Commentary on “Disruption of Neural Homeostasis as a Model of Relapse and Recurrence in Late-Life Depression”

Global burden of disease 2010 ranked major depressive disorder as a leading worldwide contributor to years of life lived with disability.¹ Between 1990 and 2010 depression’s contribution to global disability increased substantially due to population growth and notably to aging. Late-life depression (LLD) is now a major public health concern with nearly 5 million sufferers in America alone. LLD also heightens the risk for medical comorbidities, cognitive decline, poor quality of life, caregiver burden, premature mortality and suicides. As LLD became recognized as a global health priority, pharmacologic and psychotherapeutic interventions were developed to reduce symptoms. However, these are less effective in maintaining LLD remission and preventing recurrence. An enhanced understanding of the neurobiological processes that predispose to LLD recurrence is a critical step toward early identification of those at risk. As such, a scientific model of LLD recurrence, as proposed by Andreescu and coworkers,² has significant potential to provide a framework for future cutting-edge research aimed at LLD prevention.

Andreescu et al.² provide a thorough appraisal of the clinical, neuropsychological and psychosocial predictors that influence LLD recurrence risk. A s the authors note, these observations have important clinical value, but are unlikely to assist in stratifying an individual’s risk of LLD recurrence or in detecting new targets for prevention. It is against this backdrop that their review of potential neurobiological markers — primarily those identified using functional neuroimaging methods —add great value. Early seminal functional magnetic resonance imaging (fMRI) research provided a window into how core neurocognitive brain network (or circuit) measures may serve as biological signatures of LLD treatment response.³ In symptomatic LLD patients, fMRI experiments demonstrated attenuated dorsolateral and ventromedial prefrontal activation during cognitive control and emotional reactivity task paradigms, respectively; these alterations, however, only showed partial improvement after symptom resolution. The residual alterations in these brain regions, which are nodes of the executive control network (ECN) and the default mode network (DMN), may increase vulnerability for LLD recurrence. Task-dependent fMRI studies thus tracked localized activity over time in predefined regions of interest within the core neurocognitive networks. However, it was not until the development of resting-state functional connectivity MRI (R-fMRI)⁴ that neuroscientists were able to take a network approach to understanding the complex intrinsic functional architecture of the human brain.

R-fMRI is a task-independent imaging method that is used to investigate spontaneous low-frequency (i.e., <0.1 Hz) blood oxygen-level dependent (BOLD) fluctuations at “rest;” such fluctuations were initially described as noise. However, in 1995, Biswal et al. published a landmark study demonstrating that time courses of resting spontaneous BOLD fluctuations had a high degree of temporal correlation in the bilateral motor cortices as well as other regions associated with motor function. Functional connectivity (Fc) is defined as the temporal correlations of low-frequency BOLD fluctuations between spatially segregated brain regions at rest. The recognition that these activity patterns at rest are of neuronal origin has revolutionized the field of cognitive and behavioral neuroscience. Many studies in recent years have utilized R-fMRI approaches to map brain network functioning in healthy aging and to elucidate the Fc alterations underlying various late-life neuropsychiatric diseases. Andreescu et al.² provide a detailed review of the burgeoning body of literature that points to disrupted Fc within and between three well-defined core neurocognitive brain networks: the DMN, ECN and salience network. The authors also use findings from LLD treatment response studies, coupled with data extrapolated from midlife depression research, to support their hypothesis that persistent alterations in these core brain networks during remission could serve as predictors of LLD recurrence. If so, what contributes to the persistence of localized neural activity and brain network Fc alterations that predispose an older adult to depression recurrence? This is where the concept of allostatic overload and the hypothesis of homeostatic disequilibrium as a precursor to LLD recurrence is relevant.

Allostasis, a concept introduced by Sterling and Eyer,⁵ refers to the dynamic biological processes by which homeostasis is maintained through adaptation to daily stressful events. However, the cumulative effects of repeated hits from multiple stressors over a lifespan disproportionately produce wear and tear on the body as well as the brain, a phenomenon described by McEwen and colleagues as allostatic overload.⁶ The prolonged or poorly regulated allostatic response arising from chronic or repeated stress exposure causes increased activity in the hypothalamic-pituitary-adrenal (HPA) axis, autonomic nervous system and immune system. Allostatic overload can also impair mental health through its maladaptive effects on brain plasticity; these detrimental effects are further weakened with biological aging. Key neural networks involved in emotion regulation and higher cognitive functions are early brain targets of accumulated allostatic load; interestingly, these are the same neuronal systems affected in LLD. After resolution of stressful life events, the stress-induced changes in these brain networks are unlikely to completely reverse. Rather, an ongoing adaptation of brain plasticity may result in a new (sometimes fragile) homeostatic state. These findings are used as the framework to propose the homeostatic disequilibrium model as a contributor to LLD recurrence risk. Andreescu et al.² postulate that in most individuals with LLD, successful antidepressant treatment results from recovery of brain network functioning, a measure of return to stable homeostasis. However, the authors posit that in some individuals with remitted depression, an unstable homeostatic state exists, which manifests itself as persistent brain network alterations and residual physiological, cognitive and behavioral changes. These contribute to further allostatic load and homeostatic disequilibrium resulting in a recurrent LLD episode.

Andreescu et al.² discuss the novelty of using ecological momentary assessment (EMA)⁷ for early clinical detection of homeostatic disequilibrium in individuals with remitted LLD. EMA approaches, which are increasingly used in mood disorders research,⁸ involve repeated “real-time” data capture of experiences in the respondents’ natural environment during their day-to-day routines, thus substantially reducing the recall bias of retrospective studies. With increasing access to technology, EMA may allow prospective monitoring of clinical symptoms and physiological changes, and could be a powerful clinical tool to detect impending recurrence. These methods could also lead to the development of ecological momentary interventions (EMIs)⁹ with the aim of reducing LLD recurrence risk. EMIs refer to the delivery of interventions via hand-held mobile technologies while individuals are engaged in their typical daily routines. Needless to say, these innovative approaches could have high clinical relevance, and may prevent detrimental LLD-related consequences.

Though challenges and caveats exist as pointed out by Andreescu et al.,² the proposed model could stimulate LLD research in early detection of high-risk remitted individuals who are candidates for aggressive treatments, and in identification of new mechanistic targets for future preventive interventions. Novel EMA and EMI approaches could change the manner in which we monitor and treat individuals with LLD who have achieved remission, and ultimately prevent recurrence.