Trigeminal neuralgia

Trigeminal neuralgia is a facial pain in the territory of the trigeminal nerve, the fifth cranial nerve, which carries sensation from the forehead, cheek and jaw. The 2019 guideline of the European Academy of Neurology (EAN) classifies it as primary, either classical or idiopathic depending on how far a blood vessel presses on the nerve, or as secondary, when another disease such as a tumour causes it. Almost every acupuncture trial for the condition comes from China and compares acupuncture, alone or added to carbamazepine, with carbamazepine alone, open-label: the patients know which treatment they had. The reviews below pool those trials, and each states that this design limits what the pooled figures can show.

No Cochrane review

Cochrane has not reviewed acupuncture for trigeminal neuralgia. Its reviews of the condition cover non-antiepileptic drugs, last updated in 2013, and neurosurgical interventions, published in 2011.

The systematic reviews

A 2010 review by Liu and colleagues at Hong Kong Baptist University found 12 randomised trials, with 506 people having acupuncture and 414 having carbamazepine. All were of low quality, so the authors did not pool them; four trials reported acupuncture ahead of carbamazepine and eight found no difference. Later reviews pooled the trials and rated the certainty of each result with GRADE, the scale Cochrane and others use to say how much confidence a result supports, from high to very low.

Ang and colleagues, at the Korea Institute of Oriental Medicine, found pain scores lower with acupuncture than with carbamazepine across 15 trials, by a mean of 1.40 points on the visual analogue scale (VAS), a line on which the person marks how strong the pain is. The 95% prediction interval, the range in which a new trial’s result would be expected to fall, ran from 3.14 points in acupuncture’s favour to 0.34 against it. Across 13 trials, people having acupuncture reported fewer adverse effects than people taking carbamazepine. Wei and colleagues pooled 16 trials and also found more pain reduction with acupuncture, and also fewer adverse events, on evidence they rated extremely low.

Xu and colleagues, in Hangzhou, found that 37 of their 38 trials were published in Chinese, which they say raises a risk of publication and language bias. Only four trials were at low risk of bias for randomisation. They add that many trials used unvalidated or study-specific criteria and defined “response rate” differently. Their conclusion is that acupuncture added to medication may improve results over medication alone, on low-certainty evidence. Three of the six authors also ran the sham-controlled trial described below.

Wang and colleagues, at Guang’anmen Hospital in Beijing, rated one of their 23 trials at low risk of bias, five with some concerns and 17 at high risk. At the end of treatment the pooled VAS score favoured acupuncture by 1.49 points, but the result was statistically significant only in the 16 high-risk trials, not in the three trials with some concerns that reported it. Fewer than 15% of the VAS outcomes stated what period the pain score covered. Three trials measured pain three months after treatment ended and found a difference of similar size, which did not survive removing any single trial. The authors write that, with one sham-controlled trial, the evidence cannot separate a specific effect of needling from placebo response, expectation or the course a condition that relapses and remits takes on its own, and that it does not support routine clinical use.

A 2024 overview by He and colleagues assessed 13 of these reviews with AMSTAR-2, a checklist for the conduct of systematic reviews, and rated six critically low and seven low. Of the 13 outcomes it graded, none was high certainty, one moderate, four low and eight critically low. A 2022 network meta-analysis, which ranks several methods against each other through their shared comparisons, pooled 58 trials with 4,126 people and rated most of its evidence critically low.

The sham-controlled trial

A 2024 trial by Li and colleagues at two hospitals in Zhejiang randomised 120 people into four groups of 30, a 2×2 factorial design: electroacupuncture or sham electroacupuncture, each with carbamazepine at 300 mg a day or a placebo tablet, for four weeks. Electroacupuncture passes a small current between needles; Electroacupuncture describes the method. Sessions lasted 60 minutes, three times a week. The main points were ST02 (Sibai), ST07 (Xiaguan) and ST04 (Dicang), with GB01 (Tongziliao), SI18 (Quanliao) or ST06 (Jiache) added for the branch of the nerve involved, and LI04 (Hegu) and TB05 (Waiguan) on both hands and forearms. The sham needled non-points 1 cm to the side, with electrodes attached and no current. Sham acupuncture sets out the kinds of sham in use and why none is inert.

From the start to the end of treatment, the mean VAS score fell by 4.6 points with electroacupuncture plus carbamazepine, 3.7 with electroacupuncture plus placebo, 3.0 with sham plus carbamazepine and 0.9 with sham plus placebo. The trial reports electroacupuncture ahead of its sham by 1.6 points at week 4 and 0.8 at week 28, 24 weeks after treatment ended. Adverse events attributed to electroacupuncture occurred in 6 of 59 people and those attributed to carbamazepine in 15 of 59. The authors list as limitations a fixed protocol rather than individual treatment, diagnostic criteria older than the 2018 classification, and an open question whether electroacupuncture with low-dose carbamazepine compares with carbamazepine at 600 mg a day or more. This is the one trial Wang and colleagues rated at low risk of bias.

The guidelines

The EAN guideline, published in 2019, recommends carbamazepine or oxcarbazepine as drugs of first choice for long-term treatment, the carbamazepine recommendation strong on moderate-quality evidence. It recommends surgery when drugs do not control the pain or are poorly tolerated, with microvascular decompression first among the operations for classical trigeminal neuralgia, and psychological and nursing support alongside. The full text does not mention acupuncture. The American Academy of Neurology lists its 2008 guideline, written with the European Federation of Neurological Societies, as reaffirmed on 15 April 2024. That practice parameter says carbamazepine should be offered for pain control (its Level A) and oxcarbazepine too (Level B), and it covers drugs and surgery only.

In England, the NICE Clinical Knowledge Summary for primary care, last revised in January 2024, advises offering carbamazepine where there are no red-flag signs, starting at 100 mg up to twice daily and raising the dose every 2 weeks. It notes that 200 mg three or four times a day is enough for most people, and advises specialist advice if carbamazepine is unsuitable. Its management advice does not mention acupuncture. Clinical guidelines compares how guideline bodies treat acupuncture across conditions.

What the reviews could not settle

The pooled trials compare acupuncture with an open-label drug, so they cannot say how much of the difference comes from needling and how much from expectation; Wang and colleagues found one sham-controlled trial among 23. Of the same 23 trials, three reported pain three months after treatment ended. Seventeen did not report adverse events. The 2024 trial gave half or less of the daily carbamazepine dose that NICE describes as sufficient for most people, which leaves open how acupuncture compares with that dose.

AcuiQ’s trigeminal neuralgia page and the related facial pain page list the protocols individual studies prescribed, each with its citation. Temporomandibular disorders covers jaw and facial pain of another kind, and Bell’s palsy covers a disorder of the facial nerve, and Herpes zoster and postherpetic neuralgia covers the nerve pain that follows shingles.