Participants
In accordance with the Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (DSM-5) 34, a senior psychiatrist at Alhakiem Hospital in Najaf, Iraq, identified and recruited 64 individuals with FE-MDD and 32 healthy controls. All patients diagnosed with MDD were observed to be in the acute phase of the disease, with no indications of complete or partial remission. It should be noted that all FE-MDD patients met the diagnostic criteria for MDMD as detailed by Maes, Vasupanrajit, et al. (2023) 15. In addition, the senior psychiatrist recruited healthy controls from the same geographical area to serve as the control group. This group included hospital staff and patients' acquaintances. All participants were enlisted between October 2021 and March 2022.
Exclusions were made for those with chronic liver or kidney disease, women who were pregnant or nursing, and those with a history of multiple sclerosis, Parkinson's disease, stroke, or Alzheimer's disease. Similarly, psoriasis, rheumatoid arthritis, inflammatory bowel disease, cancer, type 1 diabetes, and scleroderma exhibited the same pattern. In addition, we excluded participants who had experienced an acute COVID-19 infection, severe or critical COVID-19 disease, Long COVID, or a COVID-19 infection within the previous six months. Our study did not include individuals taking immunosuppressive or immunomodulatory medications, nor did it include those taking therapeutic doses of antioxidants or omega-3 supplements (last three months).
The process of meticulously selecting the control group was accorded significant consideration. Excluded from the control group were participants with a documented lifetime history of clinical depression or dysthymia, a family history of depression, mania, psychosis, or substance use disorder, or a history of suicide. Dysthymia (excluding cases of double depression), schizophrenia, schizoaffective disorder, bipolar disorder, autism spectrum disorders, substance use disorders (excluding nicotine dependence), post-traumatic stress disorder, psycho-organic disorders, generalized anxiety disorder, and obsessive-compulsive disorder were excluded from the study.
Before participating in the study, all participants, or, if applicable, their parents or legal custodians, provided informed consent in writing. Document No. 18/2021 indicates that the Ethics Committee of the College of Medical Technology at the Islamic University of Najaf in Iraq has approved this investigation. The research was conducted in accordance with both Iraqi and international ethical and privacy regulations. In prominent documents such as the Declaration of Helsinki by the World Medical Association, the Belmont Report, the CIOMS Guideline, and the International Conference on Harmonization of Good Clinical Practice, a variety of non-exhaustive principles are outlined. Our organization's institutional review board (IRB) is committed to upholding the highest standards of quality, ensuring precise compliance with the International Guideline for the Conduct of Safe Human Research (ICH-GCP).
Clinical assessments
Using the Hamilton Depression Rating Scale (HAMD) 5, the Hamilton Anxiety Rating Scale (HAMA) 6, assessments of suicidal behaviors, and evaluations of disease recurrence scores 3, a diagnosis of MDMD can be established using machine learning techniques. Alternatively, when both the Hamilton Depression Rating Scale (HAMD) and the Hamilton Anxiety Rating Scale (HAMA) exceed twenty-two points, it may also be diagnosed. As such, the current study included only patients with FE-MDMD. The physical, mental, and behavioral health of the participants was evaluated by a senior psychiatrist who used a systematic interview and standardized procedures. The senior psychiatrist collected sociodemographic, clinical, and psychological information through semi-structured interviews.
The same professional utilized the HAMD and the HAMA to evaluate the severity of depression and anxiety, respectively. In the current study, all physiosomatic symptoms of the HAMD and HAMA were excluded to compute pure depression and pure anxiety scores, respectively. The former concept was conceptualized as a collection of symptoms including depressed mood, feelings of guilt, suicidal ideation, and decreased interest. The score for pure anxiety was determined by adding the scores for anxious mood, tension, fears, and anxious behavior observed during the interview. The physiosomatic symptom score was calculated as a z unit-based composite score based on the sum of the z scores of HAMD and HAMA physiosomatic symptoms, namely anxiety somatic, somatic gastrointestinal, and genitourinary symptoms, hypochondriasis, somatic sensory, cardiovascular, gastrointestinal, genitourinary, autonomic, and respiratory symptoms. The pure FF symptoms were calculated as the sum of the FF items after exclusion of non-physiosomatic symptoms: muscle pain, muscle tension, fatigue, autonomic, gastro-intestinal symptoms, headache, and a flu-like malaise (all FF scale items after exclusion of all non-physiosomatic symptoms, including cognitive deficits, and sadness). Gastro-intestinal symptoms (GIS) were conceptualized as a z unit-based composite score based on symptoms of the HAMA and FF scores, namely somatic gastrointestinal and GIS symptoms. Insomnia was conceptualized as a z value-based composite, namely sum of insomnia (HAMA item), insomnia early + insomnia middle + insomnia late (HAMD items) + sleep disorders (FF). Melancholia was conceptualized as a composite based on the z scores of insomnia late, psychomotor retardation, psychomotor agitation, loss of weight, and diurnal variation. This study utilized two items of the Columbia Suicide Severity Rating Scale (C-SSRS) to evaluate suicidal behaviors: the quantification of suicide attempts within the previous year and the assessment of the frequency of suicidal ideation within the previous three months 35. A composite score of suicidal behaviors (SB) was calculated by adding the weighted z scores for the HAMD suicide item, the number of suicide attempts, and the frequency of suicidal ideation.
The Adverse Childhood Experiences (ACEs) Questionnaire 36 was used to measure the extent of adverse childhood experiences. There are a total of twenty-eight items on the scale, which comprise the scoring of ten distinct domains. These include (1) mental trauma, (2) physical trauma, (3) sexual abuse, (4) mental neglect, (5) physical neglect, (6) witnessing domestic violence involving the mother, (7) presence of a family member with drug abuse issues, (8) presence of a family member with depression or mental illness, (9) experiencing the loss of a parent due to separation, death, or divorce, and (10) having an incarcerated family member. Various ACE scores were calculated, including the total sum of all ACEs (termed total ACE), the sum of physical trauma, mental neglect, and family member with substance abuse (termed ACE247), and the sum of ACE247 and a family member in prison (termed ACE24710). In addition, the negative life events (NLEs) scale was used to assess the incidence of NLEs in the preceding year 37. For the purposes of this investigation, the following factors were considered: a serious accident, the death of a family member or close friend, divorce or separation, unemployment, job loss, alcohol-related issues, drug-related issues, witnessing physical altercations or assaults, experiences of abuse or violent crime, encounters with law enforcement difficulties, problem gambling, familial incarceration, overcrowding in the household, and instances of discrimination. Consequently, we performed calculations to ascertain the combined effects of ACE and the occurrence of one or more NLEs, which we referred to as ACE + NLEs. Body mass index (BMI) was calculated by dividing the participants' weight in kilograms by the square of their height in meters. Using the diagnostic criteria specified in the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5), tobacco use disorder was evaluated.
Biochemical assays.
In this investigation, a disposable syringe and serum tubes were used to obtain a 5 mL venous blood sample from each participant while they were fasting. Blood samples were collected between 8:00 and 11:00 in the morning. The serum and blood cells were effectively separated following centrifugation at 35,000 revolutions per minute (rpm). The serum was then transferred into small Eppendorf containers and stored at -80°C until its use in the biomarker assays was deemed necessary. Bio-Plex Multiplex Immunoassay kits (Bio-Rad Laboratories Inc., Hercules, USA) were used to measure the levels of forty-eight cytokines/chemokines/growth factors in the serum of all participants. For measuring the concentrations of these proteins in blood serum, we employed a fluorescence-based method. The researchers measured the fluorescence intensity (FI) of each protein and used the values obtained after subtracting the blank value 18,38. The Electronic Supplementary File (ESF) comprises Table 1, which provides an exhaustive listing of the analytes identified by our research. In addition, each analyte's respective gene ID and alternate names are listed in this table. The coefficients of variation (CV) for all analytes in the assay were less than 11.0%. To determine the concentrations of the analytes, we employed the manufacturer-supplied standards. Subsequently, the out-of-range (OOR) concentration rate was calculated, i.e., the proportion of concentrations that fell below the minimum detectable level. In over 80% of cases, cytokines/chemokines/growth factors with concentrations below the lowest out-of-range (OOR) value were excluded from the statistical analysis. The aforementioned cytokines, including IFN-α2, IL-3, IL-7, and IL-12p40, were therefore excluded from our study. As dummy variables, we accounted for analytes with quantifiable prevalence rates between 20% and 40% when calculating our composite scores. As a result, diverse immune profiles were computed, as described in Table 2 of the ESF. These profiles covered all analytes with the exception of those with detectable concentrations below 20% 18,38,39. Table 2 of the ESF lists the variables used to generate M1, M2, M1/M2 (z M1 - z M2), Th1, Th2, Th1/Th2 (z Th-1 - z Th-2), Th-17, IRS, CIRS, IRS/CIRS (z IRS - z CIRS), and Immu-NT profiles. In addition, the individual cytokines, chemokines, and growth factors were analyzed to ascertain whether there were significant differences in immune profiles between the study groups.
Table 1
Sociodemographic and clinical data in healthy controls and patients with major dysmood disorder (MDMD) with very severe physiosomatic symptoms and immune profiles (Severe MDMD) and MDMD patients with milder physiosomatic and immune scores (Milder MDMD).
Variables | HC a n = 32 | Milder MDMD b n = 29 | Severe MDMD c n = 35 | F/X2 | df | p |
Age (years) | 32.0 ± 7.6 | 33.5 ± 9.2 | 29.8 ± 8.3 | 1.64 | 2/93 | 0.200 |
Sex (M/F) | 23/9 | 18/11 | 23/12 | 0.68 | 2 | 0.711 |
Educated (No/Yes) | 11/21 | 9/20 | 10/25 | 0.26 | 2 | 0.877 |
Single/married | 15/17 | 17/12 | 22/13 | 1.83 | 2 | 0.401 |
BMI (kg/m2) | 26.19 ± 4.11 | 24.74 ± 3.45 | 34.24 ± 3.77 | 2.33 | 2/93 | 0.103 |
Smoking (No/Yes) | 14/18 | 22/7 | 28/7 | 11.47 | 2 | 0.003 |
Previous COVID (No/yes) | 20/12 | 18/11 | 22/13 | 0.00 | 2 | 0.998 |
Duration of illness (months) | - | 2.26 ± 2.78 | 2.58 ± 2.84 | 0.21 | 1/62 | 0.649 |
Pure depression | 0.47 ± 0.76 b,c | 10.62 ± 1.54 a | 10.82 ± 1.48 a | 654.52 | 2/93 | < 0.001 |
Pure anxiety | 0.87 ± 0.87 b,c | 9.17 ± 1.28 a | 9.20 ± 1.47 a | 478.54 | 2/93 | < 0.001 |
Pure FF | 1.16 ± 0.92 b,c | 8.90 ± 4.03 a | 9.31 ± 3.06 a | 79.06 | 2/93 | < 0.001 |
Physiosomatic symptoms | -1.275 ± 0.325 b,c | 0.374 ± 0.633 a,c | 0.673 ± 0.558 a,b | 132.46 | 2/93 | < 0.001 |
Melancholia | -1.220 ± 0.358 b,c | 0.238 ± 0.541 a,c | 0.658 ± 0.690 a,b | 104.21 | 2/93 | < 0.001 |
Insomnia | -1.225 ± 0.299 b,c | 0.465 ± 0.548 a | 0.613 ± 0.752 a | 106.30 | 2/93 | < 0.001 |
GIS symptoms | -1.030 ± 0.294 b,c | 0.276 ± 0.644 a,c | 0.746 ± 0.762 a,b | 76.04 | 2/93 | < 0.001 |
Autonomic symptoms | -0.300 ± 0.523 c | -0.136 ± 0.889 | 0.280 ± 1.302 a | 3.18 | 2/93 | < 0.001 |
Total ACEs | 0.41 ± 0.499 b,c | 1.52 ± 0.986 a,c | 2.00 ± 1.138 a,b | 25.93 | 2/93 | < 0.001 |
ACE + NLEs | -0.994 ± 0.552 b,c | 0.343 ± 10.705 a | 0.607 ± 0.954 a | 41.00 | 2/93 | < 0.001 |
Fluoxetine (No/yes) | - | 18/11 | 23/12 | 0.09 | 1 | 0.762 |
Amitryptiline (No/yes) | - | 27/2 | 30/5 | FEPT | - | 0.442 |
Escitalopram (No/yes) | - | 27/2 | 31/4 | FEPT | - | 0.681 |
Olanzapine (No/yes) | - | 27/2 | 32/3 | FEPT | - | 1.0 |
Mirtazapine (No/yes) | - | 23/6 | 33/2 | FEFT | - | 0.127 |
Drug naïve (No/yes) | - | 19/10 | 19/16 | 0.83 | 2 | 0.362 |
PC_immune + phenome | -1.347 ± 0.301 b,c | 0.437 ± 0.129 a,c | 0.869 ± 0.218 a,b | 849.29 | 2/93 | < 0.001 |
All results are shown as mean ± SD. F: results of analysis of variance, X2: results of analysis of contingency analysis; FEPT: Fisher’s exact probability test, a,b,c: pairwise comparisons among group means at p = 0.05. |
FF: Fibro-Fatigue, GIS: gastro-intestinal symptoms, ACE: adverse childhood experiences, NLE: negative life events, PC_immune + phenome: a principal component extracted from 9 immune variables and 6 phenome symptom domains. |
Table 2
Immune profiles in healthy controls (HC) and patients with major dysmood disorder (MDMD) with very severe physiosomatic symptoms and immune profiles (Severe MDMD) and MDMD patients with milder physiosomatic and immune scores (Milder MDMD).
Variables (z scores) | HC n = 32 | Milder MDMD n = 29 | Severe MDMD n = 35 | F/X2 | df | P |
Classical macrophage M1 | -0.604 ± 0.918 b,c | 0.005 ± 0.902 a,c | 0.548 ± 0.839 a,b | 14.16 | 2/93 | < 0.001 |
Alternative macrophage M2 | -0.479 ± 0.851 b,c | 0.043 ± 0.914 a | 0.402 ± 1.030 a | 7.42 | 2/93 | 0.001 |
zM1 – zM2 | 0.124 ± 0.724 | -0.039 ± 0.489 | 0.146 ± 0.581 | 1.73 | 2/93 | 0.184 |
Thelper (Th)-1 | -0.592 ± 1.086 b,c | 0.017 ± 0.695 a,c | 0.527 ± 0.841 a,b | 13.17 | 2/93 | < 0.001 |
Th-2 | -0.398 ± 0.975 c | -0.150 ± 0.844 c | 0.489 ± 0.961 a,b | 8.12 | 2/93 | < 0.001 |
zTh-1 – zTh-2 | -0.194 ± 0.605 b | 0.168 ± 0.484 a | 0.0380 ± 0.454 | 3.86 | 2/93 | 0.025 |
Th-17 | -0.392 ± 1.305 c | -0.098 ± 0.700 c | 0.440 ± 0.696 a,b | 6.71 | 2/93 | 0.002 |
IRS | -0.632 ± 1.051 b,c | 0.002 ± 0.679 a,c | 0.577 ± 0.829 a,b | 16.10 | 2/93 | < 0.001 |
CIRS | -0.564 ± 0.933 b,c | -0.055 ± 0.629 a,c | 0.561 ± 1.023 a,b | 13.43 | 2/93 | < 0.001 |
zIRS - zCIRS | -0.163 ± 1.002 | 0.136 ± 0.977 | 0.037 ± 1.025 | 0.71 | 2/93 | 0.494 |
Immu-NT | -0.575 ± 1.108 b,c | -0.043 ± 0.665 a,c | 0.560 ± 0.820 a,b | 13.71 | 2/93 | < 0.001 |
All results are shown as mean ± SD. F: results of analysis of variance. |
IRS: immune-inflammatory response system, CIRS: compensatory immunoregulatory system. Immu-NT: immune-associated neurotoxicity. |
Statistics
This study's statistical analyses were performed with IBM SPSS 29, Windows version. To compare continuous variables between study groups, statistical tests such as analysis of variance (ANOVA) and the Kruskal-Wallis test were utilized. In contrast, nominal variable comparisons were performed using contingency table analysis, specifically the Chi-square test. In addition, Pearson's and point-biserial correlation coefficients were used to analyze the relationships between scale variables and binary variables. Due to the observed interconnections between cytokines, chemokines, and growth factors within the immune (cytokines and chemokines) and growth factor networks, it was decided not to implement false discovery rate (FDR) p-value correction 38,40. The researchers used manual multiple regression analysis to investigate the impact of adverse childhood experiences (ACEs), negative life events (NLEs), and additional demographic factors on the immune profiles. In a similar manner, these analyses examined the influence of different predictor variables, namely immune profiles, adverse childhood experiences (ACEs), and negative life events (NLEs), on the manifestations of depression. In addition, automatic forward stepwise regressions were employed. A significance level of p = 0.05 was used to determine the inclusion of variables, and a significance level of p = 0.06 was used to determine their exclusion. This method made it easier to determine which variables should be included in the final regression model and which ones should be excluded. For each variable included in the final regression models, the standardized coefficients, t-statistics, and exact p-values were calculated. In addition, we determined the F statistics, their respective p-values, and the effect magnitude using the partial Eta squared. Using appropriate statistical measures, the presence of multicollinearity and collinearity in the data was extensively examined. A tolerance limit of 0.25 and a threshold for the variance inflation factor of four were used. In addition, for the purpose of this analysis, we utilized the condition index and variance proportions derived from the collinearity diagnostics table. Using the White test and a modified variation of the Breusch-Pagan test, heteroskedasticity was identified. All the preceding analyses employed two-tailed tests. A significance level of 0.05 or lower was regarded as statistically significant. As required, we utilized transformations such as logarithmic, square-root, rank-based inversed normal (RINT), and a Winsorization technique to obtain a normal distribution for our data indicators.
Partial least squares (PLS) analysis was used to investigate the causal relationships between ACEs and NLEs, immunological profiles, and the physiosomatic symptoms of depression. The output variable was a latent vector derived from the diverse symptom domains, and ACE + NLEs and immune profiles as explanatory variables. In addition, the immune profiles were allowed to mediate the effects of ACE + NLEs on the clinical assessment scores. Only when both the external and internal models met the predetermined quality criteria was a comprehensive partial least squares (PLS) analysis conducted. These quality criteria are: a) Confirmatory tetrad analysis (CTA) verifies that the latent vectors derived from the indicators have not been incorrectly specified as reflective models. b) The blindfolding procedure reveals that the cross-validated redundancy of the construct is adequate. c) The latent vectors exhibit strong construct and convergence validity, as indicated by composite reliability values greater than 0.80, Cronbach's alpha values greater than 0.70, and average variance extracted (AVE) values greater than 0.5. d) At a significance level of p 0.001, all loadings on the extracted latent vectors exceed 0.65. e) The model fit is regarded satisfactory if the standardized root squared residual (SRMR) is less than 0.08. Consequently, a thorough pathway analysis is conducted using PLS-Structural Equation Modeling (PLS-SEM). In the analysis, SmartPLS software and 5,000 bootstrap samples were utilized. Path coefficients and their respective p-values were calculated. In addition, specific indirect effects, total indirect effects (mediated effects), and total effects were calculated if the model quality data met the specified conditions. The estimated minimum sample size is 103 based on a power analysis (conducted with G*Power 3.1.9.4) using a linear multiple regression analysis with an estimated effect size of 0.111 (corresponding to approximately 10% of the variance explained), a significance level (alpha) of 0.05, and statistical power of 0.8, with three covariates.