Melancholia has been proposed as a qualitatively distinct depressive subtype associated with a characteristic symptom profile (psychomotor retardation, profound anhedonia) and a better response to biological therapies. Existing work has suggested that individuals with melancholia are blunted in their display of positive emotions and differ in their neural response to emotionally evocative stimuli. Here, we unify these brain and behavioural findings amongst a carefully phenotyped group of seventy depressed participants, drawn from an established Australian database (the Australian Genetics of Depression Study) and further enriched for melancholia (high ratings of psychomotor retardation and anhedonia). Melancholic (n = 30) or non-melancholic status (n = 40) was defined using a semi-structured interview (the Sydney Melancholia Prototype Index). Complex facial expressions were captured whilst participants watched a movie clip of a comedian and classified using a machine learning algorithm. Subsequently, the dynamics of sequential changes in brain activity were modelled during the viewing of an emotionally evocative movie in the MRI scanner. We found a quantitative reduction in positive facial expressivity amongst participants with melancholia, combined with differences in the synchronous expression of brain states during positive epochs of the movie. In non-melancholic depression, the display of positive affect was inversely related to the activity of cerebellar regions implicated in the processing of affect. However, this relationship was reduced in those with a melancholic phenotype. Our multimodal findings show differences in evaluative and motoric domains between melancholic and non-melancholic depression through engagement in ecologically valid tasks that evoke positive emotion. These findings provide new markers to stratify depression and an opportunity to support the development of targeted interventions.

忧郁症被认为是一种性质独特的抑郁亚型，与特征性症状（精神运动迟缓、极度快感缺乏）和对生物疗法的更好反应相关。现有的研究表明，患有忧郁症的人表现出积极情绪的能力较弱，并且对情绪唤起刺激的神经反应也有所不同。在这里，我们将这些大脑和行为发现统一到一个由 70 名抑郁症参与者组成的仔细表型组中，这些参与者来自一个已建立的澳大利亚数据库（澳大利亚抑郁症遗传学研究），并进一步丰富了忧郁症（精神运动迟缓和快感缺乏的高评级）。忧郁（ n = 30）或非忧郁状态（ n = 40）是通过半结构化访谈（悉尼忧郁原型指数）来定义的。当参与者观看喜剧演员的电影剪辑时，复杂的面部表情被捕获，并使用机器学习算法进行分类。随后，在 MRI 扫描仪中观看令人情绪激动的电影时，对大脑活动连续变化的动态进行了建模。我们发现患有忧郁症的参与者的积极面部表情数量减少，并且在电影的积极时期大脑状态的同步表达存在差异。在非忧郁型抑郁症中，积极情感的表现与处理情感的小脑区域的活动呈负相关。然而，这种关系在具有忧郁表型的人中减弱。 我们的多模式研究结果表明，通过参与唤起积极情绪的生态有效任务，忧郁型抑郁症和非忧郁型抑郁症之间的评估和运动领域存在差异。这些发现为抑郁症分层提供了新的标记，并为支持制定有针对性的干预措施提供了机会。