2026 ECNP Congress

PS04-4044

Longitudinal multidomain maternal profiles and neonatal outcomes in perinatal mental health

Sara Peri — E. Tassi, C. Cabasino, G. Delvecchio, P. Enrico, A. Pigoni, I. Tesic, L. Squarcina, A. Ferro, Y. Barone, S. Pozzoli, V. Favalli, A. Marano, V. Tombola, A. Finardi, C. Pini, S. Busilacchi, E. Rossi, S. Boito, C. Scacchetti, I. Cetin, N. Persico, M. Fumagalli, R. Furlan, P. Brambilla · Foundation IRCCS Ca' Granda Ospedale Maggiore Policlinico, Department of Neurosciences and Mental Health, Milan, Italy

2026 ECNP Congress

Background

The perinatal period is characterized by major biological and psychological adaptations, including dynamic immune changes [3] and increased vulnerability to anxiety and depressive symptoms [1], [2]. Although perinatal mental health is increasingly recognized, most studies still focus on single domains using cross-sectional symptom-based approaches. Methodological frameworks integrating longitudinal multimodal sources, as biological and clinical data, remain limited. This study aimed to develop and apply a longitudinal statistical framework to stratify neonatal outcomes based on multimodal maternal data including immunological, clinical and sociodemographic set.

Methods

Our longitudinal cohort study enrolled 394 pregnant women (Department of Psychiatry, Policlinico Hospital, Milan), including healthy controls (N=193), individuals with a history of anxiety/depression (N=139), and with current anxious/depressive symptoms (N=62). Maternal psychiatric status was assessed, and peripheral blood samples were collected at three gestational timepoints (week 16-18, 20-22, 30-33). Neonatal health outcomes at birth included standard perinatal measures. Clinical evaluations utilized validated self-report and clinician-administered scales to assess maternal psychopathology. Concurrently, maternal peripheral immune gene expression was longitudinally profiled using whole blood RNA extraction, cDNA synthesis, and custom OpenArray PCR arrays. A multistep statistical learning pipeline was implemented to integrate three maternal data blocks: 43 immunological markers, 9 clinical/psychometric scales, and 13 baseline sociodemographic/clinical covariates. First, longitudinal variables were modelled using linear mixed-effects models to account for repeated observations and isolate within-subject variation [4]. Principal component analysis was applied to the immunological and clinical blocks. Block-specific distance matrices were then combined. Principal coordinates analysis provided a Euclidean representation for Ward hierarchical clustering. Between-cluster differences in neonatal outcomes were assessed using global and pairwise comparisons.

Results

The final Ward solution identified three maternal profiles. Significant differences among clusters emerged for neonatal length (p = 0.012) and Apgar score at 5 minutes (p = 0.043). Cluster 3 showed lower mean neonatal length compared with Cluster 1 and Cluster 2, and lower Apgar score at 5 minutes compared with Cluster 1. Notably, these neonatal measures were within the following ranges: Apgar scores 8-10, length: 44-55.5 cm. Cluster 1 was the largest group (n = 241; 61.2%) and showed a relatively stable, lower-risk multidimensional profile, with moderate longitudinal dynamics across immunological and clinical domains. Cluster 2 (n = 85; 21.6%) was characterized by greater clinical variation of HAM-A, HAM-D, BPRS and GAF over pregnancy. Cluster 3 (n = 68; 17.3%) showed the most distinctive biological profile, with stronger longitudinal variation in immune gene expression (GATA3, ENTPD1, TNF) and a more heterogeneous clinical configuration, particularly of EPDS, SCL-GSI and SCL-PST. The overall silhouette width was moderate (0.157), indicating partially overlapping but interpretable maternal profiles.

Conclusions

These findings suggest that integrating longitudinal immunological, clinical and contextual maternal data can identify interpretable multidimensional profiles during pregnancy, capturing distinct patterns of immune regulation (GATA3, ENTPD1, TNF) and psychopathology (EPDS, SCL distress measures, HAM-A/HAM-D). Selective associations with neonatal length and Apgar score support the relevance of maternal longitudinal profiling for early risk stratification. This framework may provide a reproducible basis for future studies on perinatal mental health and maternal–neonatal outcomes.

References

  1. Howard, L. M., Molyneaux, E., Dennis, C. L., Rochat, T., Stein, A., & Milgrom, J. (2014). Non-psychotic mental disorders in the perinatal period. The Lancet, 384(9956), 1775-1788.
  2. Stein, A., Pearson, R. M., Goodman, S. H., Rapa, E., Rahman, A., McCallum, M., … & Pariante, C. M. (2014). Effects of perinatal mental disorders on the fetus and child. The Lancet, 384(9956), 1800-1819.
  3. Mor, G., & Cardenas, I. (2010). The immune system in pregnancy: a unique complexity. American journal of reproductive immunology, 63(6), 425-433.
  4. Laird, N. M., & Ware, J. H. (1982). Random-effects models for longitudinal data. Biometrics, 963-974.