Energy-Related symptom phenotype is associated with immuno-metabolic dysregulations in bipolar disorder: a data-driven approach

Individuals with bipolar disorder (BD) exhibit a life expectancy reduced by 10–20 years compared to the general population, largely attributable to cardiovascular and metabolic comorbidities [1]. Converging evidence has identified an immuno-metabolic depression subtype characterised by energy-related (ER) depressive symptoms (hypersomnia, hyperphagia, psychomotor retardation, fatigue, anhedonia) co-occurring with low-grade inflammation and metabolic abnormalities [2]. This concept has been developed predominantly in unipolar depression, with immuno-metabolic dysregulations affecting 30-40% of individuals with depression [2][3]. However, evidence for its transdiagnostic extension to BD remains limited. Given evidence that this association persists after lifestyle adjustment [4] and that ER phenotypes are suspected in BD, we hypothesised that data-driven symptom stratification would identify a phenotype with distinct symptom profile and immunometabolic alterations.

Data from the FACE-BD cohort (baseline, N = 6 203) were used. Latent class analysis (LCA) was performed on domain-level indicators from the QIDS-SR16. Immuno-metabolic outcomes included waist circumference, BMI, systolic blood pressure, fasting glucose, HDL-cholesterol, triglycerides, CRP (continuous inflammatory index and binary thresholds ≥1 and ≥3 mg/L), Pan-immune Inflammation Value (PIV), composite metabolic and combined immuno-metabolic indices, and metabolic syndrome (modified NCEP-ATP III). Three nested models were fitted: M1 (age, sex, education), M2 (M1 plus smoking, alcohol use disorder, illness duration), and M3 (M2 plus atypical antipsychotics, lithium). Linearity of continuous confounders was verified using fractional polynomials. Analyses were triangulated using a standardised continuous ER symptom score (SAEA: Sleep-Appetite-Energy-Anhedonia) and a binary ER versus non-ER comparison. Additional data-driven approaches aren’t presented here.

Among 5 362 participants with class assignment, LCA identified four profiles: Low-symptom (n = 1 532, reference), Melancholic-like (n = 786), ER moderate (n = 939), and ER severe (n = 2 105). In the primary model (M2), both ER classes were associated with greater adiposity (waist circumference: +2.61 and +2.19 cm, both p < 0.001), adverse lipid profile (lower HDL, both p < 0.02; higher triglycerides, both p < 0.001), elevated PIV (both p < 0.05), higher composite metabolic index (both p < 0.001), and increased metabolic syndrome risk (OR = 1.58 [1.20–2.08] and 1.41 [1.12–1.78]). No significant associations were found for CRP or systolic blood pressure. The Melancholic-like class showed no immuno-metabolic alterations despite comparable depressive severity to the ER moderate class (median QIDS = 9 for both), supporting symptom-profile specificity over a severity effect. Associations persisted after medication adjustment (M3; N ≈ 1 600). The SAEA score confirmed a dose–response relationship (metabolic syndrome: OR per SD = 1.30 [1.19–1.42], p < 0.001).

These findings provide novel data-driven evidence that an ER symptom profile in BD is independently associated with a distinct immuno-metabolic phenotype. The consistency of results across three exposure operationalisations, three nested adjustment models, and fractional polynomial sensitivity analyses strengthens the robustness of these findings. Critically, the Melancholic-like class showing no such alterations despite similar depressive severity indicates that metabolic risk is carried by the ER symptom profile specifically, not by depression in general. These results support the transdiagnostic extension of immuno-metabolic depression to BD and highlight the clinical relevance of ER symptom stratification for targeted cardiometabolic monitoring.

References

1. Godin, O., Etain, B., Henry, C., Bougerol, T., Courtet, P., Mayliss, L., Passerieux, C., Azorin, J.M., Kahn, J.P., Gard, S., Costagliola, D., Leboyer, M., 2014. Metabolic syndrome in a French cohort of patients with bipolar disorder: results from the FACE-BD cohort. J. Clin. Psychiatry 75 (10), 1078-1085.
2. Penninx, B.W.J.H., Lamers, F., Jansen, R., Berk, M., Khandaker, G.M., De Picker, L., Milaneschi, Y., 2024. Immuno-metabolic depression: from concept to implementation. Lancet Reg. Health Eur. 48, 101166.
3. Vancampfort, D., Vansteelandt, K., Correll, C.U., Mitchell, A.J., De Herdt, A., Sienaert, P., Probst, M., De Hert, M., 2013. Metabolic syndrome and metabolic abnormalities in bipolar disorder: a meta-analysis of prevalence rates and moderators. Am. J. Psychiatry 170 (3), 265-274.
4. Lamers F, Milaneschi Y, Vinkers CH, Schoevers RA, Giltay EJ, Penninx BWJH. Depression profilers and immuno-metabolic dysregulation: Longitudinal results from the NESDA study. Brain Behav Immun. 2020 Aug;88:174-183.