Chronic Obstructive Pulmonary Disease (COPD) is a prevalent and debilitating respiratory disease, characterized by persistent and usually progressive airflow limitation (O'Donnell et al., 2007; GOLD, 2017). Dyspnea is the cardinal respiratory symptom of COPD, causing significant reductions in patients' exercise capacity and quality of life, and is frequently linked to comorbid anxiety and depression (Schlecht et al., 2005; Martínez Francés et al., 2008; Maurer et al., 2008; Roche, 2009; Yohannes and Alexopoulos, 2014; Waschki et al., 2015; GOLD, 2017). Dyspnea is usually experienced as highly aversive and threatening (American Thoracic Society, 1999; Parshall et al., 2012). Consequently, many patients avoid situations associated with dyspnea, especially physical activity (O'Donnell, 2006). This maladaptive avoidance behavior results in progressive deconditioning, ultimately increasing dyspnea at lower activity levels and contributing to disease progression (Reardon et al., 2006; Troosters et al., 2013; Waschki et al., 2015). In particular, the fearful anticipation of dyspnea is suggested to play a key role within this spiral of decline (Hayen et al., 2013).

The brain mechanisms underlying the perception and anticipation of dyspnea are still poorly understood. Previous neuroimaging studies in healthy volunteers have identified multiple dyspnea-related brain areas, presumably related to different aspects of dyspnea. Somatosensory and motor aspects are thought to be processed by sensorimotor areas (e.g., SM1, SII), while affective-motivational aspects are processed by cortico-limbic areas including anterior insula, anterior cingulate cortex (ACC), hippocampus, prefrontal cortex (PFC), and amygdala (von Leupoldt and Dahme, 2005; Davenport and Vovk, 2009; Evans, 2010). Preliminary findings suggest an involvement of several of these areas such as the insula, amygdala, ACC and the periaqueductal gray (PAG) during the fearful anticipation of impending dyspnea (Stoeckel et al., 2015, 2016; Faull et al., 2016). This converges with studies on other aversive states, e.g., fear, aversive learning, and pain, reporting activation in similar cortico-limbic structures (Peyron et al., 2000; Apkarian et al., 2005; Sehlmeyer et al., 2009; Wiech and Tracey, 2013). Unfortunately, neuroimaging studies with high spatial resolution (e.g., functional magnetic resonance imaging [fMRI]) on the anticipation and perception of dyspnea in COPD patients are widely absent, which greatly limits our understanding of the potential contribution of disease-specific brain mechanisms to the patients' burden.

However, COPD-specific brain mechanisms due to chronic dyspnea experiences seem highly plausible (Herigstad et al., 2011). For example, contemporary models on the effects of chronic pain suggest that brain activation in pain patients shifts with chronicity from sensory to emotion-related areas, leading to increased activation preferentially in cortico-limbic structures (Apkarian et al., 2011; Mansour et al., 2014). Moreover, enhanced activation was recently demonstrated in emotion-related brain areas (i.e., PFC, ACC) of COPD patients while reading dyspnea-related statements (Herigstad et al., 2015).

This fMRI-study aimed (1) to compare brain activation during the perception and anticipation of resistive-load induced increased dyspnea between patients with COPD and healthy control subjects and (2) to evaluate potential associations with patient characteristics. Analogous to findings on chronic pain (Apkarian et al., 2005, 2011; Mansour et al., 2014; Baliki and Apkarian, 2015), we hypothesized that during dyspnea perception, both groups activate similar dyspnea-relevant brain areas, with patients showing higher activation in cortico-limbic structures. Based on findings on fear, aversive learning, and dyspnea anticipation (Sehlmeyer et al., 2009; Wiech and Tracey, 2013; Stoeckel et al., 2015; Faull et al., 2016), we further hypothesized that patients show enhanced activation in emotion-related brain areas during dyspnea anticipation.

The present fMRI study compared brain activation during the perception and anticipation of resistive-load induced dyspnea between patients with COPD and matched healthy controls. Furthermore, we investigated the relationship between COPD-specific brain activation and patient characteristics.

During the perception of increased dyspnea, patients and control subjects showed widely comparable neural activation patterns in commonly observed dyspnea-relevant sensorimotor and cortico-limbic brain regions without significant differences between groups. In contrast, while anticipating increased dyspnea, enhanced neural activation was observed in patients relative to control subjects in the bilateral hippocampus and right amygdala. Notably, neural responses in these emotion-related limbic brain regions were correlated with clinical patient characteristics. Specifically, patients with higher levels of anxiety and dyspnea as well as lower exercise capacity showed higher anticipatory neural activation in the left hippocampus, whereas patients with lower health-related quality of life showed higher anticipatory neural activity in the right hippocampus and amygdala. Additional explorative findings indicated that patients with longer disease duration showed higher amygdala activation during the perception of increasing dyspnea. Taken together, these results suggest that patients with stable moderate-to-severe COPD show enhanced activation in emotion-related brain areas than healthy control subjects during the anticipation, but not during the perception of experimentally induced increased dyspnea, which further relates to important clinical characteristics.

The observed brain activation in sensorimotor (SM1, SII/operculum, SMA, and thalamus) and cortico-limbic (ACC, insula, and PFC) areas during the perception of increased dyspnea is in line with previous neuroimaging studies in healthy volunteers (Peiffer et al., 2001; Evans et al., 2002; McKay et al., 2003; von Leupoldt et al., 2008; Pattinson et al., 2009; Binks et al., 2014). Our findings extend the notion of a dual cortical pathway model of dyspnea perception with pathways subserving either sensorimotor or affective aspects of dyspnea to patients suffering from COPD (von Leupoldt and Dahme, 2005; Davenport and Vovk, 2009). Moreover, the present finding that patients and control subjects demonstrated activation in comparable dyspnea-relevant brain regions corresponds with a recent study by Higashimoto et al. using near-infrared spectroscopy (Higashimoto et al., 2015). Although limited in spatial resolution, this study also demonstrated comparable activations in pre-motoric areas between COPD patients and healthy controls. In addition, several studies on chronic pain syndromes similarly observed substantial overlap of brain activation during the perception of acute pain in healthy subjects and patients with several pain conditions (Derbyshire et al., 2002; Apkarian et al., 2005, 2011; Baliki et al., 2006). However, the absence of enhanced responses in emotion-related areas during perceived dyspnea in COPD-patients is contrasting with previous studies on chronic pain conditions such as fibromyalgia, irritable bowel syndrome, and back pain (Baliki et al., 2006; Jensen et al., 2009; Hashmi et al., 2013). For example, Hashmi and colleagues (Hashmi et al., 2013) demonstrated a shift toward more activation in emotion related areas in those patients suffering from lower back pain, who developed chronicity. The lack of comparable changes in our study might be related to differences between studies regarding sensory modalities (e.g., thermal pain vs. load induced dyspnea), experimental designs (with vs. without anticipation conditions; different control conditions), disease duration, or patient characteristics (see below) including different exacerbations phenotypes of COPD (Scioscia et al., 2017), which deserves further investigation.

Importantly, during the anticipation of increased dyspnea, patients with COPD compared to healthy control subjects showed enhanced neural activation in hippocampus and amygdala, both key areas in the processing of fear, aversive learning, and pain (LeDoux, 2003; Sehlmeyer et al., 2009; Wiech and Tracey, 2013). Recent findings in healthy individuals demonstrated activation of the amygdala during the fearful anticipation of resistive load induced dyspnea (Stoeckel et al., 2015). Similarly, Apkarian et al. (2011) reported increased amygdala activation prior to a pain stimulus peak in healthy subjects, indicating a role of the amygdala in the anticipation of impending pain. The importance of amygdala and hippocampus in pain-related fear acquisition and memory processes was furthermore confirmed for patients with chronic irritable bowel syndrome (Icenhour et al., 2015). According to the aforementioned model regarding brain circuitry involved in the transition from acute to chronic pain (Apkarian et al., 2011; Mansour et al., 2014), the hippocampus and the amygdala are key regions of the limbic circuitry crucially involved in emotional enhancement processes. This limbic circuitry is presumed to “translate” sensory signals into more emotional, cognitive suffering states, e.g., by modulating and amplifying the emotional character of aversive nociceptive input signals over time (Mansour et al., 2014). Our finding of prominent recruitment of hippocampus and amygdala in COPD patients during the anticipation of increased dyspnea most likely reflects such an amplified emotional, in particular more fearful, evaluation of upcoming dyspnea. This might provide, at least partly, the neural basis for subsequent fearful avoidance behavior in patients with COPD and contribute to a spiral of decline in form of activity avoidance, deconditioning, increased dyspnea and anxiety, and further reductions in quality of life (Reardon et al., 2006; Troosters et al., 2013). Support for this assumption comes from the present observation that increased activation of hippocampus and amygdala during increased dyspnea anticipation was closely related to important clinical characteristics such as reduced exercise capacity and quality of life as well as higher levels of dyspnea and anxiety. Notably, using structural MRI, morphological changes of reduced gray matter volume in hippocampus and amygdala were recently observed in patients with COPD relative to control subjects (Esser et al., 2016), which further supports the important role of these areas in COPD.

Finally, post-hoc analyses revealed a positive correlation between disease duration and amygdala activation during the perception of increased dyspnea. Although explorative in nature, this finding suggests that the chronic course of COPD results in a functional modulation of a brain region strongly related to fear and anxiety (LeDoux, 2003; Sehlmeyer et al., 2009; Wiech and Tracey, 2013). Interestingly, a recent study showed that longer COPD disease duration was also related to reduced gray matter volume in the ACC (Esser et al., 2016), a key structure for antinociception and the regulation of emotional states that is tightly interconnected with the amygdala (Tracey and Mantyh, 2007). Given the fact that many patients with COPD develop comorbid anxiety over time (Maurer et al., 2008; Yohannes and Alexopoulos, 2014), it might be speculated that this is related to increased amygdala activation during dyspnea anticipation and perception, paralleled by reduced antinociceptive ACC capabilities. However, future studies are required directly addressing this assumption.

When interpreting the present results, several limitations need to be taken into account. We exclusively studied a small group of highly motivated patients with stable health status and relatively good physical condition without clinical levels of comorbid anxiety and depression and no exacerbation history within the last year. This limits the generalizability of our findings and motivates future studies in other patient populations with larger sample sizes, different and/or more severe forms of COPD and related comorbidities. Moreover, the use of resistive-load breathing for the experimental induction of short-lasting dyspnea only mirrors some aspects of dyspnea, i.e., the sense of increased work and effort of breathing, and might not be comparable to sustained dyspnea experiences outside the lab. Therefore, future studies are necessary to extend the present findings to other qualities of dyspnea, for example air hunger, which might be perceived and/or anticipated differently (Banzett et al., 2008). Finally, future longitudinal studies are needed to directly address how changes in the neural processing of perceived and anticipated dyspnea relate to the subsequent development of activity avoidance, deconditioning, increased dyspnea and anxiety, and reduced quality of life.

This study suggests that patients with stable moderate-to-severe COPD show enhanced activation in emotion-related brain areas compared with healthy control subjects during the anticipation, but not during the perception, of increased resistive-load induced dyspnea. Brain activation in these areas was related to important clinical characteristics and might contribute to the development of a downward spiral including fearful activity avoidance, deconditioning, increased dyspnea and anxiety, and reduced quality of life. Taken together, our findings contribute to our understanding of brain processes in COPD and their relation with clinical outcomes and might provide potential targets for future psychological interventions aimed at improving the burden for patients.

RE, MS, HM, and AV contributed to the conception and study design; RE, MS, AK, HW, KT, and KL contributed to the data acquisition; RE, MS, CB, and AV contributed to the data analysis; RE, MS, and AV drafted the paper; all authors contributed to the interpretation of data, the editing of the paper, provided critical revisions and approved the final version of the manuscript.