Atypical & Neuropathic Facial Pain

1.The Spectrum, and Why Labels Matter

Facial pain that is not classic TN spans several overlapping entities. Postherpetic trigeminal neuralgia follows zoster, most often in the ophthalmic division, and is a deafferentation pain. Anesthesia dolorosa is pain in a numb territory, the feared sequela of an overly aggressive ablative procedure for TN — a self-inflicted deafferentation syndrome. Trigeminal neuropathic pain follows identifiable nerve injury, from trauma, surgery, or tumor. And atypical facial pain — better termed persistent idiopathic facial pain — is a diagnosis of exclusion that should be approached with great caution.

2.The Organizing Rule

For these pains, the decision logic inverts the TN framework. In TN, you add a lesion to interrupt paroxysmal transmission. In deafferentation pain, the generator is the disinhibited dorsal horn (or its trigeminal homologue), and the question becomes whether to ablate that generator or to modulate the network around it. The practical rule runs from least to most destructive. Where peripheral trigeminal function is at least partly preserved, a reversible peripheral trigeminal nerve stimulation trial is a reasonable first surgical step. Motor cortex stimulation (and, less often, deep brain stimulation) is the non-destructive option when the pain is diffuse, when the nerve is too deafferented to generate a useful paresthesia, or when peripheral stimulation has failed. The nucleus caudalis DREZ operation is the most powerful ablative option, reserved for focal, well-characterized deafferentation facial pain — postherpetic, anesthesia dolorosa, trigeminal neuropathic — in which a one-way destructive procedure is justified. The ordering matters because added destruction can deepen a pain that is already a disorder of deafferentation.

3.The Rationale and the Target

The dorsal root entry zone procedure was first developed for the deafferentation pain of brachial plexus avulsion, then applied to the face. Facial pain and temperature sensation are carried through the trigeminal nerve and gasserian ganglion to the nucleus caudalis, the most caudal subdivision of the spinal trigeminal nucleus, extending from the obex down to roughly the C2 level. After deafferentation, the second-order neurons of the nucleus caudalis become hyperactive and generate pain — so destroying that hyperactive zone, just as a spinal DREZ lesion destroys the hyperactive dorsal horn, is the rationale. The nucleus is somatotopically organized along its length, and two different maps must be kept apart. Peripherally — in the nerve, the gasserian (trigeminal) ganglion, and the post-ganglionic root — sensation is arranged by division, giving the familiar V1/V2/V3 territories, so a peripheral or ganglion-level lesion produces loss confined to a division. Centrally, the spinal trigeminal nucleus follows a concentric “onion-skin” (Dejerine) pattern: the perioral and central face (lips, nose, mouth) is represented most rostrally, near the obex, while the peripheral face and scalp (forehead, lateral cheek, preauricular region) are represented progressively more caudally, toward C2 — so a lesion at a given rostrocaudal level produces a ring-shaped sensory change that crosses divisional boundaries rather than respecting them.

After exposure of the lower brainstem and upper cervical cord, a right-angled DREZ electrode (developed specifically to reduce injury to the adjacent dorsal spinocerebellar tract) is used to place a series of radiofrequency lesions along the nucleus between the obex and C2. Historically this meant two rows of lesions roughly 1 mm apart over a 15–20 mm length — commonly 20 to 30 lesions — with thermal lesions on the order of 75 °C for 15 s each.

4.Surgical Technique: The Steps of the Operation

The operation is performed under general anesthesia with the patient prone, the head flexed and fixed in a skull clamp to open the craniocervical angle and present the cervicomedullary junction. Anesthesia follows intraoperative-monitoring principles: after induction, total intravenous anesthesia (propofol-based, with an opioid infusion) is maintained, halogenated inhalational agents are minimized or avoided because they attenuate evoked responses, and — critically — neuromuscular blockade is not continued beyond intubation, so that limb electromyography stays responsive for the corticospinal-tract stimulation checks described below. Suppressing the monitoring would remove the very safety signal the operation depends on.

Exposure is through a small suboccipital craniectomy and a C1 (with or without C2) laminectomy. The dura is opened in the midline, the cerebellar tonsil is gently elevated, and the dorsolateral cervicomedullary junction is brought into view. The surgeon then identifies the key landmarks that frame the target: the obex rostrally, the dorsal root entry of C2 caudally, the dorsolateral sulcus, and the emerging accessory nerve rootlets. The nucleus caudalis occupies the triangular zone between the dorsolateral sulcus and the line of the rootlets, tapering caudally as it joins the cervical dorsal horn.

Lesions are then placed along this trajectory with the angled DREZ electrode, typically as a serial row at roughly 1 mm intervals extending from a few millimeters above the obex down toward the C2 rootlets, at a controlled depth of about 3–4 mm and angled into the dorsolateral sulcus rather than straight down (to follow the nucleus and stay away from the corticospinal tract). Each radiofrequency lesion is made on the order of 75–80 °C for 15 s. Watertight dural closure and a standard layered suboccipital closure complete the procedure. The number and exact placement of lesions is where intraoperative monitoring transforms the operation — rather than blanketing the whole nucleus, the surgeon lesions only the physiologically confirmed, pain-corresponding sites, which is covered next.

5.The Neighboring Tracts

The nucleus caudalis sits in dangerous company, and understanding the adjacent anatomy is the whole story of this operation's complication profile. Three structures lie immediately nearby: the corticospinal tract (anterior in the medulla, but immediately ventral to the nucleus in the upper cervical cord), the dorsal spinocerebellar tract (immediately dorsolateral), and the dorsal column (immediately dorsomedial). Injure the corticospinal tract and you cause hemiparesis; injure the dorsal spinocerebellar tract or dorsal column and you cause ataxia. In the older landmark-only series, these complications were common — ataxia in roughly a third to over half of patients in the early experience, with persistent ataxia and motor problems in a substantial minority at follow-up. The high complication rate is precisely why this operation has remained in the hands of a few experts.

6.Intraoperative Neurophysiological Monitoring — the Heart of the Operation

This is the portion of the operation that determines whether it succeeds safely, and it deserves to be understood in detail. The goal of monitoring is twofold: to localize the divisions of the nucleus that correspond to the patient's pain, so that lesions are made only where they are needed, and to protect the adjacent corticospinal tract and dorsal column from injury. Done well, it allows the surgeon to achieve the same pain relief as the old high-count technique with a small fraction of the lesions — dramatically reducing the complication rate.

Setup. After anesthesia and before lesioning, stimulating needle electrodes are placed to selectively activate each trigeminal division and a control somatic nerve: the supraorbital nerve (activating V1/ophthalmic), the infraorbital nerve (V2/maxillary), the mental nerve (V3/mandibular), and the median nerve at the wrist (to activate the dorsal column pathway). Multichannel electromyographic electrodes are placed in upper- and lower-extremity muscles ipsilateral to the surgery to detect any motor activation.

Localization by trigeminal evoked potentials. The right-angled DREZ electrode itself is used as the recording electrode, connected to an evoked-potentials machine. A reference grid is established — the point just posteromedial to the entry of the rostralmost C2 rootlet is labeled the 0/0 origin, with movement recorded in millimeters along horizontal and vertical axes. Each trigeminal division is stimulated in turn (supraorbital, infraorbital, and mental nerves at low intensity, on the order of 5 mA, 0.1 ms, around 2 Hz), and the DREZ electrode is moved across the nucleus, recording the evoked response at each coordinate. The site producing the largest-amplitude response to the nerve serving the painful division is the target. If pain involves V1 and V2, the surgeon lesions where supraorbital and infraorbital responses are maximal; if V3, where the mental response is maximal.

Protecting the dorsal column. The median nerve response is the key safety signal for ataxia of dorsal-column origin: it localizes the dorsal column so the surgeon knows precisely where not to lesion. The dorsal column representation is identified and deliberately avoided. A useful practical refinement: because the dorsal column and the nucleus run roughly parallel along the cervicomedullary junction, the median-nerve dorsal-column response can be tracked continuously and held as a constant medial reference as lesioning moves rostrally toward the obex — exactly where the C2 rootlet landmarks are no longer available. Keeping that reference steady helps guard against drifting medially into the dorsal column.

Protecting the corticospinal tract — the pre-lesion check. This is the single most important safety maneuver, and it should be done before every lesion. Once a candidate target is chosen, stimulation is delivered through the lesioning electrode at the target site (around 2 Hz, intensity increased gradually to 1 V) while watching the limb EMG channels. If compound muscle action potentials appear in any muscle group, the electrode is too close to the corticospinal tract: do not lesion. Withdraw, reposition, and re-test until no motor activation is seen, then lesion. This simple stimulate-before-you-burn discipline is what prevents the hemiparesis that the proximity of the corticospinal tract would otherwise threaten.

7.What Monitoring Buys, and Its Limits

The payoff is substantial. With physiological mapping, the same clinical relief that once required 15 to 20 lesions can be achieved with only a handful targeted precisely at the painful divisions — immediate relief in the high-80s percent and sustained relief in roughly 70% at intermediate follow-up, with markedly fewer complications than the landmark-only era. Two honest limitations remain. First, the technique depends on preserved peripheral trigeminal function: when the nerve is severely deafferented — advanced postherpetic neuralgia, anesthesia dolorosa after multiple prior procedures — the evoked responses from the painful division may be absent, and the surgeon must fall back on responses from adjacent divisions and known anatomical relationships, with the corticospinal-tract stimulation check as the one remaining real-time safety control. Second, there is currently no reliable way to monitor the dorsal spinocerebellar tract directly, so ataxia from that tract cannot be fully designed out; the right-angled electrode and precise targeting remain the main defenses.

The order in which these options are offered matters as much as the options themselves. Because deafferentation pain is so easily worsened by further destruction, the reversible, non-destructive neuromodulatory approaches are generally tried before an open ablative operation. In practice that means a course of peripheral neuromodulation is often the first surgical step in a patient with preserved — or only partly lost — trigeminal function, with motor cortex stimulation and the open nucleus caudalis DREZ held in reserve for failures.

8.Peripheral Trigeminal Neuromodulation — the Usual First Step

Before escalating to a central operation, many patients with refractory trigeminal neuropathic pain, postherpetic neuralgia, or persistent idiopathic facial pain are offered peripheral nerve / nerve-field stimulation of the affected trigeminal branch. Thin subcutaneous leads are tunneled along the symptomatic distribution — most commonly the supraorbital nerve for V1 pain, the infraorbital nerve for V2, and occasionally a mental or nerve-field array for V3 — positioned to generate a field of paresthesia that overlaps the painful territory. The mechanism is modulatory rather than destructive: the leads sit on intact peripheral nerve and add no sensory loss.

The decisive practical advantage is that peripheral neuromodulation is inherently testable. Leads are placed and externalized for a trial of roughly one to two weeks; only a patient who achieves meaningful relief (conventionally at least 50% pain reduction) and tolerates the paresthesia proceeds to a permanent implanted system. This trial gate, the reversibility, and the absence of added numbness are exactly why peripheral stimulation is a reasonable first surgical move where an ablative procedure would be a one-way commitment. When peripheral stimulation fails or cannot capture the pain, escalation to motor cortex stimulation or — for a focal, mechanistically suitable target — the nucleus caudalis DREZ becomes the next consideration.

Selection. The best candidates have focal neuropathic pain in a mappable peripheral-branch territory — post-traumatic or post-surgical neuropathy of the supraorbital or infraorbital nerve, trigeminal postherpetic neuralgia, trigeminal neuropathic pain, persistent idiopathic facial pain, and atypical (type 2) trigeminal neuralgia — in whom secondary and structural causes have been excluded by imaging. A practical targeting clue is to find the hyperalgesia/allodynia strip at the border of the painful zone, the territory the lead should cover. Candidacy is confirmed only by a successful externalized trial (≥ 50% relief). Relative contraindications mirror other implanted neuromodulation: active infection, untreated coagulopathy, an inability to manage the device, and uncontrolled psychiatric or secondary-gain factors; pain that cannot be captured by paresthesia or has no discrete branch territory is unlikely to respond.

9.Motor Cortex Stimulation

Motor cortex stimulation (MCS) is the central neuromodulatory option for deafferentation facial pain, and in many centers it is preferred specifically for trigeminal neuropathic pain and anesthesia dolorosa — the very situations where adding more ablation to an already deafferented system is least appealing, and where the trigeminal nerve may be too damaged for peripheral stimulation to generate a useful paresthesia. An epidural electrode is placed over the facial representation of the motor cortex; stimulation modulates the pain network without any destructive lesion and without adding sensory loss. The mechanism is incompletely understood, but the clinical niche is clear: MCS is reversible, adds no numbness, and is well suited to the patient whose face is already numb and painful. It is the modulatory answer when the ablative answer would only deepen the deafferentation.

10.Deep Brain Stimulation and Ganglion/Rootlet Stimulation

Deep brain stimulation has a long but more limited history in facial and craniofacial pain, targeting sensory thalamus or periventricular/periaqueductal gray in selected refractory cases. Stimulation of the trigeminal ganglion or rootlets is another modulatory option in patients with preserved trigeminal sensibility. These are second- and third-line tools, reserved for pain that has defeated the better-established options.