Morning Cortisol Levels Track Anxiety and Core Depression Symptoms in First-Episode Patients

People experiencing a first episode of major depressive disorder, before any antidepressant treatment has begun, show measurably higher levels of the stress hormone cortisol in their blood, and the degree of that elevation appears to mirror the severity of several core symptom clusters, according to a new cross-sectional study published in Frontiers in Psychiatry. Researchers at Xiamen Xianyue Hospital, affiliated with Xiamen Medical College and Fujian Medical University, measured early-morning serum cortisol in 103 patients with acute, first-episode, drug-naïve major depressive disorder and compared the values with 99 age- and gender-matched healthy controls. The patients averaged 307.03 ± 77.80 nmol/L of cortisol, significantly higher than the control group’s 261.13 ± 58.15 nmol/L, a difference the authors report as statistically significant after non-parametric testing.

The study’s central contribution lies in going beyond total depression scores. Most previous work linking the hypothalamic-pituitary-adrenal (HPA) axis to depression has examined overall severity ratings, but major depressive disorder is notoriously heterogeneous, and the team wanted to know whether cortisol levels correspond to specific symptom dimensions. To do this, they rated each patient on the 24-item Hamilton Depression Rating Scale (HAMD-24), an observer-administered instrument whose 24 items are grouped into seven factors: anxiety/somatization, weight, cognitive impairment, diurnal variation, retardation, sleep disorder, and hopelessness. Patients also completed the Zung Self-Rating Depression Scale (SDS), which captures the subjective experience of depressive symptoms.

The results were strikingly patterned. Serum cortisol correlated positively and significantly with the anxiety factor (Spearman ρ = 0.642), the retardation factor (ρ = 0.481), the despair factor (ρ = 0.399), the cognitive impairment factor (ρ = 0.365), and the total HAMD-24 score (ρ = 0.602), with all of these correlations surviving false-discovery-rate correction for multiple comparisons. It also correlated positively with the self-rated SDS score (ρ = 0.460). By contrast, cortisol showed only non-significant positive trends with the weight and diurnal variation factors and a non-significant negative trend with the sleep factor. In other words, the hormone’s signal was concentrated in the symptom domains most often considered central to the illness rather than in its vegetative features.

The anxiety finding was the strongest of all. The HAMD-24 anxiety factor bundles psychic anxiety, somatic anxiety, gastrointestinal symptoms, general symptoms, hypochondriasis, and insight, spanning both mental and bodily manifestations of distress. The authors argue that this makes the correlation particularly relevant to the so-called anxious depression phenotype, and they suggest that in patients whose depression is dominated by somatic anxiety symptoms, HPA axis hyperactivity may be especially pronounced. Previous studies of depression with prominent somatic symptoms, cited in the paper, found that plasma cortisol was significantly higher in such patients than in those without somatic symptoms, lending external consistency to the new result.

To understand why cortisol has attracted decades of attention in depression research, it helps to recall its role in the body’s stress architecture. The hypothalamic-pituitary-adrenal axis operates as a three-stage hormonal cascade: the hypothalamus releases corticotropin-releasing hormone, which prompts the pituitary gland to secrete adrenocorticotropic hormone, which in turn stimulates the adrenal cortex to produce cortisol. When the intensity or duration of stress exceeds the body’s normal response capacity, the authors explain, the axis may become hyperactivated, show rhythm disturbances, or even enter a state of exhaustion after over-activation. A substantial body of clinical evidence, cited in the paper, indicates that roughly 60 to 70 percent of patients with major depressive disorder exhibit elevated serum cortisol and disrupted nocturnal rhythms, which are considered key biomarkers of the illness.

The design of the new study gives its findings particular weight within this research tradition. Because all participants were experiencing their first depressive episode and had never received antidepressant medication, psychotherapy, or physical treatments such as modified electroconvulsive therapy or repetitive transcranial magnetic stimulation, the researchers could examine HPA axis activity in a state unconfounded by prior treatment. The exclusion criteria were correspondingly strict: patients with clinically significant cardiovascular, cerebrovascular, neuroendocrine, or hematological disease were excluded, as were those with psychoactive substance use, hormonal medication use, or a family history of mental illness, all factors that could independently disturb cortisol secretion.

The laboratory methodology also merits attention. Blood samples were drawn at eight in the morning after an overnight fast, a standardized time point that reduces the noise introduced by diurnal fluctuation. Serum cortisol was quantified using a competitive-binding electrochemiluminescence immunoassay on automated Roche cobas e 402/e 801 analyzers, a method traceable to an isotope dilution gas chromatography-mass spectrometry reference standard, with a measuring range extending from 1.5 to 1750 nmol/L. Specimens were processed within two hours of venipuncture, and lipemic or hemolyzed samples were discarded to protect assay integrity. These quality-control measures matter because cortisol assays can be vulnerable to pre-analytical variation, and rigorous handling strengthens confidence that observed group differences reflect biology rather than technique.

Statistically, the investigators took precautions appropriate to the complexity of their question. Because cortisol values in both groups departed from a normal distribution, they relied on non-parametric comparisons and Spearman rank correlations, and they applied false-discovery-rate correction to guard against spurious positives across the many symptom factors examined. Bootstrap confidence intervals, computed with bias-corrected and accelerated methods, provided a robust picture of the precision of each correlation estimate. Hierarchical polynomial regression models, adjusted for age, gender, and illness duration, then tested whether the relationships between cortisol and symptom scores were strictly linear or followed more complex curves. For the anxiety, cognitive impairment, retardation, and despair factors and the total Hamilton score, only the linear term proved significant, indicating no evidence that symptoms reverse direction once cortisol passes some threshold.

That changed, however, in the relationship between cortisol and the Zung Self-Rating Depression Scale. While the observer-rated Hamilton total score showed a straightforward linear association with cortisol, the self-rated scale displayed a U-shaped relationship in the regression models. According to the fitted quadratic function, subjective depressive distress was lowest at an intermediate cortisol concentration, calculated at approximately 208.07 nmol/L in this sample, and rose both above and below that inflection point. The authors interpret this through the lens of glucocorticoid receptor sensitivity: some patients may develop glucocorticoid resistance, in which tissues respond inadequately to cortisol’s signals, while others show a blunted HPA response, and both deviations from an optimal hormonal midpoint could impair the brain’s stress-buffering and emotion-regulation capacities. They are careful to note that this inflection point derives from a single-sample fit and should not be adopted as a clinical threshold without replication.

The mechanistic account the authors offer for the symptom-specific correlations rests on well-studied neuroanatomy. Persistently elevated cortisol is thought to amplify arousal circuits centered on the amygdala, plausibly contributing to the psychic and somatic anxiety captured by the Hamilton anxiety factor, while simultaneously acting through autonomic pathways to produce gastrointestinal discomfort, generalized aching, and hypochondriacal concern. Elevated cortisol may also disrupt prefrontal-limbic information processing, and the authors cite work suggesting that cortisol dysregulation could impair glymphatic function in the brain, potentially reinforcing maladaptive cognitive patterns such as self-blame, paranoia, and rumination, though they acknowledge that this pathway remains speculative and is not directly validated by their data. Damage to hippocampal and prefrontal regions, which support self-referential evaluation and cognitive control, offers a plausible route by which hormonal excess could give rise to the worthlessness, hopelessness, and diminished sense of ability that constitute the despair factor.

The clinical implications follow logically from these correlations. If a simple morning blood draw can flag patients whose depression is accompanied by prominent anxiety, cognitive distortion, suicidal ideation, or psychomotor retardation, cortisol could help clinicians identify high-risk subgroups early in the course of illness. Patients with elevated cortisol and significant despair or cognitive symptoms may warrant closer monitoring, given their apparently heightened vulnerability, and the findings lend support to treatment development aimed at the cortisol pathway itself, including glucocorticoid receptor modulators and agents targeting corticotropin-releasing hormone receptors, approaches the authors describe as having potential value in alleviating depressive symptoms.

Several limitations temper the conclusions and chart the course for future work. The cross-sectional design captures a single moment and cannot establish whether cortisol elevation precedes the symptom clusters it correlates with, drives them, or results from them; the depressive episode itself constitutes a significant stressor that may further dysregulate the axis, creating a feedback loop that only longitudinal measurement could disentangle. The reliance on a single morning cortisol sample means that circadian rhythm disturbances, arguably as informative as absolute levels, went unmeasured, as did the upstream hormones corticotropin-releasing hormone and adrenocorticotropic hormone that would clarify where along the axis the dysfunction resides. Factors known to influence cortisol, including smoking status, body mass index, menstrual cycle phase, and wake time, were not systematically accounted for, and the sample, though adequately powered for the hypothesized effect size, remains modest. The authors indicate that dynamic observation, rhythm assessment, and fuller covariate control represent their next steps.

Subject of Research: Correlation between serum cortisol and clinical symptom dimensions in first-episode drug-naïve major depressive disorder

Article Title: Correlation between hypothalamic-pituitary-adrenal function and clinical characteristics in the acute phase of first-episode drug-naive major depressive disorder

Article References: Liu, L., Yang, X., & Lin, D. (2026). Correlation between hypothalamic-pituitary-adrenal function and clinical characteristics in the acute phase of first-episode drug-naive major depressive disorder. Frontiers in Psychiatry, 17, Article 1985217. https://doi.org/10.3389/fpsyt.2026.1985217

Image Credits: AI Generated

DOI: 10.3389/fpsyt.2026.1985217

Keywords: major depressive disorder, cortisol, HPA axis, HAMD-24, Zung Self-Rating Depression Scale, first episode, drug-naive, biomarker, anxiety, psychiatry, stress hormones, cross-sectional study