Neurobiological and Genomic Predictors of Relapse in Depression

When Should Antidepressant Treatment Be Discontinued? A Prospective Healthy-Controlled Case-Control Study

Background First-line treatments for Major Depressive Disorder (MDD) include second-generation antidepressants-such as selective serotonin reuptake inhibitors (SSRIs) and serotonin-norepinephrine reuptake inhibitors (SNRIs)-and/or evidence-based psychotherapies (American Psychological Association, 2019). While the efficacy of antidepressants in preventing relapse is well-established, approximately 40% of patients experience recurrence after discontinuing treatment (Kato et al., 2021). Due to the lack of clear guidelines on optimal treatment duration, some have suggested lifelong maintenance therapy. However, despite their relatively low side-effect profile, long-term antidepressant use is often reported as distressing by up to half of patients (Cascade et al., 2009).

Therefore, it is essential to identify predictive markers that can guide clinicians in deciding when it is safe to discontinue medication without increasing the risk of relapse. A recent meta-analysis conducted by our group found that demographic and clinical variables such as age, sex, or treatment resistance failed to predict recurrence (Arikan et al., 2023). Consequently, the focus has shifted to biological, electrophysiological, and genetic markers.

Objective The primary purpose (objective) of this study is to investigate whether various biological indicators provide sufficient metrics for discontinuing treatment in patients diagnosed with Major Depressive Disorder (MDD).

Study Design

The study design is described as a hybrid cohort study that integrates both retrospective and prospective data. It also incorporates an analysis strategy consistent with an unmatched case-control study methodology for sample size calculation and comparison against a Healthy Control group.

This study will enroll individuals diagnosed with MDD who respond to antidepressant treatment. Participants whose remission is maintained for 6-12 months will undergo a tapered discontinuation of their medication. They will be monitored monthly for 6-12 months following the discontinuation.

1. Study Overview and Objectives The core purpose of this research is to identify a set of multimodal biomarkers capable of predicting relapse in Major Depressive Disorder (MDD). The study is designed to answer the fundamental clinical question: "Even when a patient feels better, are they biologically recovered?" The project aims to develop a predictive algorithm that can inform clinicians on whether to continue or discontinue antidepressant medication based on objective biological data, rather than relying solely on clinical symptom assessments.

The study utilizes a hybrid cohort design, integrating retrospective data from patients with known clinical histories and prospective data from a newly recruited cohort followed over time.

II. Experimental Design and Cohorts The study is structured around three distinct participant groups to enable comprehensive comparisons.

• Group A (Retrospective / Known Phenotype): This group consists of patients with a confirmed history of either a single depressive episode or recurrent depression. Data for this group is collected at a single time point (Baseline/T0).
• Group B (Prospective / Follow-up): This cohort includes patients whose long-term outcome is unknown at the start of the study. They are enrolled, treated, and monitored over time. At the study's conclusion, they will be retrospectively categorized as "Relapsers" or "Non-Relapsers (in Remission)."
• Group C (Healthy Control): This group is composed of individuals with no history of psychiatric illness. Data is collected at a single time point to establish a healthy baseline for all biomarkers.

III. Measurement and Follow-up Timeline (Prospective Group B) The prospective cohort undergoes a series of comprehensive assessments at four key time points.

Time Point Description Measurements Taken T0 Baseline: Assessment before the initiation of antidepressant medication. All parameters measured: EEG, P300, REM, Genotype, Epigenetic i.e., Methylation, messenger Ribonucleic Acid (mRNA) expression, small ribonucleic acid (sRNA), Quality of Life (WHOQOL-BREDF) Scale, Mini Mental State Examination, Hamilton Depression Rating Scale (HDRS-17), Depression and Anxiety Stress Scale (DASS-21) T1 Early Response (4-8 weeks): An initial evaluation of treatment effect. EEG and P300 and clinical scales, i.e, HDRS-17, DASS-21 T2 Treatment Discontinuation: After a minimum of 6 months of treatment and achieving clinical remission treatment medication reduced gradually and all the measurements repeated again EEG, p300, REM, mRNA-sRNA expression, methylation, WHOQOL-BREF, HDRS-17, DASS-21 T3 End of Follow-up: The final assessment after 6 months without medication to determine relapse status. All the parameters at T2 repeated.

IV. Core Analytical Hypotheses The statistical analysis is organized into three distinct sections, each designed to test a specific set of hypotheses.

1. Baseline "Trait" Marker Identification This analysis seeks to identify stable, pre-existing biological markers that differentiate individuals with a vulnerability to recurrent depression from those who experience a single episode and from healthy controls.

• Data Pool: Combines all data from the retrospective Group A and the baseline (T0) data from the prospective Group B.
• Methodology:
• ANOVA / Kruskal-Wallis: To compare continuous biomarker data across three groups: Recurrent, Non-Recurrent, and Healthy Controls.
• Chi-Square Test: To analyze the relationship between categorical genetic data (SNPs) and relapse status.
• Expected Outcome: The Recurrent Depression group is expected to show biomarker profiles that are significantly different from the Healthy Control group. The Non-Recurrent group is hypothesized to fall somewhere between the Recurrent and Healthy Control groups, or to be similar to the healthy baseline.

1. Longitudinal Analysis of Biological Normalization This is the most critical component of the study, designed to test whether biological recovery aligns with clinical recovery. The central question is, "Does the biology of patients who will relapse look different from those who won't, at the moment medication is stopped?"

• Data Pool: Exclusively uses the longitudinal data from the prospective Group B (T0, T2, T3).
• Methodology: Repeated Measures ANOVA or Linear Mixed Models (LMM) will be used to analyze changes over time. The key statistical target is a significant "Group x Time" interaction effect.
• Hypothesized Scenarios:

1. Non-Relapsing Group: Biomarkers that are abnormal at baseline (T0) are expected to normalize by the time of medication cessation (T2), becoming statistically indistinguishable from the Healthy Control group.
2. Relapsing Group: Despite achieving clinical remission, this group's biomarkers are expected to remain abnormal at T2. The biological dysfunction will persist, indicating an incomplete recovery that predisposes them to relapse.
3. Early Treatment Response as a Predictor This analysis aims to determine if very early changes in brain activity can predict the final outcome months later.

• Data Pool: Uses T0 and T1 data (EEG/P300/HDRS-17/DASS-21) from Group B.
• Methodology: The percentage of change in measurements from T0 to T1 will be calculated. This "delta" value will then be tested for its ability to predict relapse status at T3 using point-biserial correlation or logistic regression.

The Clinical Decision Support Algorithm The final and most translational output of the study is the development of a predictive model to guide clinical practice.

• Methodology: Receiver Operating Characteristic (ROC) curve analysis will be performed.
• Process:

1. Creation of a Normalization Index (NI): For each patient in Group B, a score will be calculated at T2 that quantifies how much their key biomarkers deviate from the healthy control average (e.g., as a Z-score).
2. Prediction Test: The NI and other biomarker values from T2 will be used to predict the actual relapse outcome observed at T3.
3. Performance Evaluation: The Area Under the Curve (AUC) will be calculated to measure the model's predictive accuracy. An AUC value above 0.80 is considered excellent.
4. Cut-off Value Determination: The analysis will identify a specific biomarker threshold (a "cut-off" point) that provides the optimal balance of sensitivity and specificity for predicting relapse.

• Example Clinical Recommendation: "Based on the model, patients whose P300 latency remains above 320ms or whose specific mRNA expression is below value 'X' at the time of planned medication discontinuation have an 85% risk of relapse. It is recommended that these patients continue treatment, even if they report feeling well." V. Anticipated Conclusions and Clinical Impact

Upon completion, this study is positioned to make several powerful and field-advancing claims:

1. Identification: "The phenotype of relapsing depression is distinguishable from single-episode depression at baseline through a distinct profile of P300 amplitude deficits and specific RNA expression patterns."
2. Mechanism: "While antidepressant treatment can resolve clinical symptoms within six months, the underlying electrophysiological dysfunctions-a form of 'biological scar'-persist in the group that is destined to relapse."
3. Clinical Recommendation: "The decision to terminate antidepressant therapy should not be based solely on clinical remission as measured by scales like Hamilton. It must be guided by evidence of biological normalization, as determined by biomarker values at the point of remission."

Healthy Control Recruitment

Healthy controls will be recruited from various departments of a large corporate organization. Following informed consent, each participant will undergo psychiatric interviews conducted by board-certified psychiatrists using the Structural Clinical Interview for DSM-5 (SCID-5) and a sociodemographic questionnaire. Interview results will not be disclosed to participants. Subsequent assessments will include resting-state qEEG, event-related potentials (P300), overnight sleep EEG, and blood sampling for genetic analysis. Devices and protocols will match those used in the depression group to ensure consistency.