Discussion

We have presented the most extensive systematic review published on brain imaging in catatonia, including the analysis of 137 case reports (161 individual patients) and 18 group studies representing 186 individual patients with catatonia.

The analysis of case reports highlights that very disparate lesions can be associated with catatonia, including diffuse or focal abnormalities, which can occur in almost any brain region. Brain imaging is abnormal in the majority of cases, but abnormalities more frequently diffuse than localised. Most of the case reports of functional imaging (PET and SPECT) are abnormal, usually showing frontal, temporal, or basal ganglia hypoperfusion. In numerous cases, a further functional neuroimaging assessment after clinical improvement showed at least partial normalisation, as also illustrated by Escobar's SPECT study (Escobar et al., Reference Escobar, Rios, Montoya, Lopera, Ramos, Carvajal and Herrera2000).

Group studies were also analysed. Two head CT studies (Joseph et al., Reference Joseph, Anderson and O'Leary1985; Wilcox, Reference Wilcox1991) showed the potential implication of the cerebellum in the pathophysiology of catatonia and the CT head study of Northoff et al. (Reference Northoff, Waters, Mooren, Schlüter, Diekmann, Falkai and Bogerts1999d) showed a more extensive diffuse cerebral atrophy in schizophrenia patients with catatonia than in other schizophrenia patients. To date, structural brain MRI studies are very scarce in the catatonia literature, mostly showing diffuse cerebral atrophy. The majority of the case reports show diffuse lesions of white matter, in a wide range of brain regions. SPECT group studies show hypoperfusion in frontal and parietal regions, correlated with the severity of catatonia. Hypoactivation in a ¹²³I iomazenil SPECT study suggested implication of the GABAergic system (Northoff et al., Reference Northoff, Steinke, Czcervenka, Krause, Ulrich, Danos and Bogerts1999c). An fMRI study of Scheuerecker et al. (Reference Scheuerecker, Ufer, Käpernick, Wiesmann, Brückmann, Kraft and Meisenzahl2009) shows hypoactivation of the contralateral motor cortex during spontaneous motor task with hypoactivation of SMA in asymptomatic schizophrenia patients who have had a catatonic episode. Northoff's group studies show an abnormal regulation of negative emotions with hypoactivation of OFC and hyperactivation of median PFC, associated with GABAergic dysregulation after a catatonic episode. The studies of schizophrenic patients that have included subjects with catatonia suggest that hypodopaminergia may be present in catatonia, but no studies have investigated this hypothesis as yet.

However, our study suffers from several limitations. Concerning clinical cases, there is the potential for publication biases in the literature. Case reports are usually published to discuss a clinical issue of interest. In the case of catatonia, this increases the probability of presenting a case in which brain imaging is abnormal. The few structural imaging group studies we have reported confirm the high frequency of brain abnormalities in this population. However, high quality data lack to accurately assess the proportion of abnormalities found on brain imaging for this indication. In terms of group studies, we used the keyword ‘imaging’, which exposes us to the risk of missing articles that would only include the imaging modality in their title or abstract (head CT, fMRI…). However, a careful search within the reference sections allowed us to identify a large number of publications. Finally, another limitation, which concerns both clinical cases and group studies, is the use by the authors of different diagnostic criteria. As we pointed out in the introduction, diagnostic criteria have an influence on the estimated prevalence of the pathology. Overall, this limits the comparability of the different studies we have presented, also limiting the possibility of extrapolating mechanistic hypotheses.

Brain imaging studies have allowed authors to elaborate hypotheses of the pathophysiology of catatonia. In the seminal model developed by Northoff (Reference Northoff2000, Reference Northoff2002a, Reference Northoffb), catatonic symptoms are mainly associated with orbitofrontal-prefrontal/parietal cortical dysfunction and an abnormal ‘horizontal’ cortical-cortical modulation. According to this model, catatonia is associated with GABA-ergic orbitofrontal deficits following negative emotional processing; there is abnormal connectivity between prefrontal and motor areas, resulting in a frontal lobe syndrome and motor anosognosia (i.e. that the patient is unaware of his or her own motor activity). This dysconnectivity could lead to alteration of the caudate and other basal ganglia nuclei via cortical-subcortical loop dysfunction. This hypothesis is based on the fact that, according to Northoff, the SMA is not impaired in catatonia, unlike in ‘pure’ motor cortical-subcortical conditions like in Parkinson's disease or in neuroleptic malignant syndrome (NMS). Moreover, Northoff et al., have demonstrated that emotional dysregulation is associated with OFC hypoactivation and he also argues that there are clinical differences between the cogwheel rigidity of Parkinson's disease and the waxy flexibility of catatonia. According to this model, the motor anosognosia of catatonia is associated with dysfunction of the right ventrolateral and dorsolateral prefrontal cortex and right posterior parietal cortex dysfunction. This model can also explain the importance of the frontal symptoms (such as stereotypies, echolalia and echopraxia) in catatonia. In this model, the reduction in the GABA-A receptor density in patients with catatonic syndrome and the efficacy of benzodiazepines suggest GABA-ergic dysfunction.

However, there are several limitations to Northoff's model. Indeed, studies have found hypoactivation of the SMA in catatonic patients with schizophrenia both in the resting state and in spontaneous movements (Scheuerecker et al., Reference Scheuerecker, Ufer, Käpernick, Wiesmann, Brückmann, Kraft and Meisenzahl2009; Walther et al., Reference Walther, Schäppi, Federspiel, Bohlhalter, Wiest, Strik and Stegmayer2016). Interestingly, as mentioned in the introduction, clinical observations after lobotomies and frontal lobe lesions, and experiments performed on non-human primates have highlighted the specificity of SMA dysfunction in the occurrence of waxy flexibility and negativism (Dziadziuszko et al., Reference Dziadziuszko, Szurowska, Pienkowska, Jassem and Dziadziuszko2014; Joseph, Reference Joseph1999; Vancaester & Santens, Reference Vancaester and Santens2007). According to Walther, hyperactivation of the SMA in catatonia is secondary to blockade of the cortical-subcortical loop, at a subcortical level. The pro-catatonic effects of antidopaminergic drugs and PET studies suggesting reduced dopamine in basal ganglia nuclei also support the importance of subcortical dysfunction in catatonic syndrome. Interestingly, these two main hypotheses about catatonia (cortical or subcortical dysfunction) are based mainly on pharmacological and clinical arguments and the brain imaging results seem to account for a limited part of the argumentation, especially in Northoff's model. Moreover, these models have been criticized because lesions of diverse causes and localisations, including the frontal lobe, limbic system, brain stem, cerebellum and basal ganglia have also been reported to produce catatonia (Fig. 2). In fact, the brain imaging studies analysed in this paper show that lesions of frontal regions and subcortical structures such as the basal ganglia can both lead to a catatonic syndrome. In the case reports collected here, some patients have only subcortical lesions, and others have only frontal lesions, with the same apparent catatonic syndrome (see supplementary materials). Indeed, we believe that catatonia is still not ready for a unified theory (Bearden & Monterosso, Reference Bearden and Monterosso2002).

Fig. 2.

Simplified Northoff's model. Alterations in the GABA-ergic mediated deficit in orbitofrontal cortex (OFC) following negative emotional processing (white arrow) and associated with an abnormal connectivity between frontal and motor regions (striped arrow) are responsible for a frontal syndrome and for motor anosognosia. Secondary, this dysconnectivity lead to alteration of caudate and other basal ganglia through a cortical-subcortical loop dysfunction (white arrows).

A holistic hypothesis in which catatonic syndrome could be the result of two different pathways can be considered, i.e. that both frontal cortex dysregulation and ‘vertical’ abnormal modulation of cortical-subcortical loop or cortical-subcortical primitive dysfunction can lead to catatonia. Either or both pathways can be affected. This framework allows us to give a full account of the diversity of causes of catatonic syndrome, which is considered as a final common pathway. Interestingly, Walther and colleagues recently suggest that the more recent brain imaging findings indicate an alteration in the motor system in catatonia characterised by hyperactivity within the SMA and preSMA, arguing that these structures are involved in motor control, movement selection, initiation, timing, and inhibition. According to their findings, these researchers suggest that SMA hyperactivity could result from different pathways (increased feedforward stimulation of the subthalamic nucleus, activation of other cortical areas exerting inhibitory control, or from an attempt to overcome inhibitory processes) (Walther et al., Reference Walther, Stegmayer, Wilson and Heckers2019).

Twenty years after Northoff's hypothesis, we are still far from proposing a fully integrated framework explaining catatonia and it seems difficult today to extract an integrative pathophysiological model of catatonia from structural and functional abnormalities of such a different nature. In fact, brain imaging studies published in the literature have serious limitations. First, most of the studies were performed on patients who had experienced catatonic syndrome but were no longer symptomatic. We may have some doubts about the appropriateness of these studies to hypothesise regarding the pathophysiology of catatonia. The abnormalities found in this kind of study could be associated with the physiopathology of catatonia, but one might also suggest that they could be sequelae of the catatonia episode, or that they could be associated with the underlying disorder more than a manifestation of catatonia itself; they may be trait markers of catatonia more than a state marker. Second, most of the studies have been carried out on patients suffering from schizophrenia. Yet, we still do not know if the pathophysiology of catatonic syndrome in schizophrenia is the same as that in other disorders. Third, some studies had no appropriate controls, comparing catatonic schizophrenia patients with healthy subjects (instead of non-catatonic schizophrenia patients). Fourth, most of the studies (except one of Walther's studies) were performed with patients with the akinetic form of catatonia, while excited forms have been almost entirely neglected. Due to these limitations, brain imaging studies provide very little information on the actual pathophysiology of catatonic syndrome, and the main pathophysiological hypothesis are today the same as those that were proposed before the brain imaging era (Baruk & de Jong, Reference Baruk and de Jong1930). Given the growing literature of cases of inflammatory disorders associated with catatonia and the specific association with NMDAR encephalitis, the role of brain inflammation and glutamatergic dysfunction has been suggested as a novel line of enquiry (Rogers, Pollak, Blackman, & David, Reference Rogers, Pollak, Blackman and David2019). Recommendations for future research are summarised in Box 2.

Finally, we consider that clinical and diagnostic issues remain to be resolved in order to advance the understanding of the pathophysiology of catatonia. Hirjak and colleagues (including Northoff as the last author) recently published a review of brain imaging studies in catatonia. According to these authors, brain imaging studies using criteria based on behavioural and motor symptoms mainly found implication of cortico-subcortical pathway for the pathophysiology of catatonia. Meanwhile, brain imaging studies using criteria including behavioural, motor, and also affective symptoms (like the NCRS) mainly highlighted the importance of cortical regions like the orbito-frontal cortex and long-distance communication between cortical regions. These regions are largely modulated by GABAergic system, which is for the authors in line with the empirical observation of the efficacy of GABAergic treatment in catatonia (Hirjak, Kubera, Wolf, & Northoff, Reference Hirjak, Kubera, Wolf and Northoff2019b). It is difficult to objectively weigh the evidence from subtle and inconclusive neuroimaging findings against very clear clinical observations. This discrepancy between different brain imaging studies raises the debate of what catatonia really is. We also believe that to understand and illustrate this discrepancy, it is important to raise the question of the nature of NMS. In fact, for numerous authors, NMS appear as a manifestation of catatonia in a potentially lethal drug-induced form (Carroll, Reference Carroll2000; Fink, Reference Fink2013; Fricchione, Mann, & Caroff, Reference Fricchione, Mann and Caroff2000). For Northoff, NMS differs from catatonia by lacking the affective dimension (Northoff, Reference Northoff2002a). This important conceptual perspective has a major impact on models that we can conclude from brain imaging studies. It is also worth noting that using DSM-5 criteria, traditionally using an atheoretical approach, and without any affective symptoms for the definition of catatonia, NMS appears as form of catatonia (American Psychiatric Association et al., 2013).

In conclusion, catatonia is often associated with structural brain imaging abnormalities, such as diffuse white matter abnormalities and frontal lobe or basal ganglia abnormalities. During catatonic episodes, functional imaging shows hypometabolism in the frontal and temporal lobes, with hyperactivation of the SMA in catatonia associated with schizophrenia. This is consistent with work showing that psychomotor poverty in schizophrenia is associated with reduced regional cerebral blood flow in an extensive region of the pre-frontal cortex. (Liddle et al., Reference Liddle, Friston, Frith, Hirsch, Jones and Frackowiak1992) These abnormalities seem to be alleviated after successful treatment of clinical symptoms. Group studies assessing brain imaging after catatonic episodes show that emotional dysregulation is related to the GABAergic system, with hypoactivation of the OFC, hyperactivation of the medial PFC, and dysconnectivity between frontal and motor areas. Asymptomatic patients with catatonia who have had catatonic symptoms may have hypoperfusion in the contralateral motor cortex during spontaneous movement. Finally, most of brain imaging studies have serious limitations. Catatonia is a recently described syndrome that still fascinates clinicians and researchers. It is a complex, severe, and frequent condition. Nevertheless, the understanding of its pathophysiology remains in its infancy. Brain imaging studies published so far suffer from serious limitations and for now the different models presented in the literature do not explain most of the cases. There is an important need for further studies. We propose here a new integrative model, which could guide future research endeavours.