Once developed, trigeminal neuralgia (TN) is likely to have an exacerbating and remitting course. Over time, the pain-free intervals appear to diminish, and the pain becomes progressively more medically intractable. Temporary spontaneous remission may occur at any time, but permanent remission never occurs.
Without treatment, typical TN (TN1) may transform over time to become atypical TN (TN2), with a change in the character of the pain to more constant and background pain and the development of sensory impairment. Therefore, some authorities recommend early intervention to give the opportunity of pain relief without sensory deficits.
Over time, the drugs used for the treatment of TN often lose effectiveness as patients experience breakthrough pain. In some studies, more than 50% of patients with TN eventually had some kind of surgical procedure. Experience would indicate that medical management eventually fails in most patients with TN, and those patients undergo surgery. For patients in whom medical therapy has failed, surgery is a viable and effective option.
According to Dalessio, 25-50% of patients eventually stop responding to drug therapy and require some form of alternative treatment.  The clinician then may consider referral to a surgeon for 1 of the procedures discussed below. Among patients who develop TN when younger than 60 years, surgery is the definitive treatment.
Neurosurgery is generally more helpful in those patients with paroxysmal rather than constant pain and in patients whose pain follows the anatomic distribution of 1 or more trigeminal distributions rather than being spread diffusely. The various operations often fail after 1 or several years of initial relief. Such failure necessitates a repeat procedure, often with improved but still incomplete results. Thus, many patients eventually restart pain medication after surgery. Surgery appears to be less effective for TN secondary to multiple sclerosis (MS).
The timing for surgery is debatable, and no randomized study has addressed this question. However, it seems that the earlier a surgical technique is applied, the better the outcome is likely to be. At least 2 medication trials should be performed and carefully evaluated before more invasive techniques are instituted. 
The following clinical guidelines have been published:
International RadioSurgery Association - Stereotactic radiosurgery for patients with intractable typical trigeminal neuralgia who have failed medical management 
American Academy of Neurology and the European Federation of Neurological Societies - Practice parameter: The diagnostic evaluation and treatment of trigeminal neuralgia (an evidence-based review) 
Surgical treatment is indicated for patients whose TN is intractable despite medical therapy, in those who are intolerant to the adverse effects of the medications, and in those in whom previous procedures failed.
MVD is usually indicated for patients younger than 70 years who are at lower risk for complications during general anesthesia, although healthy older patients can tolerate it well. Percutaneous approaches (eg, radiofrequency ablation, glycerol injection, balloon compression, radiosurgery) are more frequently offered to elderly patients, those in poor medical condition, those with MS, and those in whom previous MVD has failed.
In elderly patients with limited life expectancy, radiofrequency rhizotomy (ie, PRTG) is sometimes preferred because it is easy to perform, has few complications, and provides symptomatic relief for a period.
Many operations have been offered to patients in recent decades. Local ablation of the peripheral nerve and wide sectioning of the sensory roots have largely been abandoned. In the past, alcohol injection was given to the affected nerve; rhizotomy or tractotomy was recommended if pharmacologic treatment was unsuccessful.
Alcohol or phenol injection of the trigeminus can be performed at various locations along the nerve, and the goal is to destroy selective pain fibers. Although it is an easy procedure, the success rate is low, in part because of a low selectivity of effect on the fiber type with this substance. Recurrence rates are around 50% at 1 year.
Three operative strategies now prevail: percutaneous procedures, gamma knife radiosurgery (GKRS), and microvascular decompression (MVD). Ninety percent of patients are pain-free immediately or soon after any of the operations, though relief lasts far longer with MVD. Pain-free intervals last on average 1.5–2 years after two of the percutaneous procedures (percutaneous retrogasserian glycerol rhizotomy [PRGR] and percutaneous balloon microcompression [PBM]), 3–4 years after GKRS, 4–5 years after another percutaneous procedure (radiofrequency trigeminal gangliolysis), and 10–12 years after MVD. 
Percutaneous procedures make sense for older patients with medically unresponsive TN. Younger patients and those expected to do well under general anesthesia should first consider microvascular decompression–presently the most cost-effective surgery.
More recently, however, both MRI imaging and posterior fossa exploration have frequently revealed a structural cause for neuralgia such as an ectatic artery impinging on the trigeminal nerve root. In such cases, simple decompression and separation of the compressing vessel from the nerve root produces long-lasting relief of symptoms.
The cost per quality adjusted pain-free year was $6,342, $8,174, and $8,269 for glycerol rhizotomy, microvascular decompression, and stereotactic radiosurgery, respectively, according to Pollock.  As of 2005, approximately 8000 patients with TN were undergoing surgery each year in the United States, at an estimated cost exceeding $100 million.
The primary complications of surgery include troublesome sensory loss over the face or the dysesthetic syndrome of anesthesia dolorosa. Anesthesia dolorosa can be disabling, occasionally is worse than the original TN, and is difficult to treat. For this reason, procedures with the best long-term success and the least risk of a residual facial dysesthetic syndrome are the most promising.
In 2008, Tatli et al reviewed surgical options, which mostly included MVD and radiofrequency rhizotomy.  Their review suggests that each surgical technique for treatment of TN has merits and limitations. They also found that MVD provides the highest rate of long-term patient satisfaction with the lowest rate of pain recurrence.
The trigeminal nerve is the largest of all the cranial nerves. It exits laterally at the midpons level and has 2 divisions—a smaller motor root (portion minor) and a larger sensory root (portion major). The motor root supplies the temporalis, pterygoid, tensor tympani, tensor palati, and mylohyoid muscles, as well as the anterior belly of the digastrics muscle. The motor root also contains sensory nerve fibers that particularly mediate pain sensation.
The gasserian ganglion is located in the trigeminal fossa (Meckel cave) of the petrous bone in the middle cranial fossa. It contains the first-order general somatic sensory fibers that carry pain, temperature, and touch. The peripheral processes of neurons in the ganglion form the 3 divisions of the trigeminal nerve: ophthalmic, maxillary, and mandibular (see the image below). The ophthalmic division exits the cranium via the superior orbital fissure; the maxillary and mandibular divisions exit via the foramen rotundum and foramen ovale, respectively. The ophthalmic and maxillary nerves are purely sensory. The mandibular nerve has sensory and motor functions.
The proprioceptive afferent fibers travel with the efferent and afferent roots. They are peripheral processes of unipolar neurons located centrally in the mesencephalic nucleus of the trigeminal nerve.
For more information about the relevant anatomy, see Trigeminal Nerve Anatomy.
When surgical procedures are contemplated, appropriate and routine preoperative laboratory tests are in order. In cases of typical (TN1) and atypical trigeminal neuralgia (TN)—referred to as TN1 and TN2, respectively—magnetic resonance imaging (MRI) of the brain with contrast is required to evaluate for vascular malformations or other lesions. MRI is sensitive for the exclusion of intracranial lesions, which can cause TN in rare cases. Devote particular attention to the posterior fossa. High-resolution imaging of the nerve at the brainstem entry zone may reveal vascular compression of the nerve. [7, 8]
Percutaneous procedures usually can be performed on an outpatient basis, with the patient under local or brief general anesthesia, at an acceptable or minimal risk of morbidity. For these reasons, they commonly are performed in debilitated persons or those older than 65 years.
Brief general anesthesia is indicated for percutaneous radiofrequency trigeminal gangliolysis (PRTG) and percutaneous balloon microcompression (PBM); no anesthesia is required for percutaneous retrogasserian glycerol rhizotomy (PRGR).
General anesthesia is indicated for microvascular decompression (MVD).
The success rate of surgical treatment if trigeminal neuralgia (TN) varies according to the experience of the surgeon or the anesthesiologist; therefore, surgical treatment of TN should be performed only by experienced surgeons.
Surgical therapy can be divided into procedures on the trigeminal nerve or gasserian ganglion (external or percutaneous, usually performed by pain management specialists) and procedures on the nerve root (open-skull microvascular decompression [MVD], performed by neurosurgeons, and gamma knife surgery [GKS], performed by radiation therapists). MVD has an overall better success rate, but it is more invasive. Percutaneous techniques may be more accessible to elderly patients who are at high surgical risk.
Three percutaneous procedures for TN have been described: percutaneous radiofrequency trigeminal gangliolysis (PRTG), percutaneous retrogasserian glycerol rhizotomy (PRGR), and percutaneous balloon microcompression (PBM). 
In each procedure, the surgeon introduces a trocar or needle lateral to the corner of the mouth and, under fluoroscopic guidance, into the ipsilateral foramen ovale. The ganglion is lysed at this location. Patients are left with minor, local, residual facial numbness after PRTG, may occasionally lose sensation after PRGR, and rarely do so after PBM.
Percutaneous radiofrequency trigeminal gangliolysis
PRTG (radiofrequency rhizotomy) is an outpatient procedure performed by placing a needle into the gasserian ganglion, through which an electrical current passes, heating the probe and producing a thermal lesion in the ganglion. Of the percutaneous procedures, it has the lowest reported rate of pain recurrence, with the average patient experiencing 4–5 years of good pain relief. 
PRTG, like PBM, is a relatively inexpensive and accessible technique and is less invasive than open surgery, with a lower (long-term) efficacy-to-recurrence ratio. The result is highly dependent on the surgeon’s skill.
Percutaneous balloon microcompression
With PBM, the operator inserts a balloon catheter through the foramen ovale into the region of the ganglion and inflates it for 1-10 minutes. PRTG has gained wide acceptance according to several investigators, because the patient recovers quickly, and goes home the day of the procedure or the next day.
According to Tan et al, the recurrence rate approaches 25% with PRTG, and occasionally patients suffer complications of jaw weakness and corneal anesthesia. As related by Meglio and Cioni, some surgeons report excellent results with PBM, which are comparable to those with PRTG. 
Percutaneous retrogasserian glycerol rhizotomy
Glycerol injection of the gasserian ganglion to destroy the pain-transmitting fibers has been used for a long time. It is easy to perform and has a higher efficacy rate and a lower recurrence rate than alcohol injection.
In PRGR, as in other percutaneous procedures, a spinal needle penetrates the face, this time to the trigeminal cistern, at which point a cisternogram is obtained with water-soluble contrast material. After removing this material, the surgeon instills anhydrous glycerol, asking the patient to remain seated for an additional 2 hours until the nerve is fully ablated.
In the view of some, PRGR is the favored procedure, in that it includes only a minimal risk of disturbed facial sensitivity postoperatively. However, Cappiabianca et al  and Taha and Tew,  who favor the radiofrequency rhizotomy, argue that PRGR has the highest rate of pain recurrence.
Gamma Knife Surgery
In 1953, Leksell irradiated the trigeminal nerve in 2 patients with good initial success, but the data were not published until 1971.  Progress with GKS began to accelerate when, in the 1990s, surgeons learned to target the nerve precisely with stereotactic magnetic resonance imaging (MRI), to determine the proper radiation dose for quickly relieving pain without incurring facial sensory loss, and to ascertain the length of nerve to be radiated.  Since 2000, GKS has become more widely available.
Stereotactic GKS is among the newest techniques for treating TN. It is less technically demanding and less operator-dependent than the percutaneous procedures, and it is the least invasive of all the surgical procedures. It appears to be about as effective as the percutaneous procedures, even in patients in whom prior surgery or medication trials failed, but it often takes weeks to months to bring relief and costs considerably more.
GKS consists of multiple rays (>200) of high-energy photons concentrated with high accuracy to deliver a 70-90 Gy dose to the target (ie, the trigeminal nerve root). [16, 17] This treatment destroys specific components of the nerve.
GKS can be used after a patient does not respond to any of the abovementioned procedures, including GKS itself. The device contains a stable source of radiation (60-Co) that frees this technique from requiring an external source of radioactivity (eg, cyclotron).
Of those treated with GKS, 60% of patients are immediately free of pain, and more than 75% of patients have greater than 50% relief after 1.5 years. Recurrence rates are around 25% between 1 and 3 years, and 50% at 3–4 years.
In a study of 106 subjects, Kondziolka et al found that most patients already had no relief with either microvascular decompression or glycerol rhizotomy.  At a median follow-up point of 18 months, 60% of patients were pain free, 17% were moderately improved, and 23% were minimally or not improved. They concluded that this technique is minimally invasive, is associated with a low risk (10%) of facial paresthesias or sensory loss, and offers a high rate (86%) of significant, initial pain relief.
Henson compared GKS with PRGR in 188 patients and concluded that GKS improved pain more consistently and induced less residual facial paresthesia. Pollack reviewed a group of 121 individuals who underwent one or the other procedure and found the rates of complete pain relief similar, about 60% at 6 months and 54% at 24 months.
In a study of 151 patients, Sheehan et al reported that 47% were pain-free after 1 year and 34% after 3 years  ; 9% suffered incurred new facial numbness after the procedure. In a similar study of 49 patients followed for a mean of 49 months, one third of whom had also failed either MVD or PRGR, GKS provided complete relief, even off medications, in 14 patients (32%) and partial but durable relief in 27 patients (61%) (McNatt). Others report rather disparate complete pain relief, from 42% (Jawahar) to 59 % (Drzymala) to 60% (Pollack) to 80% (Urgosik).
In these prospective but open and uncontrolled trials, complete pain relief predictably waned substantially by year 3-5, as with the percutaneous procedures. Pain relief also arrives much more slowly, often coming only 3-12 weeks after the procedure, too long a wait perhaps for some individuals. New facial numbness or paresthesias develop slowly over the first 12-15 months after GSK, reaching bothersome levels in 9-20% of patients.
MVD is the classic surgical procedure for TN, and the most effective one. Its rationale is based on the hypothesis that compression from vessels in the vicinity of the trigeminal nerve leads to abnormal nerve activity. Usually, it requires the demonstration of true contact and compression by the artery on the nerve (neurovascular compression [NVC]) , but some series report almost equally good results without any demonstrated abnormality on imaging or even frank compression shown preoperatively.
NVC can now be imaged preoperatively using MRI and MRA scans, fused into one 3-dimensional image volume (see video below). The fidelity of this type of imaging and intraoperative findings of NVC is high (96% sensitive). 
This video shows the 3-dimensional MRA/MRI on a TN patient whose intraoperative video shows identical NVC to that predicted by the preoperative imaging. This video also shows the microvascular decompression of the nerve and placement of Teflon "cotton" between the nerve and artery.
MVD is commonly performed in younger, healthier patients, especially those with pain isolated to the ophthalmic division or in all 3 divisions of the trigeminal nerve and in those with secondary TN. It is now the most common surgical procedure performed for TN.
In MVD, the skin is incised behind the ear and a 3-cm craniotomy performed. The dura is retracted to expose the trigeminal nerve, and the vascular elements compressing the nerve as it enters the pons are identified. Teflon felt is then used to pad the nerve away from the offending artery or vein. Patients spend 3–4 days in the hospital and another week convalescing at home. Thus, recovery is more prolonged than with percutaneous procedures.
Large series have been published, and the initial efficacy is greater than 80%. The recurrence rates after MVD, compared with those after other invasive treatments, are among the lowest (20% at 1 y, 25% at 5 y). [20, 6]
Complications include chemical meningitis, ipsilateral hearing loss, and facial sensory loss or palsy. Mortality rates in experienced centers are less than 0.5%. Mortality for MVD approaches 0.5%. Serious morbidity includes dizziness, temporary facial palsy, cerebrospinal fluid leaks, meningitis, cerebellar stroke, and hearing loss, which may occur in 1-5% of cases.
In patients undergoing MVD in whom no NVC can be found either by imaging, or at the time of surgery, a nerve disruptive procedure, trigeminal internal neurolysis (TIN) can be performed in which the surgeon thoroughly separates the internal fibers of the nerve as it enters the brainstem. This procedure produces pain relief of equivalent quality and duration to MVD, and in most cases results in nominal facial sensory loss. 
Peripheral neurectomy, although safe and effective, is rarely used; however, it may be of value in patients who have TN and a limited life span. 
Trigeminal tractotomy (cutting of the descending trigeminal tract in the medulla), retrogasserian glycerol instillation in the trigeminal cistern, peripheral alcohol blocks, and partial trigeminal rhizotomy via a posterior fossa craniotomy are other rarely used options.
Accurate data on surgical outcome are still difficult to obtain. Most surgical series do not meet modern criteria for evidence-based medicine. Recently, an evidence-based approach has been applied to both the diagnosis of trigeminal neuralgia (TN) and its surgical management. These results were published in Neurology in August 2008 and can be viewed at Medscape.
The challenges of the field are exemplified by a paper published by Zakrzewska and Lopez, who assessed the quality of 222 reports of surgical management of TN and found only 3 randomized controlled trials (RCTs) on peripheral techniques.  The vast majority of the evidence derived from case series reports (class 3 evidence), and a very high proportion was of poor quality.
The difficulties center on the following important issues:
Lack of clear diagnostic criteria and baseline assessments
Poor methodology - Low numbers, short follow-up period, high percentage lost to follow-up, mixture of cases (eg, previous surgery, including repeated treatments)
Lack of Kaplan-Meier assessment of pain relief, poorly defined outcome measures (eg, partial success), incomplete reporting of all complications, and no quality-of-life evaluations
Lack of independent evaluation
Certain principles seem to transcend even the relatively poor outcomes studies performed thus far Seemingly, the longer the duration of symptoms, the poorer the chances of success. Of all the procedures, microvascular decompression (MVD) carries the lowest rate of facial dysesthesia (0.3%). Compared with the percutaneous procedures, MVD rarely causes facial numbness (rate, 0.15%). In addition, it is the preferred procedure in younger patients who desire no sensory deficit. It is also the treatment most likely to yield sustained postoperative pain relief.
One study found that 70% of patients had excellent results (defined as a cure or significant pain relief) 10 years after the procedure.  Possible reasons for failure include new vascular compression from scarred implants or other sources, but these are rarely identified during posterior fossa reexploration for failed MVD. After an initial 10% risk of recurrence of TN within 1 year after MVD, the risk of pain recurrence is about 3.5% every succeeding year.  The reasons for this recurrence are not clear.
In a 1999 study, cerebellar injuries and hearing loss occurred in fewer than 1% of the patients, and cerebrospinal fluid (CSF) leakage occurred in 1.85%.  As expected, these rates were inversely proportional to the total number of procedures performed.
Burchiel reported that 90% of patients are pain free after any of the operations mentioned.  Those in whom the first percutaneous procedure fails may undergo a repeat procedure, which usually provides relief. Sweet reported that pain-free intervals last 1.5-2 years after percutaneous retrogasserian glycerol rhizotomy (PRGR) and percutaneous balloon microcompression (PBM) and last 3-4 years after percutaneous radiofrequency trigeminal gangliolysis (PRTG). For MVD, 15 years of relief is typical.
Morbidity associated with trigeminal nerve decompression stems from hemorrhage, infection, and possible damage to the brainstem around the area of decompression. Adverse effects of surgery include corneal anesthesia, facial numbness outside of the trigger zone, new facial pain, facial dysesthesias, and intracranial hemorrhage (rare). Anesthesia dolorosa (TN pain associated with dense hypesthesia) is usually a result of surgical treatment; this is difficult to treat.
In centers where MVD is frequently performed, complications include facial dysesthesia (0.3%), facial numbness (0.15%), cerebellar injuries and hearing loss (< 1%), and CSF leakage (< 2%).
Complications of PRTG depend on the amount of numbness created by the lesion. Dysesthesia has been reported in up to 5-25% of patients, corneal numbness in up to 15%, and masseter weakness in about 4%. These complications are markedly reduced if the numbness produced by the procedure is limited. Some of these complications may be reversible. To avoid ophthalmic complications, some experts do not recommend this approach when the ophthalmic division is involved. 
The rate of pain recurrence with PRGR is between that of radiofrequency ablation and that of percutaneous compression. The rate of significant facial numbness (5%) is low. About half of the patients have pain recurrence at 2 years.  PBM carries about the same complications and average pain-free outcome as PRGR does (2 y). [30, 31]