目的 探讨交泰丸单用及联用选择性5-羟色胺再摄取抑制剂（selective serotonin reuptake inhibitors，SSRIs）治疗抑郁症心肾不交证时，中医证候与外周血神经递质水平变化的相关性，并探索性构建疗效预测模型。方法 基于随机对照试验的事后分析集（n＝91），分为交泰丸组（n＝31）、交泰丸联合SSRIs组（n＝30）和SSRIs组（n＝30）。治疗8周后，以汉密尔顿抑郁量表减分率≥50%定义应答。比较3组治疗前后神经递质与中医证候变化。采用Spearman相关分析评估中医证候变化值与神经递质5-羟色胺（5-hydroxytryptamine，5-HT）、去甲肾上腺素（norepinephrine，NE）和多巴胺（dopamine，DA）变化值的关联。采用Firth惩罚似然Logistic回归（校正年龄、性别等混杂因素）分析各组基线特征与应答的独立关联，筛选潜在预测因子。基于有统计学意义的预测因子分别构建预测模型，并采用Bootstrap法（1 000次重复）进行内部验证以校正过拟合，报告乐观校正后的曲线下面积（area under curve，AUC）评估模型判别效能。结果 3组间应答率无统计学差异（P＞0.05）。治疗前后比较显示，3组患者5-HT、NE水平及大部分中医证候评分均较基线显著改善（P＜0.05）。相关性分析显示，交泰丸组5-HT变化值与失眠多梦变化值（r＝0.521，P＝0.003）、咽干口燥变化值（r＝0.438，P＝0.014）呈正相关。多因素分析显示，交泰丸组基线“健忘”（OR＝0.60，P＝0.022）是疗效的负向预测因素；交泰丸联合SSRIs组基线“心悸”是疗效的正向预测因素（OR＝1.62，P＝0.029），“遗精或月经不调”是负向预测因素（OR＝0.22，P＝0.007），但后者在治疗后其症状改善不显著。预测模型分析显示，交泰丸组基于“健忘”的模型校正后AUC为0.669（原始AUC＝0.838）。交泰丸联合SSRIs组中同时包含“心悸”与“遗精或月经不调”的复合模型预测效能最佳，其校正后AUC为0.809（原始AUC＝0.937），提示具有初步的区分潜力，但也提示存在过拟合现象。结论 通过探索交泰丸治疗抑郁症心肾不交证患者的中医证候与神经递质变化关联及疗效预测模型，为临床以“心火亢盛”和“肾精亏虚”为主的抑郁症心肾不交证亚型的治疗方案选择提供了初步依据，为交泰丸的精准应用提供了探索性线索，但预测效能需在前瞻性大样本研究中进一步验证。

Objective To preliminarily explore the correlation between changes in traditional Chinese medicine (TCM) syndromes and peripheral neurotransmitter levels in patients with depression of heart-kidney non-interaction syndrome treated with Jiaotai Pill (交泰丸, JTW) alone or in combination with selective serotonin reuptake inhibitors (SSRIs), and to exploratorily develop efficacy prediction models. Methods This post-hoc analysis utilized data from a randomized controlled trial (n = 91). Patients were divided into the JTW group (n = 31), JTW combined with SSRIs group (n = 30), and SSRIs group (n = 30). Response after eight weeks of treatment was defined as a ≥ 50% reduction rate in the Hamilton depression scale (HAMD) score. Changes in neurotransmitter levels and TCM syndrome scores before and after treatment were compared among the three groups. Spearman correlation analysis was used to assess the associations between changes in individual TCM syndrome scores and changes in neurotransmitter levels of 5-hydroxytryptamine (5-HT), norepinephrine (NE), and dopamine (DA). Firth's penalized-likelihood logistic regression, adjusted for confounders such as age and sex, was employed to analyze the independent associations between baseline characteristics and treatment response within each group, identifying potential predictors. Predictive models were constructed based on statistically significant predictors. Bootstrap internal validation (1 000 repetitions) was performed to correct for overfitting, and the optimism-corrected area under the curve (AUC) was reported to evaluate the discriminatory performance of the models. Results No statistically significant difference in response rates was observed among the three groups (P > 0.05). Comparisons before and after treatment showed that 5-HT and NE levels, as well as most TCM syndrome scores, improved significantly from baseline in all three groups (P < 0.05). Correlation analysis revealed that in the JTW group, Δ5-HT was positively correlated with Δinsomnia/dreamfulness (r = 0.521, P = 0.003) and Δdry throat and mouth (r = 0.438, P = 0.014). Multivariate analysis indicated that in the JTW group, baseline “forgetfulness” was a negative predictor of efficacy (OR = 0.60, P = 0.022). In the combination group, baseline “palpitations” was a positive predictor (OR = 1.62, P = 0.029), while “spermatorrhea or irregular menstruation” was a negative predictor (OR = 0.22, P = 0.007). Notably, the symptom of “spermatorrhea or irregular menstruation” itself did not show significant improvement after treatment. Predictive model analysis showed that in the JTW group, the model based on “forgetfulness” had a corrected AUC of 0.669 (original AUC = 0.838). In the combination group, the composite model including both “palpitations” and “spermatorrhea or irregular menstruation” demonstrated the best predictive performance, with a corrected AUC of 0.809 (original AUC = 0.937), suggesting preliminary discriminatory ability but also indicating potential overfitting. Conclusion By exploring the associations between TCM syndromes and neurotransmitter changes in depressed patients with heart-kidney non-interaction syndrome treated with JTW and by developing efficacy prediction models, this study provides preliminary evidence to guide treatment selection for clinical subtypes predominantly characterized by “exuberance of heart fire” or “kidney essence deficiency”. These findings offer exploratory clues for the precise application of JTW, though its predictive performance requires further validation in prospective and large-scale studies.

R285.64

国家自然科学基金面上项目（82074220）