WEBB FELLOW: Using Neural Subtyping to Examine the Heterogeneity of Postpartum Depression

Abstract:

Background:

Postpartum depression (PPD) is a common disorder and carries significant morbidity for affected mothers and compromised outcomes in their infants (Slomien et al., 2019). PPD relates to alterations in cognition and hormones with emerging novel treatments targeting the maternal-infant dyad (Werner et al., 2016) or using injectable allopregnanolone (Meltzer-Brody et al., 2018); yet, these associations are inconsistent across the literature (Schiller et al., 2015). We will apply a machine learning approach to determine whether neural-based PPD subtypes show more consistent cognitive and hormonal profiles.



Methods:

We will utilize neural-clinical “biotypes” identified in major depressive disorder (MDD; Drysdale et al., 2017) to parse brain connectivity patterns in PPD. First, we will characterize whether patients with PPD predominantly fall within specific MDD-defined neural subtypes. We will obtain resting state functional MRI (rsfMRI) scans in mothers with and without PPD during the first two months after childbirth. We will relate resting state functional connectivity patterns to clinical symptoms using canonical correlation. Each patient will then be matched to MDD-defined biotypes. Patients will provide a blood sample for serum allopregnanolone quantification and complete a cognitive task measuring infant-related social attention. Finally, we will collect depression inventories at intake and 3-month intervals for one year. We will compare allopregnanolone levels, infant-related attention, and clinical outcomes between PPD patients in different rsfMRI-based subtypes.



Results:

We plan data collection starting later this year.



Discussion:

We anticipate that while PPD will represent a more neurally homogenous disorder than MDD, patients will demonstrate some heterogeneity. We predict connectivity-based patient subgroups will also exhibit distinct hormonal and cognitive profiles. This information could guide treatment assignment based on biological phenotypes. Patients with more disrupted infant-related attention might benefit more from a mother-infant dyadic intervention while those with lower allopregnanolone levels might respond better to neurosteroid-based treatment. This work will also set the stage to test whether a subtyping approach can prospectively predict PPD risk.



Conclusion:

We are only beginning to unpack the pathogenesis of PPD. Delineating PPD’s neural heterogeneity could clarify discrepancies in previous hormonal and cognitive studies, guide intervention, and improve treatment outcomes.



References:

Slomien J, Honvo G, Emonts P, et al: Consequences of maternal postpartum depression: A systematic review of maternal and infant outcomes. Women’s Health 2019;15:1-55.

Werner EA, Gustafsson HC, Lee S, et al: PREPP: Postpartum Depression Prevention through the mother-infant dyad. Arch Women’s Ment Health 2016;19(2):229-242.

Meltzer-Brody S, Colquhoun H, Riesenberg R, et al: Brexanolone injection in post-partum depression: two multicenter, double-blind, randomized, placebo-controlled, phase 3 trials. Lancet 2018;392:1058-70.

Schiller CE, Meltzer-Brody S, Rubinow DR: The role of reproductive hormones in postpartum depression. CNS Spectrums 2015;20(1):48-59.

Drysdale AT, Grosenick L, Downar J, et al: Resting-state connectivity biomarkers define neurophysiological subtypes of depression. Nature Medicine 2017;23(1):28-38.