Real-time fMRI neurofeedback (rtfMRI-nf) left amygdala (LA) training is a promising intervention for major depressive disorder (MDD). We have previously proposed that rtfMRI-nf LA training may reverse depression-associated regional impairments in neuroplasticity and restore information flow within emotion-regulating neural circuits. Inflammatory cytokines as well as the neuroactive metabolites of an immunoregulatory pathway, i.e. the kynurenine pathway (KP), have previously been implicated in neuroplasticity. Therefore, in this proof-of-principle study, we investigated the association between rtfMRI-nf LA training and circulating inflammatory mediators and KP metabolites. Based on our previous work, the primary variable of interest was the ratio of the NMDA-receptor antagonist, kynurenic acid to the NMDA receptor agonist, quinolinic acid (KynA/QA), a putative neuroprotective index. We tested two main hypotheses. i. Whether rtfMRI-nf acutely modulates KynA/QA, and ii. whether baseline KynA/QA predicts response to rtfMRI-nf.

Twenty-nine unmedicated participants who met DSM-5 criteria for MDD based on the Mini-International Neuropsychiatric Interview and had current depressive symptoms (Montgomery-Åsberg Depression Rating Scale (MADRS) score > 6) completed two rtfMRI-nf sessions to upregulate LA activity (Visit1 and 2), as well as a follow-up (Visit3) without rtfMRI-nf. All visits occurred at two-week intervals. At all three visits, the MADRS was administered to participants and serum samples for the quantification of inflammatory cytokines and KP metabolites were obtained. First, the longitudinal changes in the MADRS score and immune markers were tested by linear mixed effect model analysis. Further, utilizing a linear regression model, we investigated the relationship between rtfMRI-nf performance and immune markers. After two sessions of rtfMRI-nf, MADRS scores were significantly reduced (t[58] = −4.07, p = 0.009, d = 0.56). Thirteen participants showed a ≥ 25% reduction in the MADRS score (the partial responder group). There was a significant effect of visit (F[2,58] = 3.17, p = 0.05) for the neuroprotective index, KynA to 3-hydroxykynurenine (3-HK), that was driven by a significant increase in KynA/3-HK between Visit1 and Visit3 (t[58] = 2.50, p = 0.03, d = 0.38). A higher baseline level of KynA/QA (β = 5.23, p = 0.06; rho = 0.49, p = 0.02) was associated with greater ability to upregulate the LA. Finally, for exploratory purposes correlation analyses were performed between the partial responder and the non-responder groups as well as in the whole sample including all KP metabolites and cytokines. In the partial responder group, greater ability to upregulate the LA was correlated with an increase in KynA/QA after rtfMRI-nf (rho = 0.75, p = 0.03). The results are consistent with the possibility that rtfMRI-nf decreases metabolism down the so-called neurotoxic branch of the KP. Nevertheless, non-specific effects cannot be ruled out due to the lack of a sham control. Future, controlled studies are needed to determine whether the increase in KynA/3HK and KynA/QA is specific to rtfMRI-nf or whether it is a non-specific correlate of the resolution of depressive symptoms. Similarly, replication studies are needed to determine whether KynA/QA has clinical utility as a treatment response biomarker.

实时 fMRI 神经反馈 (rtfMRI-nf) 左杏仁核 (LA) 训练是治疗重度抑郁症 (MDD) 的一种很有前景的干预措施。我们之前曾提出 rtfMRI-nf LA 训练可以逆转与抑郁相关的神经可塑性区域损伤，并恢复情绪调节神经回路中的信息流。炎症细胞因子以及免疫调节途径的神经活性代谢物，即犬尿氨酸途径 (KP)，以前曾与神经可塑性有关。因此，在这项原理验证研究中，我们调查了 rtfMRI-nf LA 训练与循环炎症介质和 KP 代谢物之间的关联。基于我们之前的工作，感兴趣的主要变量是 NMDA 受体拮抗剂犬尿酸与 NMDA 受体激动剂喹啉酸 (KynA/QA) 的比率，推定的神经保护指数。我们测试了两个主要假设。一世。rtfMRI-nf 是否会急剧调节 KynA/QA，以及 ii. 基线 KynA/QA 是否预测对 rtfMRI-nf 的反应。

根据 Mini-International Neuropsychiatric 访谈，符合 DSM-5 MDD 标准并有当前抑郁症状（蒙哥马利-Åsberg 抑郁量表 (MADRS) 评分 > 6）的 29 名未服用药物的参与者完成了两次 rtfMRI-nf 会议以上调 LA 活动（访问 1 和 2），以及没有 rtfMRI-nf 的随访（访问 3）。所有访问均以两周为间隔进行。在所有 3 次访问中，参与者都接受了 MADRS，并获得了血清样本，用于量化炎性细胞因子和 KP 代谢物。首先，通过线性混合效应模型分析测试 MADRS 评分和免疫标志物的纵向变化。此外，我们利用线性回归模型研究了 rtfMRI-nf 性能与免疫标志物之间的关系。经过两次 rtfMRI-nf 疗程后，t [58] = -4.07，p  = 0.009，d  = 0.56）。13 名参与者的 MADRS 评分降低 ≥ 25%（部分反应组）。访问 ( F [2,58] = 3.17, p  = 0.05) 对神经保护指数 KynA 对 3-羟基犬尿氨酸 (3-HK) 有显着影响，这是由 KynA/3-HK 的显着增加驱动的访问 1 和访问 3 之间（t [58] = 2.50，p  = 0.03，d  = 0.38）。较高的 KynA/QA 基线水平（β = 5.23，p  = 0.06；rho = 0.49，p = 0.02) 与上调 LA 的能力更强相关。最后，出于探索性目的，在部分响应者和非响应者组之间以及包括所有 KP 代谢物和细胞因子的整个样本中进行了相关性分析。在部分反应组中，上调 LA 的能力更强与 rtfMRI-nf 后 KynA/QA 的增加相关（rho = 0.75，p = 0.03）。结果与 rtfMRI-nf 降低 KP 所谓神经毒性分支的代谢的可能性一致。然而，由于缺乏假控制，不能排除非特异性影响。未来，需要对照研究来确定 KynA/3HK 和 KynA/QA 的增加是否特定于 rtfMRI-nf 或它是否与抑郁症状的消退具有非特异性相关性。同样，需要重复研究来确定 KynA/QA 作为治疗反应生物标志物是否具有临床效用。