group experienced a waist circumference reduction of 10.0 cm, total body fat reduction of 4.8%, a decrease in metabolic syndrome severity score by 0.58, and a reduction in the inflammatory marker hs-CRP by 68%, with donor-derived strains persistently colonizing the recipient gut (PMID 40877311). Microbiome intervention is not merely “changing the microbiome and finishing the job,” but rather shifting the entire metabolic trajectory.

    TCM itself also has practical applications matching this logic: a 2025 clinical study of 30 cases treating Stage 1 Phlegm-Dampness constitution hypertension with modified Ze Xie Tang showed that after 42 days of intervention, blood pressure decreased, microbiota diversity increased, and short-chain fatty acids (SCFA) significantly rose, with mechanisms pointing to the “gut-immune-kidney axis” (PMID 40606611). Phlegm-Dampness constitution is a high-risk phenotype for metabolic diseases; this formula effectively acts through the microbiome channel—objectively constituting “targeted microbiome intervention based on constitution typing.”

    Third Piece of the Puzzle: Connecting the Two Events

    The logical loop now emerges: constitution scores can be driven by intervention (herbal medicine + lifestyle), microbiome intervention can alter metabolic outcomes (FMT/dietary fiber), and the same interventions act simultaneously on constitution scores and microbiome fingerprints (Yang Deficiency RCT). Constitution, microbiome, and disease form a cycle that can be rotated by external intervention. However, one must honestly acknowledge the gap: no head-to-head RCT has yet directly compared “microbiome intervention selected by constitution typing” versus “one-size-fits-all microbiome intervention.” Therefore, the incremental value of “syndrome differentiation-guided microbial selection” over “generalized microbial use” remains a hypothesis awaiting proof rather than an established fact.

    Evidence and Limitations

    Verified: Herbal medicine + lifestyle interventions can reduce imbalanced constitution scores while simultaneously restoring microbiota diversity (PMID 34107861); constitution interventions exhibit objective immunological and transcriptomic markers (PMID 41246353); FMT combined with fiber can improve insulin sensitivity (PMID 34226737), with long-term follow-ups showing improvements in metabolic and inflammatory outcomes (PMID 40877311); Phlegm-Dampness formulas can modulate microbiota to lower blood pressure via the gut-immune-kidney axis (PMID 40606611).

    Partially Verified: The description of post-intervention microbiota “returning to Peaceful Constitution levels” is an approximation; no study yet has established a gold-standard reference library comparing microbiome fingerprints across the nine constitutions. FMT study results show inconsistencies (some trials showed no difference in clinical indicators), with effect sizes modulated by donor source, dosage, and follow-up duration.

    To Be Verified: Head-to-head RCT comparing constitution-typed microbiome intervention vs. one-size-fits-all intervention; dose-response curves for microbiome intervention effects on constitution score metrics; differential responses to the same microbiome intervention across different constitutions (stratified by responders/non-responders); the kinetics of reversion for both constitution and microbiota after intervention cessation.

    Related Reading

    Evidence Level: A
    References: PMID 34107861, PMID 41246353, PMID 34226737, PMID 40877311, PMID 40606611


    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.

  • 中医九种体质的多中心GWAS重复验证:从初步关联到可复现的表型指纹

    Evidence level: A

    体质分型已有多条遗传学证据线,但要让”九种体质=可遗传表型指纹”成为被现代遗传学接受的结论,必须完成多中心大样本GWAS+跨群体meta这条最后验证链路。

    正文

    问题:体质量表能测出你是”痰湿质”还是”平和质”,但这套分型在基因组层面站得住吗?已有答案是”有初步证据”,但距离”严格验证”还差最后一公里——多中心重复。

    一、体质分型的遗传学证据已经存在

    体质差异的遗传基础并非空谈。2009年,韩国庆熙大学对四象体质医学做了首次全基因组关联研究(GWAS),在353,202个SNP位点上发现四象类型两两比较有15,944个SNP显著差异,其中5个SNP在所有比较中共同显著(PMID 19954339)。同一时期,北京中医药大学王琦团队对痰湿质做基因表达谱分析,发现4个上调基因(COPS8、GNPDA1、CD52、ARPC3)与6个下调基因(GSPT2、CACNB2等),功能指向高脂血症与糖尿病易感性(PMID 23548117)——”痰湿质容易得代谢病”在分子层面有了第一个锚点。

    候选基因研究进一步补上机制细节:痰湿质与LDL受体基因Pvu II/Ava II多态性显著关联,且痰湿质人群甘油三酯、空腹血糖、餐后胰岛素均显著更高(PMID 17898945);高血压人群痰湿质血清脂联素显著更低,脂联素基因rs1063537的T等位基因携带者是其遗传特征(PMID 20681271);韩国卒中人群中NPY基因C-399T的T等位基因与痰湿证负相关(PMID 22110543)——同一”湿浊-代谢”关联在两个族裔中得到交叉印证。更宏观的证据来自双生子研究:四象体质遗传度估算高达47%-81%,共享环境效应几乎为零,说明体质很大程度上是”先天写好的蓝图”。

    二、为什么必须多中心重复验证

    上述证据真实但零散,存在三个结构性弱点:样本量小(几十到几百人,远低于现代GWAS的数万级门槛)、单中心单族裔(四象研究集中在韩国、九种体质研究集中在中国)、方法学不统一(不同量表、不同芯片、不同阈值)。这正是现代遗传学的”多中心GWAS+meta分析”要解决的问题——用统一表型定义(《中医体质分类与判定》量表)、统一质控流程,在多个独立中心采集大样本基因型,再跨群体合并统计量。2024年Yap与Morris的综述系统梳理了多族裔GWAS meta分析(MAGMA)的协议、统计方法与软件(PMID 39022911),指出跨族裔整合能提高检出功效并改善因果变异的精细定位;2026年Dias等人的大规模模拟与UK Biobank(约32.4万人)+All of Us(约20.7万人)实测表明,合并分析在控制群体分层后统计功效更优(PMID 40162269)。这些方法学成熟到可以直接套用——缺的不是工具,而是”九种体质”这个表型被拿去做大样本GWAS这件事本身。

    三、这条路走通意味着什么

    一旦九种体质在多个独立中心的大样本GWAS中复现出稳定位点,并跨族裔合并出体质多基因风险评分(PRS),体质学就能从”经验分类”升级为”可遗传表型指纹”:体质量表变成一张”表型PRS”,用症状和体征做代理,直接指向疾病易感性方向。这与4P医疗(个性化、预测、预防、参与)的框架完全同构——也是体质学真正融入精准医学的门票。反过来,如果多中心GWAS大面积复现失败,那将是对”体质可分”假说的最严厉证伪——这正是科学该有的样子。

    证据与局限

    已验证: 体质类型与特定基因多态性存在可重复关联(四象GWAS的SNP信号、痰湿质表达谱、LDL-R/APN/NPY/HLA多态性);痰湿质-代谢病易感在分子层面可解释;多族裔GWAS meta分析方法学成熟(PMID 39022911, 40162269)。

    部分验证: 九种体质中仅痰湿、阴虚、阳虚等少数类型有较充分遗传证据;现有GWAS多为韩国四象体质样本,中医九种体质尚无达到现代标准的大样本全基因组研究。

    待验证: ①统一体质量表下多中心大样本GWAS能否复现已知位点并发现新位点;②体质PRS能否独立预测疾病结局并跨族裔迁移;③体质-基因关联在不同环境暴露下是否稳定;④”体质可调”是否对应可测量的基因组/表观遗传变化。

    相关阅读

    证据等级: A
    参考文献: PMID 19954339, PMID 23548117, PMID 17898945, PMID 20681271, PMID 22110543, PMID 39022911, PMID 40162269


    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。

    Multi-center GWAS replication of the nine traditional Chinese medicine constitutions: from preliminary associations to reproducible phenotypic fingerprints

    Evidence level: A

    Multiple lines of genetic evidence support constitution classification, yet establishing “Nine Constitutions = Heritable Phenotypic Fingerprints” as a conclusion accepted by modern genetics requires completing the final validation link: multi-center large-sample GWAS combined with cross-population meta-analysis.

    Main Text

    Question: Can body mass scales identify whether you have the “Phlegm-Dampness Constitution” or the “Harmonious Constitution”? Does this classification hold up at the genomic level? The current answer is “there is preliminary evidence,” but it still lacks the final mile of “rigorous validation”—multi-center replication.

    I. Genetic Evidence for Constitution Classification Already Exists

    The genetic basis for constitution differences is not theoretical. In 2009, Kyung Hee University in Korea conducted the first genome-wide association study (GWAS) on four-constitution medicine, identifying 15,944 SNPs with significant differences between pairs of constitution types across 353,202 SNP loci, with 5 SNPs showing joint significance across all comparisons (PMID 19954339). Around the same time, Wang Qi’s team at Beijing University of Chinese Medicine performed gene expression profiling on the Phlegm-Dampness Constitution, identifying 4 upregulated genes (COPS8, GNPDA1, CD52, ARPC3) and 6 downregulated genes (GSPT2, CACNB2, etc.), pointing functionally toward susceptibility to hyperlipidemia and diabetes (PMID 23548117)—providing the first molecular anchor for the notion that “Phlegm-Dampness constitution predisposes to metabolic disorders.”

    Candidate gene studies further elucidate mechanistic details: the Phlegm-Dampness Constitution shows significant association with polymorphisms in the LDL receptor gene (Pvu II/Ava II), and individuals with this constitution exhibit significantly higher triglycerides, fasting glucose, and postprandial insulin (PMID 17898945); in hypertensive populations, serum adiponectin is significantly lower in the Phlegm-Dampness Constitution, with the T allele of adiponectin gene rs1063537 serving as a genetic characteristic (PMID 20681271); in Korean stroke populations, the T allele of the NPY gene C-399T shows a negative correlation with Phlegm-Dampness syndrome (PMID 22110543)—cross-validating the same “dampness-turbidity-metabolism” association across two ethnic groups. Broader evidence comes from twin studies: heritability estimates for four-constitution types range from 47% to 81%, with shared environmental effects nearly zero, indicating that constitution is largely a “pre-written blueprint.”

    II. Why Multi-Center Replication Is Necessary

    The above evidence is real but fragmented, exhibiting three structural weaknesses: small sample sizes (tens to hundreds of individuals, far below the tens-of-thousands threshold for modern GWAS), single-center/single-ethnicity (four-constitution studies concentrated in Korea, nine-constitution studies in China), and methodological heterogeneity (different scales, different chips, different thresholds). This is precisely what modern genetics addresses through “multi-center GWAS + meta-analysis”—using unified phenotypic definitions (e.g., the TCM Constitution Classification and Judgment scale), standardized quality control protocols, collecting large sample genotypes across independent centers, and then merging statistics across populations. A 2024 review by Yap and Morris systematically outlined protocols, statistical methods, and software for multi-ethnic GWAS meta-analysis (MAGMA) (PMID 39022911), noting that cross-ethnic integration improves detection power and refines causal variant localization; a 2026 large-scale simulation by Dias et al., alongside empirical data from UK Biobank (~324,000 individuals) and All of Us (~207,000 individuals), demonstrates that merged analysis yields superior statistical power after controlling for population stratification (PMID 40162269). These methodologies are mature enough for direct application—the missing component is not the tools, but whether the “Nine Constitutions” phenotype itself will be subjected to large-sample GWAS.

    III. What This Pathway Entails If Successful

    If the Nine Constitutions consistently replicate stable loci across multi-center large-sample GWAS and yield a constitution polygenic risk score (PRS) upon cross-ethnic merging, the field of constitution studies can evolve from “empirical classification” to “heritable phenotypic fingerprints”: the body mass scale becomes a “phenotype PRS,” using symptoms and signs as proxies to directly point toward disease susceptibility directions. This aligns perfectly with the 4P medical framework (personalized, predictive, preventive, participatory)—also serving as the ticket for constitution studies to truly integrate into precision medicine. Conversely, if multi-center GWAS fails to replicate broadly, it would constitute the most rigorous falsification of the “constitution categorization” hypothesis—exactly how science should operate.

    Evidence and Limitations

    Verified: Reproducible associations exist between constitution types and specific gene polymorphisms (SNP signals from four-constitution GWAS, expression profiles for Phlegm-Dampness, and polymorphisms in LDL-R/APN/NPY/HLA); the Phlegm-Dampness-metabolic disorder susceptibility is explainable at the molecular level; multi-ethnic GWAS meta-analysis methodologies are mature (PMID 39022911, 40162269).

    Partially Verified: Only a few constitution types within the Nine Constitutions (e.g., Phlegm-Dampness, Yin Deficiency, Yang Deficiency) have substantial genetic evidence; existing GWAS predominantly utilize Korean four-constitution samples, and no large-sample whole-genome study for TCM Nine Constitutions meets modern standards.

    Pending Verification: ① Can multi-center large-sample GWAS under a unified constitution scale replicate known loci and discover new ones? ② Can constitution PRS independently predict disease outcomes and transfer across ethnicities? ③ Are constitution-gene associations stable under different environmental exposures? ④ Does “modifiable constitution” correspond to measurable genomic or epigenetic changes.

    Related Reading

    Evidence Level: A
    References: PMID 19954339, PMID 23548117, PMID 17898945, PMID 20681271, PMID 22110543, PMID 39022911, PMID 40162269


    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.