**2. Anatomical Background**

### *2.1. Somatosensory System and Relevant Inputs*

Pain-sensing neurons, nociceptors, are pseudounipolar cells with somata, located in the dorsal root ganglion within the dorsal root of spinal nerves. Nociceptors were first described by Sherrington in 1906 [36]. Their afferent processes, either unmyelinated C fibers or myelinated Aδ fibers, project to peripheral tissues and typically end with free nerve endings capable of sensing painful mechanical or thermal stimuli. The efferent processes project to the thalamus, hypothalamus, and several midbrain areas, mainly via the contralateral anterolateral spinal cord. The disruption of information transmission along this tract via either surgical transection (first described by Spiller and Martin in 1912 [37]) or electrical stimulation has shown notable success in ameliorating some forms of chronic pain [38].

The descending pain modulatory pathway travels from the cerebrum to the ventrolateral periaqueductal gray (PAG) matter and rostral ventromedial medulla [39], and it projects to the dorsal horn of the spinal cord, as shown in Figure 2. The PAG matter in the brainstem contains bidirectional nociceptive pathways from medullary nuclei, allowing for modulation through both projections. As with other medullary nuclei, analgesia occurs via inhibitory serotonergic, enkephalin, mu-opioid, and GABA projections to the dorsal horn through descending pain control pathways. The other efferent nuclei from the medulla and pons appear to have a similar mechanism of analgesia. These are the same pathways thought to underlie opiate-induced analgesia [30]. Other notable efferents from the dorsal horn include the parabrachial nuclei, which project directly to the amygdala. The Kölliker-Fuse nucleus is found within the parabrachial nucleus and demonstrates noradrenergic descending inhibition [40]. Therefore, the parabrachial nuclei may be involved with salience or affective elements of pain [41,42].

**Figure 2.** Descending pain modulatory pathway. The pathway originates in the cerebrum and descends to the periaqueductal gray (PAG) matter, rostral ventromedial medulla, and projects to the dorsal horn of the spinal cord. Bidirectional nociceptive pathways through the medullary nuclei are shown. The efferent pathway from the dorsal horn goes through the parabrachial nucleus, amygdala, and thalamic nuclei.

Nociceptive projections ascend through the PAG and then project to the thalamic nuclei, which have medial and lateral nuclear groups [43]. The lateral group, consisting of ventroposterior nuclei, has small receptive fields and projects to the somatosensory cortex. Circuits involving these nuclei are therefore likely responsible for the conscious identification and localization of painful stimuli. The medial group consists of the central lateral nucleus and intralaminar complex, which have widespread projections to the basal ganglia and cortex, and therefore likely mediate the arousal response to pain.

Following the disappointing results of septal stimulation in the 1960s, electrical stimulation of the PAG was tried in humans in the 1970s [16,17]. In one of the earliest studies on PAG stimulation, Richardson and Akil [26] reported on 30 patients with electrodes implanted for patient-controlled self-stimulation. Two thirds of the patients reported good outcomes (reduction in pain of >50%). Periventricular stimulation showed similar promise around the same time, with Hosobuchi et al. [28] showing pain relief in five of six patients. Young et al. [44] showed that stimulation of the Kölliker-Fuse subnucleus alone or in combination with stimulation in the PAG matter or the somatosensory thalamic nuclei provided self-reported "excellent" pain relief in three of six patients with intractable pain. Following these, and other promising early studies, were two negative randomized controlled trials involving Medtronic devices. One trial, based on the Model 3380 electrode, was discontinued after the first-generation production. The other trial involved the Model 3387 electrode. This clinical trial was stopped because of slow enrollment, high attrition, and low efficacy [29].
