In view of these points, the amygdala may be considered a key structure in the etiopathogenesis of autistic symptoms and in mediating aggressive behavior. However, neither ablational surgery nor functional imaging have sufficient spatial resolution for correlating their results with the subnuclear organization of the amygdala and associated areas. This makes it difficult to predict the consequences of stimulation of amygdaloid subnuclei with respect to SIB and symptoms associated with autism.
In order to assess the specific amygdaloid areas stimulation of which was likely to improve the boy's symptoms, we implanted the DBS electrodes bilaterally at two different sites. One electrode targeted the paralaminar, the BL as well as the Ce nucleus in order to modulate the input (L-BL) and successive processing area (BL–Ce) of the amygdala. The trajectory of the other electrode was destined to contact the extended amygdala in order to interfere with efferent and afferent connections.
It turned out that only stimulation within the BL amygdala improved the symptoms sufficiently, whereas stimulation of any other target yielded minor and temporarily limited effects at best. This finding strongly supports the assumption that the area which has been stimulated effectively is functionally relevant in the context of the treated disorder and that unspecific effects due to possible mechanical irritation, microlesioning, or inflammation presumably do not account for the observed outcome.
The interpretation of the results of BL stimulation remains tentative considering the complexity of the segregated circuits connected through this structure which thus might have been affected. Moreover, stimulation may not be restricted to the BL nucleus but may also affect immediately adjacent parts of the lateral and the basomedial nucleus, respectively, as a result of current spread and stimulation of intermingling dendritic trees. These nuclei and their respective connections should therefore also be considered when evaluating the effects of stimulation.
The lateral nucleus
is the main amygdaloid target for already pre-processed multimodal information from higher order association cortices (Pessoa and Adolphs, 2010
). Subcortical afferents reach the lateral nucleus predominantly from chemosensory, visual, and auditory areas of the thalamus. Accordingly, the lateral nucleus is involved in the processing and evaluation of sensory stimuli reflecting emotional and socially pertinent content (reviewed in Davis and Whalen, 2001
; LeDoux, 2003
Most of these fibers entering the lateral nucleus are glutamatergic and terminate on principal neurons that also use glutamate as transmitter. These are under inhibitory control of different classes of GABA-ergic interneurons (Lang and Pare, 1997
; Yilmazer-Hanke et al., 2007
; Morozov et al., 2011
). Their activity plays a predominant role in balancing excitation and inhibition within the lateral nucleus. An increase in excitatory tone resulting from reduced GABAergic signaling might thus result in a hyper-excitable state causing emotional tensions, rage, temper tantrums, fear and anxiety, as well as cognitive dysfunctions (Harkin et al., 1998
; Hensch et al., 1998
; Schuler et al., 2001
), and—in extreme cases—epileptic seizures. These states are major characteristics comprising the autistic syndrome.
The BL nucleus
is functionally interposed between the lateral and CE and therefore regarded as a communication channel between both nuclei (see Freese and Amaral, 2009
). Besides its afferents from the lateral nucleus it has major reciprocal connections with the orbitofrontal and the anterior cingulate cortex. The concurrent processing of the input from these regions may be regarded as a gate control system, where the “danger” signal, as well as other emotionally and socially relevant signals, is evaluated in the context of conscious experiences or expectations. Its activity modulates the CE but also cortical areas that are involved in the regulation of various cognitive functions (Phelps, 2006
Hence, information propagation from cortex and subcortex through the lateral to the BL nucleus is presumably modulated by the above mentioned interaction between glutamatergic and GABAergic chemical synaptic transmission thus being particularly relevant for autism. Genetic studies/genome scans have revealed mutations in some patients with autism- associated symptoms. These genetic changes affect the transmission of both, GABA- and glutamatergic synapses (Hussman, 2001
; Jamain et al., 2002
; Derwińska et al., 2009
, reviewed by Rubenstein and Merzenich, 2003
; State, 2010
). Reduced GABAergic signaling of interneurons of the L and BL nuclei might yield a “hyperexcitability-state” of the respective nuclei as found by Markram et al. (2008
) and reviewed by Markram and Markram (2010
) in the rat-valproic acid-model of autism and in consequence impair both intraamygdaloid and amygdalo-cortical circuitry. This might well contribute to the pathogenesis of autism as proposed by the former authors and considering the well-established role of the amygdala in rage processing also of SIB and explain the beneficial effects of DBS in the BL nucleus which have been achieved in our patient possibly through interference of DBS at high frequencies with hyperexcitable cell ensembles in the intra-amygdaloid relay-nucleus.
Nevertheless, when discussing a clinical case, i.e., a human subject, we feel ethically obliged to duly take into account that it remains controversial, whether neuroscientific results obtained from animal experimentation in phylogenetically rather primitive species, especially in rodents with their simple brain architecture and their rudimentary neocortex, are readily transferable to apparently related conditions in the hypercomplex human brain, as critically remarked by Markram et al. (2008
; Markram and Markram, 2010
), although such results certainly do provide a valuable starting point for future in-depth studies.
However, at this point we would like to indicate that the reported results might not be restricted to the BL nucleus alone. It is thoroughly conceivable that these beneficial effects are also related to DBS-evoked modulations exerted by the BL nucleus on a distributed network involving not only the amygdala, but also other subcortical structures such as the striatum or other basal ganglia. There are strong indications for efficient connectivity between the striatum and the amygdala, as demonstrated by Popescu et al. (2009
) as well as for links with orbitofrontal regions. Recently Le Jeune et al. (2010
) have shown that DBS in the subthalamic nucleus affects limbic and associative circuits. Furthermore, it was demonstrated that such modulations were related to emotional disturbances (Péron et al., 2010
). Hence, these kinds of distant modulations, encompassing large-scale network effects, might also have had an important part in successfully treating the case reported here.
The improvement of core symptoms of autism by stimulation of the BL amygdala supports hypotheses ascribing the amygdala a dominant role in the pathogenesis of autism (Baron-Cohen et al., 2000
; Amaral et al., 2003
as reviewed by Bachevalier, 1994
; Schumann et al., 2006
; Markram et al., 2008
; Markram and Markram, 2010
). We cannot exclude, however, that these improvements might partly or even totally be due to psychosocial alterations, caused by reducing SIB and/or pathological anxiety through DBS.
Our finding that stimulation of other amygdaloid nuclei, especially of the CE as well as of the extended amygdala has been ineffective is surprising and cannot be explained at present.