Individuals with obsessive-compulsive disorder (OCD) show persistent avoidance behaviors, often in the absence of actual threat. Quality-of-life costs and heterogeneity support the need for novel brain-behavior intervention targets. Informed by mechanistic and anatomical studies of persistent avoidance in rodents and nonhuman primates, our goal was to test whether connections within a hypothesized persistent avoidance–related network predicted OCD-related harm avoidance (HA), a trait measure of persistent avoidance. We hypothesized that 1) HA, not an OCD diagnosis, would be associated with altered endogenous connectivity in at least one connection in the network; 2) HA-specific findings would be robust to comorbid symptoms; and 3) reliable findings would replicate in a holdout testing subsample. Using resting-state functional connectivity magnetic resonance imaging, cross-validated elastic net for feature selection, and Poisson generalized linear models, we tested which connections significantly predicted HA in our training subsample ( = 73; 71.8% female; healthy control group = 36, OCD group = 37); robustness to comorbidities; and replicability in a testing subsample ( = 30; 56.7% female; healthy control group = 15, OCD group = 15). Stronger inverse connectivity between the right dorsal anterior cingulate cortex and right basolateral amygdala and stronger positive connectivity between the right ventral anterior insula and left ventral striatum were associated with greater HA across groups. Network connections did not discriminate OCD diagnostic status or predict HA-correlated traits, suggesting sensitivity to trait HA. The dorsal anterior cingulate cortex–basolateral amygdala relationship was robust to controlling for comorbidities and medication in individuals with OCD and was also predictive of HA in our testing subsample. Stronger inverse dorsal anterior cingulate cortex–basolateral amygdala connectivity was robustly and reliably associated with HA across groups and in OCD. Results support the relevance of a cross-species persistent avoidance–related network to OCD, with implications for precision-based approaches and treatment.

中文翻译：

---

内源性功能连接的特定模式与强迫症的伤害避免相关


患有强迫症（OCD）的人通常在没有实际威胁的情况下表现出持续的回避行为。生活质量成本和异质性支持对新的大脑行为干预目标的需求。根据啮齿动物和非人类灵长类动物持续回避的机制和解剖学研究，我们的目标是测试假设的持续回避相关网络中的连接是否可以预测与强迫症相关的伤害回避（HA），这是持续回避的一种特征测量。我们假设 1) HA，而不是强迫症诊断，将与网络中至少一个连接的内源性连接改变相关； 2) HA 特定的发现对于合并症症状是稳健的； 3）可靠的发现将在坚持测试子样本中复制。使用静息态功能连接磁共振成像、用于特征选择的交叉验证弹性网络和泊松广义线性模型，我们测试了哪些连接可显着预测训练子样本中的 HA（= 73；71.8% 女性；健康对照组 = 36，强迫症组 = 37)；对合并症的稳健性；以及测试子样本中的可重复性（= 30；56.7% 女性；健康对照组 = 15，强迫症组 = 15）。右侧背侧前扣带皮层和右侧基底外侧杏仁核之间较强的反向连接以及右侧腹侧前岛叶和左侧腹侧纹状体之间较强的正向连接与各组间较高的 HA 相关。网络连接不能区分 OCD 诊断状态或预测 HA 相关特征，表明对 HA 特征的敏感性。 背侧前扣带皮层与基底外侧杏仁核的关系对于控制强迫症患者的合并症和药物治疗非常有效，并且在我们的测试子样本中也可以预测 HA。更强的反向背侧前扣带皮层-基底外侧杏仁核连接与各组和强迫症患者的 HA 密切相关。结果支持跨物种持续回避相关网络与强迫症的相关性，对基于精确的方法和治疗具有影响。