Editorial - Recent Advances on the Multimodal Search for Markers of Treatment Response in Affective Disorders: From Bench to Bedside? Volume II

Martin Walter^1*Sebastian Olbrich²Nils Opel¹Maria Strauss³Bin Zhang⁴Lena Vera Danyeli¹

• ¹University Hospital Jena, Jena, Germany
• ²Psychiatric University Hospital Zurich, Zurich, Zürich, Switzerland
• ³University Hospital Leipzig, Leipzig, Lower Saxony, Germany
• ⁴Tianjin Anding Hospital, Tianjin, China

Affective disorders and their comorbidities concern the largest group of psychiatric patients due to 1) early and continuous lifelong, high incidences and 2) chronicity, often due to inadequate or delayed treatment responses (Nandi et al., 2009;Calvo et al., 2020;Wang et al., 2007). This contrasts with the comparably high level of treatment success during controlled treatment escalation, with response levels e.g. to ECT of 60-80% even in difficultto-treat patients. Although the application of stratified treatment regimens-starting with relatively well-tolerated drugs or psycho(mono-)therapy-has proven efficient compared to free treatment regimens (Adli et al., 2002), most algorithms still begin with first-line drug suggestions that offer only moderate expected response rates. For these first-line treatments, remission rates are as low as 33% (STAR-D), decreasing further at subsequent treatment levels. Consequently, poor or significantly delayed remission rates add to a complex situation that recently resulted in as few as 7% of patients (of several diagnoses) achieving timely and efficient treatment on an international level (Vigo et al., 2025). Not surprisingly, the long search for the right individual treatment, in the presence of side effects accompanying ineffective treatments, which may even worsen clinical status, adds to the number of patients, who disappointed, drop out from treatment courses which might have eventually led to a significant response. The search for (bio-)markers predicting individual treatment responses has been a task spanning clinical and scientific careers and has led to some candidate markers, that, however, have relatively low predictive values or replicability (Bartova et al., 2019). While factors predicting general treatment resistance have been reported, these, however, seem to have little value in supporting a bypass towards more targeted interventions. Given the fact that there is a very large group of patients who would, in principle, respond or remit once efficient treatments are identified and applied, and given the long course of ineffective treatment, during which patients lose adherence, identifying and implementing new, additional predictors promises substantial clinical and health economic benefits. This benefit is maximized if these markers not only allow early transition to escalated treatments but also identify subsystems that characterize biological patient subgroups with specific therapeutic targets different from those addressed by first-or second-line treatments. Earlier collections of such candidates particularly have demonstrated potential ways to incorporate additional knowledge into treatment prediction (Schmidt et al., 2019). In the past five years, evidence has accumulated on 1) new mechanisms and candidates and 2) new treatment opportunities provided by large-scale reimbursement options e.g. TMS or NMDA antagonists. These new opportunities and insights have fueled research into treatment rationales beyond monoaminergic approaches, identifying exciting new candidate markers related to glutamatergic modulation. New therapeutically driven routes towards glutamatergic mechanisms have emerged alongside growing insights into biological phenotypes underlying long-debated clinical subgroups of depression, mainly melancholic and atypical subtypes (Lorenze et al., 2021). Importantly, these subtypes, with distinct symptom and response profiles, have been increasingly characterized by multidimensional biological markers (Woelfer et al., 2019) that link regional brain imaging evidence, patient history and molecular convergence of several biological systems. Ongoing evidence on patients best characterized as "immunometabolic" (Milaneshi et al., 2020), suggested a convergence of a metabolic/adipose load with accompanying (low-grade) inflammatory signatures in these patients indicating the need for new and adapted treatment strategies.The emergence of neuropsychoimmunological patient characteristics has interestingly converged with the introduction of new treatments at the level of glutamatergic signaling, where activation of the IDO is related to downstream changes in kynurenic metabolites with NMDA-ergic properties. This development was reflected by the contributions to our second edition: Consequently, Reininghaus and colleagues extended their previous characterization of kynurenic pathway abnormalities (Reininghaus et al., 2019) in depression to explore their predictive potential for individual treatment outcomes (Reininghaus et al., 2024 -this issue). They show that statically altered kynurenic acid (KYN) could not only help identify non-responders, but that KYN increases over the course of treatment signifying treatment response. Importantly, such metabolic signatures became a recent focus of non-invasive imaging studies (Sen et al., 2021). In this issue, Moreau and colleagues review evidence from 20 different studies on imaging treatment predictors in OCD (Moreau et al., 2025 -this issue). As a result, they conclude that mainly structural assessments failed to provide consistent markers of treatment response. Their findings may be interpreted along the lines that relevant processes may better be captured by dynamic i.e. metabolic or neurophysiological investigations. Such metabolic investigations became possible though the introduction of feasible MR spectroscopy sequences (MRS), which can capture a wide range of metabolic markers in certain brain regions. Watling et al. applied this technique to an emerging candidate of metabolic abnormalities, namely glutathione (GSH), the brain's most abundant antioxidant (Watling et al., 2023 -this issue). While they failed to find differential GSH levels in a group of PTSD patients, they reported initial evidence for altered metalloproteinase (MMP)-9 and myeloperoxidase (MPO) levels, which were also related to disease progression. Previous accounts for the application of quantitative EEG in treatment prediction (Schiller, 2019) have been followed up by the last contribution to this second volume: Kim et al. demonstrated that a QEEG allowed for successful identification of depression patients with an accuracy of 92.31 and a 10-fold cross-validation loss of 0.13% (Kim et al., 2022 -this issue). While these numbers already lie considerably higher than those reported for MRI-based imaging signals (Winter et al., 2024), the true power of electrophysiological signals, however, might be found in the correct labeling of treatment responders for various treatment options in depression (Arns et al., 2023;Ip et al., 2021).In conclusion, the most recent advances concern both new targets and techniques with an expanding set of potential biomarkers across multiple modalities which, however, may depict different aspects of similar mechanisms or patient characteristics. Therefore, the search for suitable treatment markers, guiding towards specific interventions, remains an open quest with high potential pains and gains.