||Background. Etiological research of depression and anxiety disorders has been hampered by
diagnostic heterogeneity. In order to address this, researchers have tried to identify more
homogeneous patient subgroups. This work has predominantly focused on explaining interpersonal
heterogeneity based on clinical features (i.e. symptom profiles). However, to explain
interpersonal variations in underlying pathophysiological mechanisms, it might be more
effective to take biological heterogeneity as the point of departure when trying to identify subgroups.
Therefore, this study aimed to identify data-driven subgroups of patients based on
Methods. Data of patients with a current depressive and/or anxiety disorder came from the
Netherlands Study of Depression and Anxiety, a large, multi-site naturalistic cohort study
(n = 1460). Thirty-six biomarkers (e.g. leptin, brain-derived neurotrophic factor, tryptophan)
were measured, as well as sociodemographic and clinical characteristics. Latent class analysis
of the discretized (lower 10%, middle, upper 10%) biomarkers were used to identify different
Results. The analyses resulted in three classes, which were primarily characterized by different
levels of metabolic health: ‘lean’ (21.6%), ‘average’ (62.2%) and ‘overweight’ (16.2%).
Inspection of the classes’ clinical features showed the highest levels of psychopathology, severity
and medication use in the overweight class.
Conclusions. The identified classes were strongly tied to general (metabolic) health, and did
not reflect any natural cutoffs along the lines of the traditional diagnostic classifications. Our
analyses suggested that especially poor metabolic health could be seen as a distal marker for
depression and anxiety, suggesting a relationship between the ‘overweight’ subtype and internalizing