Table 2.

Neurophysiological and clinical correlates of candidate brain-based depression subtypes.

Study	Grouping	Symptoms	Treatment response	Brain circuit
[88]	Dimension 1	Mood (feeling sad, suicidality, anhedonia, irritability, persecutory/suspicious, loss sense of self)		Positively correlated    Intra-DMN    DMN to frontoparietal    Salience-ventral attention and frontoparietal    Frontoparietal
	Dimension 2	Psychosis (auditory perceptions, odd/intrusive thoughts, reality confusion, audible thoughts, superstitions, overly energetic, pressured speech)		Positively correlated    Intra-DMN    DMN to executive (frontoparietal and salience)    Frontoparietal
	Dimension 3	Fear (of traveling—agoraphobia, social phobia)		Negatively correlated    Intra-DMN    Salience to ventral attention Positively correlated to frontoparietal circuits
	Dimension 4	Externalizing behavior (trouble following instructions, irritability to unfairness, attention issues, losing temper)		Negatively correlated    Intra-DMN    Frontoparietal to dorsal attention Positively correlated    Frontoparietal    Salience to frontoparietal
[89]	Dimension 1	Externalizing/internalizing Sex Positively correlated with age		Positively correlated    Attentional Networks    FPTC Negatively correlated    Limbic    Intra-subcortical
	Dimension 2	Emotional well-being vs. distress Negatively correlated with age		Negatively correlated    Attentional Networks    FPTC Positively correlated    Limbic    Intra-subcortical
[95]	Subgroup A	19% of depressed subjects (50% of healthy)		No change in connectivity during positive mood induction task
	Subgroup B	81% of depressed subjects (50% of healthy) Difficulty sustaining positive affect during task Negative bias on reaction time Higher symptom severity		Hyperconnectivity in ventral affective network during positive mood induction task
[96]	Subgroup A	71% patients 65% female		No change in DMN connectivity
	Subgroup B	Recurrent depression (63%) Comorbid anxiety (42%) 87% female		Change in the circuit path direction of the dACC (in the DMN)
[97]	Subgroup 1	Comorbid anxiety with severe depression (16% vs. 9% in Subgroup 2) Longer duration (mean 57.5 months)
	Subgroup 2	Shorter duration (mean 37 months)
[98]	Subgroup 1	Anxiety Insomnia Fatigue	Strong therapeutic effect for TMS at DMPFC target (82.5% of patients improved significantly)	Decreased RSFC in the    Frontoamygdala    ACC    Orbitofrontal circuits
	Subgroup 2	Fatigue Lower severity	No significant effect for TMS at DMPFC target	Decreased RSFC in the    ACC    Orbitofrontal circuits
	Subgroup 3	Anhedonia Psychomotor retardation	Mild therapeutic effect for TMS at DMPFC target (61% of patients improved significantly)	Increased RSFC in the    Thalamic    Frontostriatal    Subcortical circuits
	Subgroup 4	Anhedonia Anxiety Insomnia	No significant effect for TMS at DMPFC target	Decreased RSFC in the frontoamygdala circuit Increased RSFC in the    Thalamic    Frontostriatal    Subcortical circuits
[101]	Subgroup 1	High CATS High RSFC Low baseline BDI High BDI 6 weeks post-treatment	Treatment-resistant to SSRI	High Mean FC (angular gyrus in DMN)
	Subgroup 2	Low CATS Moderate RSFC Low baseline BDI Low BDI 6 weeks post-treatment		Moderate Mean FC (angular gyrus in DMN)
	Subgroup 3	High CATS Low RSFC High baseline BDI Low BDI 6 weeks post-treatment	Treatment-responsive to SSRI	Low Mean FC (angular gyrus in DMN)

Efforts to identify points of convergence between these studies are complicated by methodological differences in clustering techniques and criteria, data types used for clustering, subject samples, and other technical details. This is especially true for analyses of brain circuit function, which vary widely across studies. Given the very different methods employed by these studies and their varying findings, the available data do not currently support even preliminary conclusions about points of convergence. However, two themes recur in the clinical correlates of the subtypes in multiple studies. First, the presence or absence of comorbid anxiety- and fear-related symptoms is an important feature of the subtyping results in four of six studies [88, 95, 96, 98]. Second, the subtypes differ with respect to overall depression severity in three studies [96, 98, 101] and with respect to recurrence, which often correlates with severity, in a fourth study [95].

dACC dorsal anterior cingulate cortex, BDI Beck Depression Inventory, CATS Child and Adolescent Trauma Screen, DMN default mode network, DMPFC dorsomedial prefrontal cortex, FPTC frontoparietal task control, RSFC resting state functional connectivity, SSRI selective-serotonin reuptake inhibitor, TMS transcranial magnetic stimulation.