Year-Long Study Will Track How ADHD and Borderline Personality Disorder Reshape Bipolar Disorder

Bipolar disorder has long been described as a disease of rhythm: alternating waves of mania and depression that sweep through a person’s life in patterns that clinicians have tried for decades to chart, predict, and tame. Yet for a substantial share of patients, the textbook rhythm is drowned out by something else entirely — the co-occurrence of a second psychiatric condition that may fundamentally alter how the illness unfolds. A new study protocol published in Frontiers in Human Neuroscience describes an ambitious effort to capture, in unprecedented biological detail, how bipolar disorder behaves when it travels alongside attention deficit hyperactivity disorder (ADHD) or borderline personality disorder (BPD).

The project, known as MapCoBD (Mapping the natural course of comorbid bipolar disorder via integration of multimodal data), is a 12-month naturalistic observational study led by researchers at the University of Cagliari, the University of Padova, and collaborating institutions across Italy and Canada. Its central question is deceptively simple: does the year-long clinical course of bipolar disorder differ between people with classic episodic illness and those whose bipolar disorder is complicated by ADHD or BPD? The answer could reshape how psychiatrists understand, monitor, and treat one of the most disabling conditions in medicine, which ranks among the top ten causes of disability and mortality worldwide.

The rationale for the study rests on a striking gap in the literature. Classic episodic bipolar disorder — the reference point for nearly everything psychiatry knows about the illness — is characterized by a later age of onset, low rates of comorbidity, and familial loading for the disorder. By contrast, people with bipolar disorder and comorbid BPD show greater psychiatric complexity, elevated suicide-related morbidity, and increased long-term hospital utilization, while those with comorbid ADHD face a lower quality of life, more mood episodes, higher rates of substance abuse and dependence, and worse overall functioning. Despite these well-documented clinical burdens, almost no longitudinal study has systematically compared the actual trajectories of these three groups while simultaneously measuring the biology that might underlie their differences.

MapCoBD will recruit 150 individuals with bipolar disorder at two Italian clinical sites, the University of Cagliari and the University of Catanzaro, dividing them into three groups of 50: classic episodic bipolar disorder, bipolar disorder with comorbid ADHD, and bipolar disorder with comorbid BPD. Each participant will be assessed at baseline, six months, and twelve months, with a battery spanning mood symptom ratings, neuropsychological testing, fasting blood sampling, and structural brain MRI at 3 Tesla. Crucially, the study does not stop at periodic snapshots. Every participant will complete daily mood ratings throughout the entire year, allowing researchers to capture the fine-grained texture of mood instability that episodic clinic visits inevitably miss.

The primary endpoint is a measure the researchers call illness burden: the proportion of evaluable follow-up days spent with moderate-to-severe psychopathological symptoms, derived from the prospective NIMH Life Chart Method. Each day is classified as either meeting or not meeting this threshold, and the resulting score ranges from zero — no days of significant symptoms — to one, meaning moderate-to-severe symptoms throughout the entire interval. Mathematically, the measure corresponds to the normalized area under the curve of a daily binary illness indicator, a formulation that turns a year of lived experience into a single, comparable number across all three groups. Daily affective ratings will separately characterize mood instability as a secondary outcome.

Beneath the clinical monitoring lies an equally sophisticated molecular program. At each visit, fasting blood samples collected between 8:00 and 10:00 a.m. will yield serum, plasma, and RNA for analysis of three biological systems chosen for their relevance to mood regulation and stress responsivity. The first is the endocannabinoid system, a critical modulator of synaptic plasticity and emotional processing whose dysregulation has been implicated in bipolar disorder; large-scale genetic studies have even found enrichment of endocannabinoid signaling gene sets in bipolar disorder risk loci. The second is inflammation, with a panel of 120 cytokines, growth factors, and chemokines quantified through a multiplex antibody array — a choice grounded in meta-analyses showing consistently elevated pro-inflammatory cytokines such as IL-6 and TNF-α across mood states. The third is the tryptophan–kynurenine pathway, the metabolic route that converts the amino acid tryptophan into neuroactive compounds ranging from the neuroprotective kynurenic acid to the neurotoxic quinolinic acid, thereby forming a direct biochemical bridge between immune activation and brain function.

The technical machinery behind these measurements is formidable. Endocannabinoids including anandamide and 2-arachidonoylglycerol, along with their catabolic enzymes and receptor targets, will be quantified using liquid chromatography tandem mass spectrometry, PCR, and western blotting. Kynurenine pathway metabolites will be measured with high-performance liquid chromatography and tandem mass spectrometry, with enzymatic activity inferred from ratios such as kynurenine-to-tryptophan for IDO and TDO activity, and kynurenic acid-to-quinolinic acid for the balance of NMDA receptor antagonism and agonism. MicroRNA profiles — capturing epigenetic regulation — will be sequenced on an Illumina NextSeq 2000 after careful quality control and normalization. Meanwhile, the MRI protocol includes diffusion-weighted imaging for white matter tract reconstruction and resting-state functional connectivity analyses performed with the CONN toolbox and SPM 12, examining both region-to-region and seed-to-voxel connectivity.

Perhaps the most forward-looking element of the protocol is its use of machine learning to weave these data streams together. Because the number of measured features will far exceed the 150 participants, the researchers will apply dimensionality reduction techniques such as LASSO and principal component analysis before training supervised algorithms including penalized logistic regression, random forests, support vector machines, and extreme gradient boosting. Two predictive tasks are planned: a multiclass classification distinguishing the three clinical groups, and prediction of longitudinal mood instability and illness burden. To guard against the overfitting that plagues high-dimensional psychiatric data, all preprocessing, feature selection, and hyperparameter tuning will be confined to inner folds of a repeated nested cross-validation scheme, with held-out data touched only for final evaluation. Feature importance will be interpreted using permutation importance and SHAP values, and the team is explicit that any signatures emerging from these analyses are exploratory and will require independent validation before any clinical use.

The hypotheses are clinically pointed. The researchers expect both comorbid groups to show greater illness burden and mood instability than the episodic reference group, but with distinct flavors: the BPD-comorbid group is anticipated to display greater affective instability and impulsivity, while the ADHD-comorbid group is expected to show more prominent attentional and neuropsychological impairment. This distinction matters because emotional dysregulation, impulsivity, and attentional difficulties overlap heavily across ADHD, BPD, and bipolar disorder itself, often complicating diagnosis and treatment planning. If the study finds that these overlapping symptoms track with distinguishable biological signatures — say, different inflammatory profiles, kynurenine metabolite ratios, or patterns of anterior cingulate connectivity — it would suggest that the comorbid presentations are not merely more severe versions of bipolar disorder but biologically distinct trajectories in their own right.

The implications, if the hypotheses hold, could extend from the clinic to the broader philosophy of psychiatric diagnosis. Evidence of distinguishable longitudinal profiles could push clinicians toward more systematic screening for ADHD and BPD in bipolar patients, and toward comorbidity-sensitive monitoring — closer tracking of emotional volatility in one group, more rigorous assessment of executive function in another. Repeated biological sampling may also reveal whether molecular alterations are stable traits of the comorbid profiles or state-dependent echoes of mood fluctuation, a question that cross-sectional studies have never been able to answer. The authors are careful to note the limitations: the naturalistic design precludes causal inference, a 12-month window may not capture the full cyclicity of the illness, and the sample size is modest for machine-learning analyses. Recruitment is already well underway, with 110 individuals tested at baseline. But if MapCoBD succeeds in linking the daily texture of mood to the molecular currents beneath it, the study could mark a decisive step toward precision psychiatry for one of the world’s most burdensome mental illnesses.

Subject of Research: A longitudinal multimodal study protocol comparing the clinical course and biological signatures of bipolar disorder with and without comorbid ADHD or borderline personality disorder

Article Title: Mapping the natural course of comorbid bipolar disorder via integration of multimodal data (MapCoBD): a study protocol

Article References: Comai, S., Morena, M., Steardo, L., Jr, Squassina, A., Pisanu, C., Meloni, A., Congiu, D., Lazzardi, S., Pinna, M., Contu, M., Nasini, S., Barzon, B., Zuliani, A. M., Dall'Acqua, S., Balestrieri, A., Campolongo, P., Pullano, I., Iazzolino, A. M., de Filippis, R., … Manchia, M. (2026). Mapping the natural course of comorbid bipolar disorder via integration of multimodal data (MapCoBD): a study protocol. Frontiers in Human Neuroscience, 20, Article 1924416. https://doi.org/10.3389/fnhum.2026.1924416

Image Credits: AI Generated

DOI: 10.3389/fnhum.2026.1924416

Keywords: bipolar disorder, ADHD, borderline personality disorder, comorbidity, endocannabinoids, inflammation, kynurenine pathway, microRNA, neuroimaging, machine learning, mood instability, biomarkers