The “immune hypothesis” of major depressive disorder (MDD) suggests a strong link between MDD and a dysfunctional immune system, while mounting research data indicate the significant role of several pro-inflammatory pathways in the pathophysiology of the disorder ^[1][2][3][68,69,70]. Accordingly, both immune and inflammatory biomarkers have been extensively studied with respect to their potential for predicting responses to antidepressant treatment. For example, several meta-analyses have repeatedly indicated that levels of tumor necrosis factor-alpha (TNF-α), interleukin-6 (IL-6), IL-1β, brain-derived neurotrophic factor (BDNF), IL-8 and c-reactive-protein (CRP) in peripheral blood serum are reliable biomarkers of antidepressant treatment response in MDD ^[4][5][71,72]. More specifically, higher baseline levels of TNF-α, IL-6, BDNF, IL-1β, and CRP, as well as lower IL-8 levels in the blood, are related to poorer responses to pharmacotherapy in MDD. However, there are also studies that do not confirm these results ^{[4][5][6][7][8][9][10]}[31,37,38,40,71,72,73], while especially in TRD, respective research addressing baseline immune and inflammatory biomarkers for the prediction of treatment response is relatively scarce.

ThWe researchers hhave identified three studies regarding the assessment of treatment response prediction in TRD with respect to IL-1β baseline levels. In an open study of ketamine infusion as monotherapy, in a sample of 16 TRD patients, Yang et al. reported significantly higher IL-1β serum baseline levels, as well as significantly reduced IL-1β levels 230 min and 1 day after ketamine infusion in responders compared to non-responders ^[11][41]. In contrast, in a similar study of 33 TRD patients, Kiraly et al. did not find any correlation between IL-1β baseline levels and treatment response to i.v. ketamine as monotherapy ^[12][47]. Similarly, an open study of add-on electroconvulsive therapy (ECT) in a sample of 29 patients with TRD by Kruse et al., also showed no correlation between IL-1β baseline blood levels and response to treatment ^[13][42], suggesting that IL-1β baseline levels have no prognostic value for treatment response in TRD.

With respect to IL-6, there is a better line of evidence concerning its predictive value, as several studies have reported an association between plasma IL-6 and treatment response in TRD, while a recent systematic review by Yang et al. concluded that higher baseline levels of IL-6 predicted better responses in patients with TRD, although several contradictory results were reported ^[14][74]. In particular, a study by Chen et al. assessed the add-on use of intravenous ketamine in TRD patients and pointed out a connection between higher baseline levels of IL-6 and a better response to therapy in the group subjected to 0.5 mg/kg ketamine infusion but different results in the groups receiving 0.2 mg/kg ketamine and a placebo ^[15][43]. Similarly, Yang et al. also noted significantly higher baseline levels of IL-6 in TRD responders to 0.5 mg/kg ketamine infusion monotherapy compared to the group of non-responders ^[11][41], while in the ECT add-on study by Kruse et al., higher levels of IL-6 prior to treatment also predicted lower scores in the Montgomery–Asberg Depression Rating Scale following a course of treatment in females, but not in men ^[13][42]. In addition, in an add-on study of metyrapone on SSRI treatment in 63 TRD patients, Strawbridge et al. indicated a correlation between higher baseline levels of IL-6 and poorer responses to treatment ^[16][49].

On the other hand, Kiraly et al. didn’t discover any prognostic association between IL-6 levels and responses to treatment with 0.5 mg/kg ketamine infusion monotherapy in TRD patients ^[12][47]. Similarly, Kagawa et al. found no correlation between baseline IL-6 levels and clinical responses in augmentation therapy with lamotrigine in TRD patients, nor between baseline IL-6 levels and improvements in MADRS score ^[17][44]. Allen et al. also failed to find any association between baseline IL-6 levels and responses to treatment with 0.5 mg/kg ketamine infusion or ECT monotherapy in TRD patients, although a correlation between the greater decrease of depressive symptoms and higher baseline levels of IL-6 was found only in the first 24 hours post ketamine infusion ^[18][45]. Finally, Yoshimura et al. concluded that baseline levels of IL-6 in the blood had no prognostic value with respect to treatment response in patients resistant to therapy with SSRIs/SNRIs ^[19][48].

With respect to the predictive significance of baseline IL-8 levels for treatment responses in TRD, thwe researchers ccould identify only three studies; of these, one study concluded that higher baseline levels of IL-8 were related to less serious depressive symptoms after the third infusion of 0.5 mg/kg ketamine in patients with TRD but not for other time points during treatment ^[18][45]. The two other studies could not report any prognostic correlation between baseline IL-8 levels and responses to treatment under 0.5 mg/kg ketamine infusion monotherapy ^[12][47] or ECT add-on treatment ^[13][42].

Only three studies were found that investigated the prognostic value of baseline IL-10 levels on treatment responses in TRD patients. Allen et al. ^[18][45] and Kiraly et al. ^[12][47] both could not show any correlation between baseline IL-10 and responses to treatment with 0.5 mg/kg ketamine infusion monotherapy, while Strawbridge et al. also reported no prognostic value of baseline IL-10 on treatment responses in an add-on study of metyrapone on SSRI treatment in 63 TRD patients ^[16][49].

The two studies by Allen et al. ^[18][45] and Kiraly et al. ^[12][47] were the only studies found to investigate an association between baseline INF-γ levels and treatment response to 0.5 mg/kg ketamine infusion monotherapy in TRD patients, and both failed to find any significant prognostic correlations, which is also supported by missing supportive data for INF-γ in the systematic review of Yang et al. ^[14][74].

Several studies have assessed the prognostic value of baseline TNF-a levels in blood and treatment responses in TRD; however, most of them failed to report any prognostic correlations. For example, Chen et al. found that baseline TNF-a levels were not associated with treatment outcomes for both 0.5 and 0.2 mg/kg i.v. doses of add-on ketamine treatment in 47 TRD patients ^[15][43]. Similarly, the two studies by Yang et al. ^[11][41] and Kiraly et al. ^[12][47] on the treatment response of TRD patients to 0.5 mg/kg ketamine infusion monotherapy did not find differences in baseline TNF-a levels between responders and non-responders to 0.5 mg/kg ketamine infusion monotherapy, and could not report any prognostic correlation between baseline TNF-a and treatment response. In addition, no prognostic correlation between baseline TNF-a levels and treatment response was found in studies assessing the response to ECT add-on treatment ^[13][42], SSRI/SNRI add-on therapy ^[19][48], and metyrapone add-on treatment to SSRIs ^[16][49] in patients with TRD. The systematic review oby Yang et al. also did not find any supporting data for the prognostic value of TNF-a in clinical response trials in TRD ^[14][74]. However, the experimental trial by Raison et al., assessing the treatment response of TRD patients with mild resistance to treatment with the functional TNF-a antagonist infliximab, was the only study that could show that higher baseline TNF-a levels were associated with a better treatment response ^[20][50].

A large number of studies have assessed the predictive value of baseline peripheral CRP levels within TRD treatment response, and the recent systematic review by Yang et al. actually supports a clinically significant prognostic association between higher CRP reference levels and a better treatment response ^[14][74]. For example, the study by Raison et al. found that baseline levels of hsCRP > 5mg/L predicted a larger decrease in Hamilton (HAMD)-17 scale scores in 30 TRD patients with mild resistance to treatment receiving infliximab treatment ^[20][50], while Papakostas et al. also reported a larger decrease of HAMD-17 scores in SSRI-resistant depressive patients with higher levels of hsCRP receiving adjunctive therapy with L-methylfolate ^[21][51]. However, some studies have shown conflicting or negative results. For example, no prognostic correlations between baseline peripheral CRP levels and treatment response were found in studies assessing the response of TRD patients to both 0.5 and 0.2 mg/kg i.v. doses of add-on ketamine ^[15][43] or metyrapone treatment ^[16][49]. Interestingly, in their ECT add-on study, Kruse et al. pointed out that baseline CRP levels correlated significantly with final MADRS scores at the end of treatment in the female TRD sample only, while no correlation could be found with respect to the changes in the MADRS score in the total sample of patients ^[13][42].

Growth factors, like brain-derived neurotrophic factor (BDNF) have been often investigated as response biomarkers of depression treatment and especially implicated in the rapid mechanism of action of ketamine ^[22][23][75,76]. In addition, BDNF is considered to play an important role in the neuroimmune and inflammatory pathophysiology of MDD ^[24][25][77,78]. However, to date, only a few studies have managed to show some correlation between BDNF and treatment response in TRD.

For example, the ECT add-on study of Piccini et al., in 18 patients with TRD, reported lower baseline BDNF levels in patients vs. the control subjects, an increase in BDNF after treatment and, most importantly, higher BDNF baseline levels in responders than in non-responders ^[26][52]. Similarly, Haile et al. investigated 22 patients with TRD and found that i.v. ketamine monotherapy resulted in a higher BDNF increase in the serum of the responders than in the non-responders, as well as a negative correlation between MADRS scores and BDNF levels ^[27][54]. However, there was no significant prognostic correlation between BDNF baseline levels and responses to treatment with ketamine ^[27][54]. On the other hand, a study of treatment responses to add-on riluzole or a placebo in 55 TRD patients by Wilkinson et al. reported lower baseline levels of BDNF in responders to riluzole or the placebo compared to non-responders, although the statistical significance remained within the trend level ^[28][53].

Nevertheless, most studies assessing baseline peripheral BDNF could not show any correlation with respect to treatment response. For instance, in an add-on ECT response study of 74 patients with TRD, Maffioletti et al. could not show any difference between the baseline BDNF levels in responders and in non-responders ^[29][55]. Similarly, Huang et al. also showed no correlation between the baseline levels of BDNF in serum and responses to treatment in a comparison study of ECT vs. anesthesia with ketamine and propophol in 30 TRD patients ^[30][56], as did Allen et al. in a monotherapy study with either ECT or 0.5 mg/kg i.v. ketamine infusion in a group of 35 patients with TRD ^[31][46]. Likewise, Kagawa et al. reported only minor variations of BDNF baseline levels between responders and non-responders in a study of adjunctive treatment with lamotrigine in 46 TRD patients, while there was no prognostic correlation found between baseline levels of BDNF and responses to treatment, nor any important association between BDNF-level changes and the improvement of MADRS scores ^[17][44].

There have been few studies that managed to indicate a correlation between responses to treatment of TRD and other baseline growth factors. For example, Pisoni et al. found that among all growth factors studied, only vascular endothelial growth factor-C (VEGF-C) showed any prognostic correlation between the treatment response of a group of 36 patients and TRD. In this case, lower baseline levels before treatment were related to better responses to an add-on antidepressant treatment ^[32][57]. In another response study of add-on repeated transcranial magnetic stimulation (rTMS) in 15 patients with TRD by Fuduka et al., responders to the treatment showed higher initial concentrations of VEGF, while the percentage change in VEGF levels after treatment showed a statistically-significant correlation with the changes in psychometric scores of depressive symptomology ^[33][58]. Finally, in an i.v. 0.5 mg/kg ketamine infusion monotherapy study of 33 patients with TRD, Kiraly et al. showed that, among a large number of assessed biomarkers, only baseline serum levels of fibroblast growth factor 2 (FGF-2) were associated with treatment response, where low initial levels predicted better response to treatment ^[12][47].

MDD is considered a stress-related disorder with a unique pathophysiological neuroendocrine profile, presenting distinct changes in activity and reactivity of the HPA axis ^[34][35][2,79]. The most consistent findings include correlations between: the hyperactivity of the HPA axis and increased cortisol (CORT) levels, a higher corticotropin-releasing hormone (CRH) drive and higher adrenocorticotropic hormone (ACTH) and vasopressine (AVP) levels, and a weak cortisol awakening response (CAR) and the reduced sensitivity of glycocorticoid receptors ^[36][80]. It has also been shown that effective antidepressive treatment with reductions in depressive psychopathology is often related to a consequent normalization in the HPA-axis (re-)activity ^[36][80]. Apart from the baseline and diurnal levels of several HPA axis hormones (CRH, AVP, ACTH, CORT, and dehydroepiandrosterone—DHEA), neuroendocrine suppression or stimulation tests (e.g., dexamethasone, metyrapone) have often been used to study dynamic endocrine levels with respect to their possible prognostic significance in antidepressant response ^[36][80]. Nevertheless, the studies that investigated the baseline and diurnal activity, and dynamic responsiveness of the HPA axis in TRD, are proportionally scarce.

In one of the three available studies, Markopoulou et al. measured baseline CORT and DHEA and their ratio in 28 patients with TRD, noting their association with treatment responses to add-on pharmacological treatment ^[37][59]. In this researchtudy, while CORT levels were lower after the treatment, there was no correlation between baseline and post-treatment CORT levels and treatment outcome. On the contrary, the responders to treatment had significantly lower DHEA and showed a higher CORT/DHEA ratio both at admission and discharge compared to the non-responders, suggesting that, although remaining stable across treatment, the CORT/DHEA ratio could represent a prognostic biomarker of response to TRD. In another study, Dinan et al. examined therapeutic responses to add-on therapy using dexamethasone in 10 patients with SSRIs resistance and showed that higher baseline CORT levels predicted treatment response ^[38][60]. On the other hand, in a study of treatment responses to add-on sleep deprivation or sleep phase shift in 21 patients with TRD (unipolar and bipolar), Kurczewska et al. noticed that baseline CORT levels were significantly lower in responders than in non-responders ^[39][61].

Adipokines are cytokines, with hormonal action secreted by the lipid tissue (e.g., adiponectin, leptin, and resistin), are considered to participate in the pathophysiological pathways connecting obesity with cardiovascular diseases. In a 0.5 mg/kg i.v. ketamine infusion add-on study in 8 patients with unipolar/bipolar TRD, Machado-Vieria et al. showed that apart from the prognostic value of BMI, regarding the response to treatment with ketamine, lower baseline concentrations of adiponectin in the serum could predict antidepressant responses to ketamine ^[40][62].

Blood lipidemic factors have been implicated in MDD pathophysiology and especially in intra- and inter-neuronal functioning, and have been studied as biomarkers in MDD ^[41][81]. In TRD, only two studies have been found that assess blood lipidemic factors in relation to treatment response. In an add-on study with infliximab in 26 patients with TRD, Bekhbat et al. found that baseline levels of cholesterol, LDL, and non-HDL were higher in responders and also showed significant decreases (during treatment) in those patients with higher baseline CRP ^[42][63]. On the contrary, in a study of 92 TRD patients, Papakostas et al. reported that baseline levels of cholesterol > 200 mg/dL predicted a worse response to monotherapy with nortriptyline ^[43][64].