Introduction

Depression is a prevalent and high-morbidity disorder, impacting millions globally each year1. While antidepressants are the cornerstone of treatment, achieving remission remains a challenge; indeed, only about one-third of patients reach remission in acute treatment trials2,3. This highlights the urgent need for strategies that can enhance remission rates. Although the development of groundbreaking new antidepressants would be ideal, it remains an unattainable goal in the near term. A more practical approach for clinicians involves identifying predictors of remission at initial diagnosis to tailor individualized treatment plans. To date, a variety of predictors have been explored across biological, psychological, and social dimensions2,4,5.

From a biological perspective, extensive evidence indicates that activation of the inflammatory response system contributes to the pathophysiology and chronicity of depressive disorders, with elevated peripheral cytokines repeatedly linked to greater symptom burden, cognitive dysfunction, and reduced likelihood of remission6. Inflammation has increasingly been conceptualized as a transdiagnostic mechanism contributing to treatment-resistant or difficult-to-treat depression phenotypes, in line with contemporary frameworks highlighting the multifactorial determinants of antidepressant nonresponse7. Among pro-inflammatory cytokines, interleukin-1 beta (IL-1β) is of particular relevance because it serves as an upstream initiator of the inflammatory cascade, triggering downstream production of other mediators such as IL-6, and exerting potent effects on neuroendocrine, monoaminergic, and neuroplastic pathways implicated in antidepressant resistance8. Mechanistic work has shown that IL-1β can alter synaptic transmission, inhibit hippocampal neurogenesis, activate microglia, and dysregulate the hypothalamic–pituitary–adrenal (HPA) axis—all of which are pathways linked to poor treatment outcomes.

Although IL-6 has historically been the most widely studied cytokine in depression, recent translational evidence suggests that IL-1β may play a more proximal regulatory role in driving affective symptomatology, making it a biologically compelling candidate for prediction studies. Nevertheless, findings regarding IL-1β and antidepressant treatment response remain inconsistent. Some investigations report that elevated baseline IL-1β predicts poor remission or resistance to pharmacotherapy9, whereas other studies have failed to observe significant associations10, and a few have suggested an inverse or nonsignificant relationship depending on sample characteristics and analytic approach11. These inconsistencies highlight the need for more refined approaches that move beyond single-biomarker models and consider potential moderators—such as psychological traits—that may clarify when and for whom IL-1β is most clinically relevant.

Personality-related factors have also been implicated in shaping individual differences in antidepressant treatment response12. The Five-Factor Model (FFM) of personality, encompassing Extraversion, Agreeableness, Conscientiousness, Neuroticism, and Openness, provides a comprehensive and widely accepted framework for characterizing personality structure13. Among these traits, Neuroticism—marked by heightened emotional reactivity, stress sensitivity, and a tendency toward negative affect—has been consistently associated with poorer treatment outcomes, as it influences how individuals interpret and cope with stress during pharmacotherapy14,15. Although the present study employed the brief version of FFM model due to its established validity, several other validated instruments are frequently used to assess Neuroticism in clinical populations, including the NEO Personality Inventory–Revised (NEO-PI-R) and the NEO Five-Factor Inventory (NEO-FFI)16, and the Eysenck Personality Questionnaire–Neuroticism scale (EPQ-N)17. These instruments provide more granular assessments of the trait, though their length may limit feasibility in large naturalistic cohorts. The predictive roles of the remaining four FFM traits in antidepressant response remain less clearly delineated18.

Neuroticism has also been reported to modulate the effects of biological markers on antidepressant response. Observations suggest that individuals with high Neuroticism may exhibit poorer responses to treatment, influenced by their polygenic risk scores19. Moreover, high Neuroticism was associated with higher levels of IL-6, a proinflammatory cytokine similar to IL-1β20. Based on these observations, it can be hypothesized that the impact of IL-1β on depression treatment outcomes may vary with levels of Neuroticism. However, research specifically addressing this interaction remains sparse.

In response to these research gaps, our study employs a prospective cohort of Korean patients with depressive disorders undergoing stepwise antidepressant treatment. We aim to explore the combined interactive effects of baseline serum IL-1β and Neuroticism levels on treatment outcomes over a 12-week period.

Material and methods

Study framework

This research is conducted under the MAKE Biomarker Discovery for Enhancing Antidepressant Treatment Effect and Response (MAKE BETTER) initiative. The foundational methodology for this investigation was delineated in earlier studies21. In this segment of the project, we initiated the collection of both socio-demographic and clinical information, which includes personality assessments, in addition to a comprehensive set of blood biomarkers. The study’s follow-up routine was designed to assess participants at three-week intervals for the initial 12 weeks using a stepwise, naturalistic approach to antidepressant therapy. A detailed timeline of study procedures is provided in Supplementary Figure S1. All study protocols conformed to the ethical standards sanctioned by the Institutional Review Board of Chonnam National University Hospital (CNUH 2012–2014), in alignment with prevailing ethical requirements.

Participants

The study recruited individuals diagnosed with depressive disorders according to the Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition (DSM-IV)22. Enrollment was conducted at the outpatient psychiatric department of Chonnam National University Hospital (CNUH) from March 2012 to April 2017. The research utilized broad inclusion criteria and limited exclusion criteria to ensure a diverse range of participants. Inclusion criteria were: i) aged older than 7 years; ii) diagnosed with major depressive disorder, dysthymic disorder, or depressive disorder not otherwise specified (NOS) by applying Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM-IV-TR) criteria23; iii) Hamilton Depression Rating Scale (HAMD)24 score ≥ 14; iv) able to complete questionnaires, understand the objective of the study, and sign the informed consent form. Exclusion criteria were: i) an unstable or uncontrolled medical condition; ii) unable to complete the psychiatric assessment or comply with the medication regimen, due to a severe physical illness; iii) current or lifetime DSM-IV diagnosis of bipolar disorder, schizophrenia, schizoaffective disorder, schizophreniform disorder, psychotic disorder NOS, or other psychotic disorder; iv) history of organic psychosis, dementia, epilepsy, or seizure disorder; v) history of anticonvulsant treatment; vi) hospitalization for any psychiatric diagnosis apart from depressive disorder (e.g., alcohol/drug dependence); vii) electroconvulsive therapy received for the current depressive episode; viii) pregnant or breastfeeding. Prior to participation, all individuals provided informed consent and received detailed documentation about the study.

Serum interleukin-1 beta (IL-1β)

Participants were required to fast from the night before, although drinking water was permitted, to ready themselves for blood sampling. To ensure minimal stress and stable physiological conditions, they were also requested to rest quietly for 25 to 45 min before venipuncture was performed. The measurement of sIL-1β concentrations was carried out using the Human High Sensitivity T Cell Magnetic Bead Panel provided by EMD Millipore (Billerica, USA), with the analysis conducted at the Global Clinical Central Lab in Yongin, Korea. The analytical performance of the IL-1β assay met established quality criteria. The lower limit of detection for IL-1β was approximately 0.3 pg/mL, with a lower limit of quantification of about 1.0 pg/mL. The assay demonstrated an intra-assay coefficient of variation below 10% and an inter-assay coefficient of variation below 15%, consistent with the manufacturer’s specifications for the Millipore high-sensitivity magnetic bead panel. For analytical purposes, levels of sIL-1β were divided into high and low categories based on median values and were examined as both categorical and continuous variables in subsequent statistical evaluations.

Personality assessments

Personality characteristics were measured using the Big Five Inventory (BFI)-10, a succinct 10-item adaptation of the BFI that captures the essential elements of the Five-Factor Model (FFM) of personality25. This includes two Neuroticism items: i) seeing oneself as someone who tends to criticize others, and ii) being prone to nervousness. Neuroticism generally reflects a propensity for emotional instability and low self-confidence. Participants rated themselves on these attributes using a 1 to 5 scale, with total possible scores of 10, where higher values signify greater manifestations of the Neuroticism trait. The BFI-10 has been translated and validated for Korean cohorts26. In the present sample, the two Neuroticism items demonstrated a modest but acceptable inter-item correlation (Spearman’s rho = 0.219, P < 0.001), consistent with prior validation studies of the BFI-10. For the purposes of analysis, the Neuroticism trait was analyzed both by dividing it into ‘lower’ and ‘higher’ categories based on median values—a practice aligned with prior studies27—and as continuous variables.

Baseline covariates

At study entry, comprehensive sociodemographic information was obtained for all participants, including age, sex, education in years, marital status (classified as married or unmarried), living arrangement (alone or with others), employment status, and monthly income (categorized as below or above USD $2,000). Clinical variables were recorded following procedures described in earlier methodological reports21,30 . Diagnostic classification of depressive disorders adhered to previously established criteria, with additional specification of melancholic and atypical features. The clinical dataset also included variables such as age at onset, total duration of illness, number of previous depressive episodes, duration of the current episode, family history of depressive disorders, and the presence of comorbid medical conditions. These conditions were assessed using a detailed self-report questionnaire covering 15 major body systems. For each reported condition, information about ongoing treatment was also collected. This included autoimmune diseases and chronic infections; however, medication use was not systematically assessed in participants who did not report such conditions. Baseline physical health assessments recorded Body Mass Index (BMI) and smoking status. Symptomatic evaluation was conducted using various scales, including the Hospital Anxiety and Depression Scale (HADS) for depression and anxiety (HADS-D and HADS-A)28, the Social and Occupational Functioning Assessment Scale (SOFAS)23, and the Alcohol Use Disorder Identification Test (AUDIT)29. These tools provided quantitative assessments of symptom severity in various domains.

Stepwise pharmacotherapy

The treatment regimen, previously outlined in our published works21,30, is briefly recapitulated in the Supplementary Material. The pharmacological strategy began with an initial phase of monotherapy. Based on tri-weekly assessments of each participant’s response to treatment and tolerance of side effects, subsequent modifications involved either switching medications, adding supplementary agents, or employing a combination of drugs. This adaptive approach allowed for tailored pharmacotherapy suited to individual patient needs.

Outcome

The primary endpoint of the study, remission, was defined using a Hamilton Depression Rating Scale24 score of 7 or less. Participants were assessed at 3, 6, 9, and 12 weeks, and remission required a HAMD score ≤ 7 at the final 12-week evaluation (or at the last available follow-up if the 12-week visit was completed early). Intermediate fluctuation in earlier visits did not alter the final remission classification.

Statistical analysis

Baseline characteristics were divided into four groups based on the median values of sIL-1β and Neuroticism: low Neuroticism/low sIL-1β, low Neuroticism/high sIL-1β, high Neuroticism/low sIL-1β, and high Neuroticism/high sIL-1β. Group comparisons were conducted using analysis of variance (ANOVA) or chi-square tests, depending on the type of variable. Scheffé’s post hoc analyses were applied following ANOVA, and specific pairwise comparisons were used for chi-square tests when indicated. These characteristics were also examined by remission status (remission vs. non-remission) employing independent t-tests or chi-square tests. To minimize model overfitting, covariates for adjusted models were selected based on their statistical significance in univariate analyses (P < 0.05) and the absence of multicollinearity. In addition, the number of coexisting physical disorders was included as a covariate regardless of statistical significance, given its potential influence on systemic inflammatory markers. Treatment steps reflecting the stepwise pharmacotherapy protocol—including initial antidepressant class, switching during treatment, and augmentation strategies—were also incorporated as covariates to account for potential confounding by pharmacological heterogeneity. The relationship between sIL-1β and Neuroticism was evaluated using Spearman’s rank-order correlation. Associations between predictors and the likelihood of 12-week remission were analyzed using binary logistic regression. Model assumptions were evaluated through multicollinearity diagnostics (variance inflation factors), goodness-of-fit testing (Hosmer–Lemeshow), and influence diagnostics. Interaction effects were tested by adding the product term (sIL-1β × Neuroticism) to the logistic models, and full effect size estimates (odds ratios and 95% confidence intervals) are reported. Three sensitivity analyses were performed: (i) excluding participants who reported autoimmune disorders or chronic infections; (ii) restricting the analysis to patients who received serotonergic antidepressants (e.g., SSRI/SNRI monotherapy) during the study period; and (iii) restricting the analysis to participants who completed the full 12-week follow-up. Missing data were handled using complete-case analysis because the remission endpoint required a valid HAMD score at the final 12-week assessment. No imputation procedures were applied, as remission status cannot be inferred without an observed final HAMD value. Because baseline comparisons (Tables 1 and S1) were used solely for covariate screening rather than hypothesis testing, correction for multiple comparisons was not applied. Similarly, binary and continuous specifications of sIL-1β and Neuroticism represent alternative operationalizations of the same constructs rather than independent statistical tests. All statistical tests were two-tailed, with a significance level set at P < 0.05. These statistical analyses were executed using IBM SPSS Statistics, Version 27.

Table 1 Baseline characteristics according to baseline levels of serum interleukin-1 beta (sIL-1β) and the Neuroticism of the Big Five Inventory in patients with depressive disorders (N = 1086).
Full size table

Results

Recruitment process

Details of participant recruitment are presented in Supplementary Figure S1. A total of 1,262 individuals were initially screened, among whom 1,094 (86.7%) consented to blood sampling, and 1,086 (86.1%) completed at least one follow-up visit within the 12-week treatment period. Comparative analyses between the 1,086 included participants and the 176 excluded individuals showed no statistically significant differences in baseline sociodemographic or clinical characteristics (all P > 0.1). Of those who were followed longitudinally, 490 (45.1%) achieved remission according to the study criteria. While the initial inclusion criteria allowed for the enrollment of individuals aged 8 years and older as described in the eligibility criteria, the actual participants who consented and were enrolled in the study were all aged 18 years or older.

Baseline characteristics

Table 1 summarizes baseline demographic and clinical features across four groups categorized by median serum sIL-1β levels (1.11 pg/mL) and Neuroticism scores (cutoff = 7/8). Significant group differences emerged in several domains, including age, educational attainment, marital status, presence of atypical depressive features, age at first onset, number of depressive episodes, and scores on the HADS-D, HADS-A, and AUDIT. Post hoc analysis revealed that more complex socio-clinical profiles—characterized by unmarried status, presence of atypical depressive features, higher number of depressive episodes, and elevated levels of depression, anxiety, and alcohol consumption—were prevalent in the group with higher sIL-1β and Neuroticism levels. Additional examination of these characteristics in relation to remission outcomes is presented in Supplementary Table S1. Notably, factors like younger age, lower income, earlier onset of depression, longer current episode duration, non-smoking status, and higher scores on the HADS-D and HADS-A, along with lower SOFAS scores, were linked to non-remission outcomes. Eight variables—age, marital status, monthly income, atypical features, number of depressive episodes, duration of current illness, and scores on HADS-A and AUDIT—were chosen based on statistical significance (P < 0.05) and their importance in controlling for potential multicollinearity. In addition, the number of physical disorders and treatment steps were included as adjustment factors given their potential influence on the exposure and/or outcome variables. No evidence of problematic multicollinearity was observed among these variables (all variance inflation factors < 2). Baseline serum sIL-1β concentrations showed a modest but statistically significant positive correlation with Neuroticism levels (rho = 0.099, P = 0.045).

Individual effects of sIL-1β and Neuroticism on remission

Table 2 summarizes how baseline sIL-1β and Neuroticism levels were associated with remission status at 12 weeks. When analyzed as both categorical and continuous variables, higher sIL-1β levels were consistently linked with non-remission in both the unadjusted and adjusted regression models. For Neuroticism, greater scores were significantly related to non-remission in the unadjusted analyses for both variable types, although in the adjusted model, the association remained significant only when Neuroticism was treated as a continuous measure.

Table 2 Individual effects of serum interleukin-1 beta (sIL-1β) and the Neuroticism of the big five inventory as binary and continuous variables on the probability of 12-week remission.
Full size table

Interactive effects of sIL-1β and Neuroticism on remission

Figure 1 illustrates the combined influence of sIL-1β and Neuroticism on the probability of achieving remission after 12 weeks of treatment. Among participants with lower Neuroticism, higher sIL-1β concentrations were not significantly associated with remission outcomes, even after covariate adjustment. In contrast, within the high-Neuroticism group, elevated sIL-1β was strongly related to an increased likelihood of non-remission, demonstrating a significant interaction in the adjusted model. This suggests that individuals with both elevated sIL-1β and high Neuroticism are at considerably greater risk of failing to achieve remission. When both predictors were treated as continuous variables, the interaction remained statistically significant, confirming a robust interplay between inflammatory activity and personality factors in influencing treatment response (Wald = 6.624, P = 0.009; OR = 0.93, 95% CI = 0.90–0.97). Analyses using Neuroticism as a continuous trait yielded results consistent with those obtained using the median-split categories, further reinforcing the robustness of the interaction effect. Furthermore, model diagnostic checks indicated that the logistic regression assumptions were adequately met: the Hosmer–Lemeshow goodness-of-fit test suggested acceptable model fit (P > 0.05), and influence diagnostics did not identify any observations exerting undue leverage on parameter estimates.

Fig. 1
figure 1

Interactive combined effects of baseline levels of serum interleukin-1 beta (sIL-1β) and the Neuroticism of the big five inventory on the probability of 12-week remission. aInteractive effects of sIL-1β and Neuroticism on remission status were estimated using multinomial logistic regression; and bodds ratios (95% confidence intervals) [OR (95% CI) were calculated using binary logistic regression for lower (< 1.11 pg/mL) vs. higher (≥ 1.11 pg/mL) sIL-1β levels on remission status, after adjustment for age, marital status, monthly income, atypical feature, number of depressive episodes, duration of present illness, number of physical disorders, scores on Hospital anxiety & depression scale-anxiety subscale and alcohol use disorder identification test, and treatment steps. *P < 0.05; ‡P < 0.001.

Full size image

Sensitivity analyses of the IL-1β × Neuroticism interaction

Three sensitivity analyses were conducted to assess the robustness of the IL-1β × Neuroticism interaction. First, excluding 27 participants with autoimmune disorders or chronic infections yielded results consistent with the primary analysis; the interaction remained significant in models using both binary variables (Wald = 4.509, P = 0.024; OR = 0.61, 95% CI = 0.36–0.97) and continuous measures (Wald = 6.381, P = 0.010; OR = 0.93, 95% CI = 0.91–0.97). Second, restricting the sample to patients receiving serotonergic antidepressant monotherapy (n = 327; 117 remitters, 35.8%) produced comparable findings. The IL-1β × Neuroticism interaction remained statistically significant and demonstrated slightly stronger effect sizes in both categorical (Wald = 4.806, P = 0.015; OR = 0.57, 95% CI = 0.27–0.96) and continuous models (Wald = 6.981, P = 0.007; OR = 0.92, 95% CI = 0.89–0.96). Third, when analyses were limited to participants who completed the full 12-week follow-up (n = 966; 429 remitters, 44.4%), the interaction again remained significant for both binary (Wald = 4.680, P = 0.021; OR = 0.59, 95% CI = 0.33–0.95) and continuous variables (Wald = 6.786, P = 0.008; OR = 0.93, 95% CI = 0.89–0.96). Across all sensitivity analyses, the direction and magnitude of the interaction were consistent with the main findings, indicating that the IL-1β × Neuroticism effect was not driven by immune-related conditions, treatment heterogeneity, or differential follow-up. All model diagnostic checks confirmed that logistic regression assumptions were adequately satisfied.

Discussion

In this prospective, naturalistic study designed to closely reflect routine clinical practice, we found that elevated serum interleukin-1 beta (sIL-1β) levels combined with high Neuroticism substantially reduced the likelihood of achieving remission after 12 weeks of antidepressant therapy. This association was further supported by a modest but statistically significant correlation between these two variables. The interaction effect persisted across multiple sensitivity analyses, reinforcing the stability of this association. These findings indicate a form of dynamic modulation between biological and psychological factors, suggesting that their interaction may play an important role in shaping antidepressant treatment outcomes through underlying bidirectional mechanisms.

The link between higher sIL-1β concentrations and elevated Neuroticism—both associated with lower remission rates—may be interpreted through several empirically supported pathways. Firstly, although baseline sIL-1β and Neuroticism were only modestly correlated, this small but statistically significant association may indicate that inflammatory activation and emotional reactivity overlap to some extent in shaping stress sensitivity and mood regulation20. However, given the small effect size, this relationship should be interpreted as subtle rather than synergistic, and any potential interaction between these factors is likely to be modest in magnitude. Additionally, individuals exhibiting both higher sIL-1β and higher Neuroticism tended to show more adverse socio-clinical profiles—including unmarried status, atypical depressive features, a greater number of depressive episodes, and higher levels of depression, anxiety, and alcohol use—which are well established as predictors of poorer clinical outcomes31,32. These co-occurring vulnerabilities may contribute to persistently worse treatment trajectories. Furthermore, elevated sIL-1β may disrupt monoaminergic signaling and interact with stress-related physiological responses, mechanisms that are consistently implicated in inflammation-related mood dysregulation. When combined with the behavioral and emotional characteristics typically associated with high Neuroticism—such as maladaptive coping and heightened perceived stress—these processes may create a reinforcing cycle that lowers the likelihood of remission33,34. Although genetic or epigenetic predispositions linking inflammation and Neuroticism have been proposed in prior literature19, these factors were not assessed in the present study and thus should be regarded as speculative rather than explanatory. Moreover, IL-1β represents only one component of the broader inflammatory cascade. Upstream activation of IL-1β frequently drives elevations in IL-6 and TNF-α, cytokines that have also been implicated in antidepressant nonresponse. Whether the interaction observed in this study is specific to IL-1β or reflects a more generalized inflammatory signature remains an important question for future multimarker investigations.

In contrast, the observation that elevated IL-1β did not significantly impair remission among individuals with low Neuroticism may reflect several protective mechanisms supported by previous research. People with lower Neuroticism generally demonstrate greater psychological resilience and more effective stress-coping strategies, which can buffer the emotional and physiological impact of elevated cytokines35. They may also exhibit a more regulated inflammatory response, reducing the extent to which cytokine activation interferes with mood or cognitive processes36. Moreover, lower Neuroticism is commonly associated with healthier behavioral patterns—including better sleep, more physical activity, and greater adherence to treatment—which can contribute to lower systemic inflammation and help counteract the detrimental influence of elevated IL-1β37. These pathways, while plausible and consistent with existing evidence, remain indirect and warrant further investigation using longitudinal or mechanistic study designs.

In relation to the individual associations observed in this study, the significant correlation between high Neuroticism and lower remission rates aligns with previous research findings14,15. Our study also found that high levels of IL-1β predict poorer antidepressant treatment response, a result that is consistent with some prior studies9, yet stands in contrast to the more controversial findings reported in meta-analyses6 and a broader range of research10,11. This suggests that further investigation is warranted. However, our research uniquely contributes to the ongoing debate by analyzing the modulating effect of Neuroticism on the relationship between IL-1β and antidepressant treatment response. We discovered that this effect is particularly pronounced in the high Neuroticism group, while it disappears in the low Neuroticism group, thereby providing potential clues to resolve the longstanding controversy surrounding this relationship. From a transdiagnostic perspective, this finding is notable because Neuroticism is strongly linked to anxiety-depression comorbidity and may influence adherence, engagement, and stress-related inflammatory activation. Integrating personality traits into biomarker-based risk stratification may therefore help capture broader affective vulnerability profiles.

Our study faces several limitations, most notably in the scope of inflammatory biomarker assessments, which were only conducted at baseline. This limitation restricts our ability to monitor how changes in these biomarkers, influenced by treatment, could impact antidepressant responses—a phenomenon previously demonstrated to be significant by Brunoni et al.11. Furthermore, although sensitivity analyses excluding participants with autoimmune disorders or chronic infections yielded results consistent with our main findings, we did not systematically assess the use of anti-inflammatory or immunomodulatory medications across the full cohort. This limitation warrants consideration, as such medications may alter peripheral inflammatory markers and could theoretically influence antidepressant treatment efficacy38. Another limitation is that Neuroticism was assessed using only two items from the BFI-10. Although this brief measure has demonstrated acceptable validity25,26, its restricted item range may limit measurement precision and could attenuate effect size estimates. Consistent with this, the IL-1β × Neuroticism interaction was stronger when Neuroticism was modeled as a continuous variable (interaction Wald = 6.624) rather than as a dichotomized trait (interaction Wald = 4.515), suggesting that the abbreviated scale may have underestimated the true magnitude of the association. Thus, part of the residual variability may be attributable to measurement constraints rather than psychological processes alone. Furthermore, although the BFI-10 has shown acceptable psychometric properties, its two-item Neuroticism subscale inevitably provides a limited assessment range. Future research would benefit from employing more comprehensive personality inventories—such as the NEO-inventories and EPQ-N—which offer greater measurement precision and may capture more refined variability relevant to antidepressant treatment response. In addition, systematic information on psychiatric comorbidities—such as generalized anxiety disorder, panic disorder, or post-traumatic stress disorder—was not collected. These conditions could influence both inflammatory activity and personality-related vulnerability, and their absence from the analytic models represents another important limitation. While the naturalistic design of this study offers strong ecological validity by mirroring real-world clinical settings, it also introduces certain constraints. As this was a naturalistic observational design, the associations identified should not be interpreted as causal; unmeasured or bidirectional influences between inflammation and affective vulnerability may also have contributed to the observed patterns. The absence of a standardized treatment protocol may have contributed to variability in therapeutic outcomes. The flexible, stepwise pharmacotherapy approach—encompassing differences in switching patterns, augmentation strategies, and patient-driven treatment decisions—may influence inflammatory trajectories and remission probabilities. Different antidepressant classes and combination strategies exert varying immunomodulatory effects; thus, this pharmacological heterogeneity could either obscure or exaggerate the predictive role of baseline IL-1β depending on the treatment context. However, analyses incorporating treatment-related variables, as well as a sensitivity analysis restricted to patients receiving serotonergic antidepressant monotherapy, also yielded significant and directionally consistent interaction effects, suggesting that the observed association was robust and unlikely to be explained by pharmacological heterogeneity. Moreover, the single-center design of this study may limit the external validity of our findings, as the sample might not fully capture the diversity of patient characteristics and treatment environments encountered in broader clinical settings.

Despite these limitations, the study has several notable strengths that enhance its scientific contribution. Chief among them is the large sample size, which provides strong statistical power and represents a substantial advancement over many previous investigations into biomarkers associated with antidepressant response. This robust sample allows for more reliable and generalizable conclusions. Additionally, the study achieved high follow-up completion rates, ensuring comprehensive longitudinal data and minimal participant attrition. Such methodological rigor reduces the potential for bias stemming from missing data and reinforces the overall reliability and validity of the findings. Furthermore, in a sensitivity analysis restricted to participants who completed the full 12-week follow-up, the IL-1β × Neuroticism interaction remained significant, further supporting the robustness of our findings.

Conclusion

In conclusion, our study reveals dynamic modulation of antidepressant treatment response through the interaction of sIL-1β and Neuroticism levels at 12 weeks. This finding underscores the complex interplay between biological markers and psychological traits in determining treatment outcomes, providing a deeper understanding of the individual variability in response to antidepressant therapy. The general implications of these results are significant, suggesting that the integration of biomarker and psychological assessments could lead to more personalized and effective treatment strategies for depression.

Clinical implications

From a clinical standpoint, our results point toward several potential applications. Individuals presenting with both elevated sIL-1β and high Neuroticism may represent a subgroup at heightened risk for suboptimal remission. Identifying this profile early could help clinicians prioritize closer symptom monitoring, implement more proactive follow-up during the initial treatment phase, or consider earlier augmentation or combination strategies. Moreover, because high Neuroticism is associated with maladaptive stress responses, these patients may particularly benefit from targeted psychotherapeutic interventions—such as cognitive-behavioral techniques focused on stress regulation or emotional reactivity—delivered alongside pharmacotherapy. In clinical settings where cytokine testing is feasible, IL-1β assessment may also contribute to risk stratification frameworks that guide individualized care. Additionally, these findings provide a rationale for stratified clinical trial designs in which high-IL-1β/high-Neuroticism patients are prospectively identified and evaluated for differential response to adjunctive anti-inflammatory strategies. Such stratification may enhance signal detection and improve the efficiency of future precision-psychiatry interventions.

Future research

These results also raise questions for future clinical trials. Stratifying participants by personality traits may help identify subgroups most likely to benefit from adjunctive anti-inflammatory interventions, particularly among those showing concurrent elevations in IL-1β and neurotic tendencies. Such stratification could refine trial design, enhance signal detection, and support the broader development of precision psychiatry approaches. For future research, these findings advocate for longitudinal studies that further explore the mechanisms underpinning these interactions over longer periods and across diverse patient populations. Additionally, investigating other biomarkers in conjunction with broader psychological traits could expand our understanding of their collective impact on antidepressant responses, paving the way for a more holistic approach to depression treatment. Replication studies incorporating repeated cytokine measurements—not only baseline IL-1β but also dynamic changes in IL-6, TNF-α, and other inflammatory markers—will be essential to determine whether the interaction observed here reflects a stable individual vulnerability or a more time-sensitive inflammatory–psychological coupling.