The etiology and pathogenesis of major depressive disorder (MDD) symptomatology is complex and not sufficiently understood. Adult subjects regularly exhibit affective, cognitive, behavioral, psychomotor, and somatic symptoms [1
]. The association between MDD and somatization is common and clinically highly relevant [2
]. Individuals with MDD frequently suffer from pain, nausea, constipation, heart palpitations, shortness of breath, dizziness [2
], or a common sense of physical malaise [3
]. Meta-analytic evidence suggests an association between depression and functional somatic syndromes such as irritable bowel syndrome, non-ulcer dyspepsia, fibromyalgia, and chronic fatigue syndrome [4
]. Moreover, functional somatic symptoms are positively correlated with depression severity [5
]. Thus, there is clear evidence for clinically relevant associations between MDD and functional somatic symptoms that point to disturbed afferent bodily signals [7
]. It has been shown that impaired interoception is associated with a variety of mental disorders [8
]. This correlation between mental functions and the body is the core concept of embodiment theories [9
] and might enable a deeper understanding of psychosomatic processes in MDD [11
]. Therefore, this review focuses on the interoceptive system in order to shed new light on psychopathology and potential novel targets for the treatment of MDD. However, it is not in the scope of this review to determine if disturbed processing of afferent bodily signals is a cause or the result of mental disorders.
Interoception has been defined in various ways since Sherrington introduced the term in his seminal work [12
]. In this review, we favor the broadly accepted definition by Craig [14
] defining interoception as the sense of the physiological condition of the body. Accordingly, interoception comprises sensations from the entire body (e.g., thirst, hunger, itch, dyspnea, sexual arousal, heartbeat, distensions of bladder, stomach, rectum or esophagus, as well as temperature, pain, affective touch, etc.) and not just from the viscera. Craig’s definition is based on findings of functional neuroanatomy and relates to a sensory interoceptive pathway that conveys internal bodily signals to the brain where they are processed and integrated at the conscious or unconscious level [15
]. The lamina I spinothalamocortical pathway constitutes the complementary afferent part to the efferent autonomic nervous system [14
]. It originates in the small-diameter afferent fibers (Aδ and C) that carry sensory information about the physiological condition of virtually all bodily tissues. The sensory inputs are conveyed to the superficial dorsal horn of the spinal cord, where they are relayed to the brainstem. Ascending projections provide thalamocortical representations of the physiological condition of the body. Neuroimaging studies have shown that the activity of anterior insular cortex is correlated with awareness of the physiological condition (e.g., appetite during hypoglycemia) as well as bodily feelings [16
]. Therefore, the insula has been referred to as the primary interoceptive cortex [16
]. Afferent bodily signals serve as a basis for several physiological and psychological functions. Particularly, interoception and the related lamina I spinothalamocortical pathway are closely associated with homeostasis. As a result, interoception motivates behaviors that face homeostatic needs [18
]. Furthermore, interoception is closely associated with emotions, as well as cognitive functions [18
]. It has been suggested that unspecific states like subjective well-being are built upon an integration and evaluation of interoceptive signals that are represented in the insula and associated structures [14
]. Additionally, active Bayesian inference models of interoception hold that interoceptive experience is shaped by brain predictions which are based on prior beliefs and expectations about bodily states. Prediction errors, defined as a mismatch between afferent inputs and brain’s predictions, are suspected to underlie various mental disorders, e.g., anxiety and depression [21
]. The relevance of predictive coding models for psychiatry has been discussed in detail elsewhere [15
Garfinkel et al. proposed three different dimensions of interoception which have been quantified with distinct methodological paradigms [25
]. This review focuses on interoceptive accuracy (IAc), also referred to as interoceptive sensitivity, which is defined as the “objective accuracy in detecting internal bodily sensations” [26
] (p. 67). Other facets of interoception comprise interoceptive sensibility (“self-perceived dispositional tendency to be internally self-focused and interoceptively cognisant”, p. 67) and a metacognitive dimension of interoception. IAc is commonly assessed using heartbeat tracking [27
] or heartbeat discrimination tasks [28
]. Schandry’s heartbeat tracking method has been preferred in psychiatric research due to its ease of use. Briefly, the task involves the following: A researcher asks individuals to count the number of their heartbeats by focusing on physical sensations in repeated trials at different time intervals. Reported number of heartbeats in each trial are compared to the objectively assessed number of heartbeats recorded by electrocardiography. An error index is calculated and averaged across trials that represents an outcome measure of the heartbeat perception accuracy for each participant [27
Several neuroimaging studies investigated brain function during the heartbeat perception task (for a meta-analysis, see [29
]). For example, Critchley et al. [20
] demonstrated that an activation of the right anterior insular cortex predicted higher accuracy on a heartbeat perception task. Respectively, the authors showed that grey matter volume of the right anterior insular cortex predicted higher IAc in healthy adults. However, it has been shown that insula activity negatively correlates with depression and somatic symptom severity during a heartbeat perception task in subjects with MDD [30
]. Taking these findings together, current neuroscientific knowledge suggests an involvement of the insula on heartbeat perception with MDD subjects showing hypoactivation of the insular cortex.
In summary, we will draw two conclusions: First, the hypoactivation of the insula in depressed individuals during a heartbeat perception task suggests that MDD is associated with altered interoceptive awareness of the cardiovascular system. Second, the complex psychosomatic symptomatology of MDD points to disturbed interoceptive processing [31
]. There is preliminary evidence suggesting that MDD is associated with altered IAc during a heartbeat perception task [21
]. However, previous literature reviews have been subject to main limitations since they were narrative or lacked a systematic search strategy according to recommendations of the PRISMA statement [35
]. Thus, it is unclear if available reviews have covered the present state of research. Therefore, we systematically reviewed studies that assessed performance on heartbeat perception tasks in subjects with MDD. The main research questions of this systematic review were:
1. Do differences exist between subjects with MDD and other clinical samples as well as healthy control groups regarding accuracy on a heartbeat perception task?
2. Which type, and strength of association, exists between depression severity and performance scores on a heartbeat perception task?
We further aimed to explore associations between MDD-related symptoms and interoception as well as interoception altering effects of antidepressants in retrieved studies to identify possible confounding effects.