From symptoms to systems: A comprehensive review of the clinical utility of classic and novel approaches to major depression subtyping

Major depressive disorder (MDD) is the leading cause of disability worldwide, yet current treatments are not effective for all. Efforts to identify distinct subtypes of MDD with prognostic value have yielded mixed success. This review aims to guide clinical decision-making by highlighting the relative usefulness of subtypes currently used in clinical practice, as well as proposed biologically-informed subtypes. Of the subtypes historically used in clinical practice, those that classify patients based on overall severity, the absence/presence of psychotic symptoms and proximal aetiologies (e.g. pregnancy), have some utility in informing treatment guidelines. Severity is particularly informative as it can predict overall prognosis and likelihood of treatment response, as well as likely preference between psychotherapy and pharmacotherapy. In contrast, classifying patients by symptom profile (for example melancholic, anxious or atypical) and symptom seasonality (seasonal affective disorder) has less clear prognostic utility. This appears to be because patients fail to neatly cluster as a function of symptom differences, making these subtypes arbitrary. Furthermore, the ability of clinicians to effectively identify differences in symptom seasonality appears poor. Looking forwards, emerging subtypes linked to HPA axis, immuno-metabolic and reward circuit dysfunction appear identifiable by extant validated questionnaires and fluid biomarkers and show great promise for predicting treatment response to novel drugs, including anti-glucocorticoid, anti-inflammatory and anti-anhedonic drugs. Saying this, there are some questions yet to be answered about these subtypes, such as there degree of overlaps, something that may be solved in future by further use of "big data", computational approaches to treatment response prediction.