Immune-neuroendocrine crosstalk in mood and psychotic disorders: A meta-analysis and systematic review

Bidirectional interactions between immune and neuroendocrine mechanisms are involved in mood and psychotic disorders, although individual studies report inconsistent and even contradictory findings on the nature of this crosstalk. Our objective was to perform an up to date systematic review and meta-analysis of the association between hypothalamic-pituitary-adrenal (HPA) axis and immune system functioning in mood and psychotic disorders.

We searched the Pubmed, Web of Science and Embase databases for studies reporting correlations between one or more HPA- and immune markers (IM) in patients with mood or psychotic disorders. We analyzed unchallenged correlations as well as challenge studies investigating the HPA-immune interaction through dexamethasone (DEX) and/or CRH suppression, HPA-mediated challenge of immune cell proliferation, immune challenges, or psychological stressors. Finally, genetic studies focusing on HPA x immune interrelation were evaluated. For meta-analyzable data, three primary outcome measures were defined for immune functioning, namely the pro-inflammatory index (PII) and anti-inflammatory index (AII) for the molecular IM and a composite cellular immune marker score (CCIM) for the cellular IM. Secondary analyses were performed for the individual molecular and cellular IM. Heterogeneity was evaluated with the I² statistic. Meta-regression analyses were performed to evaluate the impact of potential covariates (publication year, gender, age, symptom severity) on the primary outcome analyses.

93 studies (n = 8226) were included, of which 50 (n = 5649) contained meta-analyzable data. The majority of the included studies (k = 72) investigated major depressive disorder (MDD) patients, nineteen schizophrenia spectrum disorders (SSD) and six bipolar disorder (BD). Under physiological conditions, a poor association was found between cortisol and the PII only in the unmedicated subsample of MDD (k = 8; n = 425; r = .205; z = 2.151; p = .031) and the medicated subsample of SSD (k = 4; n = 152; r = .0.237; z = 2.314; p = .021). No significant correlation was found in MDD between the AII and cortisol (k = 3; n = 1243; r = .005; z = .188; p = .851). Similar results were found for the association between immune cell numbers and cortisol in both MDD (k = 10; n = 773; r = -.005; z = -.113; p = .894) and SSD (k = 4; n = 99; r = .167; z = 1.356; p = .175). A total of 42 studies discussed post-challenge associations between immune alterations and HPA disturbances, of which 12 (n = 389; all MDD) contained meta-analyzable data and 37 entered the systematic review (n = 1783). No post-DEX correlations were found between cortisol and PII (k = 3; n = 105; r = .074; z = .355; p = .722) or CCIM (k = 5; n = 259; r = -.153; z = -1.294; p = .196). However, a significant association was found between post-DEX cortisol/ACTH and PII produced by stimulated blood cells in vitro (k = 3; n = 61; r = .508; z = 4.042; p < .001) as well as for cortisol and CCIM score in MDD after in vitro mitogen stimulation (k = 4; n = 90; r = -.309; z = -2498; p = .012). Following a psychological stressor (k = 6; n = 121), cortisol responses tended to be blunted in all included pathologies, while immune activation was comparable to healthy controls. Genetic studies (k = 7; n = 464) demonstrate altered gene expression of glucocorticoid receptors (GR) in peripheral immune cells in MDD. Heterogeneity over studies tended to be moderate to high.

The main limitations are the heterogeneity of outcome measures (both HPA and IM) and small sample sizes of the included studies. We conclude that, in physiological conditions, associations between HPA-axis and molecular or cellular IM are absent or poor in both MDD and SSD and psychotropic medication may influence this crosstalk differently in both patient groups. Studies using challenge paradigms in MDD populations did reveal differences in the HPA-immune crosstalk. The normally expected decrease in lymphocytes after DEX distribution tended to be less pronounced in MDD, especially in glucocorticoid-insensitive non-suppressors. It is recommended that future studies should be properly powered and assess HPA functioning using multiple cortisol assessments. Challenge studies are probably more useful than baseline biomarker studies and cellular IM are more informative than molecular IM. It is recommended to broadly assess leucocyte function and, when possible, perform subgroup analyses based on HPA- and/or immune function.