Four Fibromyalgia Drugs Approved in US, None in EU. Why?

Your fibromyalgia patient has a diagnosis, a playbook, and a prescription. Only some of that stands on firm ground. The FDA has approved four drugs for fibromyalgia, European regulators refused the first three, and no European application has been identified for the fourth. The newest pooled data say a drug gives substantial extra relief to about 1 patient in 10 (Rheumatology).

The Playbook Is Settled. The Evidence Isn’t.

A 2026 clinical practice review published in The New England Journal of Medicine sets out the standard model. Treat fibromyalgia as a nociplastic pain disorder, diagnose it clinically, start with education and exercise, and add a short list of centrally acting drugs. The review calls the drug effect sizes small (NEJM).

A 2025 Cochrane overview puts numbers on “small.” For at least 50% pain relief versus placebo:

• Duloxetine 60 mg: 36% vs 23% (number needed to treat [NNT], 7.6).
• Pregabalin 450 mg: 24% vs 14% (NNT, 9.7).
• Milnacipran 100 mg: 27% vs 18% (NNT, 10) (Rheumatology).

Withdrawals for adverse events ran 6 to 9 percentage points above placebo. Every trial in the overview excluded inflammatory rheumatic disease, so these are not your clinic’s patients. The authors suggest a 4- to 6-week trial and stopping if relief is not substantial by 6 weeks (Rheumatology).

Trial size matters. In an older network meta-analysis restricted to trials with at least 100 patients per arm, only serotonin-norepinephrine reuptake inhibitors (SNRIs), pregabalin, and cognitive-behavioral therapy (CBT) held up. The authors judged the drug benefits to be of questionable clinical relevance (Ann Rheum Dis).

Same Trials, Opposite Verdicts

The FDA approved pregabalin (2007), duloxetine (2008), and milnacipran (2009). The European Medicines Agency (EMA) committee refused all three, so no fibromyalgia drug holds an EU-wide indication (Arch Womens Ment Health). The committee concluded the benefits did not outweigh the risks, finding the drugs’ effects to be inconsistent or insufficiently shown (EMA, pregabalin; EMA, duloxetine; EMA, milnacipran).

Guidelines split the same way:

• European Alliance of Associations for Rheumatology (EULAR) 2017: Exercise is the only “strong for” recommendation. Pregabalin, duloxetine or milnacipran, low-dose amitriptyline, tramadol, and cyclobenzaprine are “weak for” (Ann Rheum Dis).
• National Institute for Health and Care Excellence (NICE) 2021: Consider an antidepressant. Do not start gabapentinoids or opioids (NICE NG193). 

Critics say NICE built its guidance on a category, chronic primary pain, with no trials of its own. Defenders point to its fibromyalgia subgroup analyses (NICE evidence review). Neither side can point to a head-to-head trial in patients with fibromyalgia that settles it (Medscape News UK). 

The Newest Approval Has the Messiest Trial Record

Sublingual cyclobenzaprine (Tonmya) won FDA approval on August 15, 2025, after two positive phase 3 trials (Clin Drug Investig). No European application has been identified, so the EMA has not weighed in. The disagreement here is inside the trial record. 

• For: RESILIENT (n=457) reduced pain by 1.8 points vs 1.2 with placebo (effect size, 0.38) (Pain Med). RELIEF (n=503) showed 1.91 vs 1.51 (P=.01) (Arthritis Care Res).
• Against: A third 5.6 mg trial, RALLY (n=514), was halted for futility. Two 2.8 mg trials missed their primary endpoints (Clin Drug Investig). A four-trial meta-analysis did not include RALLY (Clin Rheumatol).

The differences of 0.4 to 0.7 points sit well below the 2-point threshold used in a large earlier review (JAMA Intern Med). The trials ran 14 weeks, enrolled 95% women, and were sponsor-run, with Tonix employees among the authors (Pain Med). 

Amitriptyline, Naltrexone, Cannabidiol: Real or Hype?

Low-dose amitriptyline is the off-label favorite. Its case is stronger on sleep, fatigue, and quality of life than on pain (JAMA Netw Open). But the Cochrane NNT of 4.1 comes from four trials and 275 participants, rated very low quality (Cochrane). A 2026 cost-effectiveness model favors it, but draws its inputs from a network meta-analysis by the same group (JAMA Netw Open).

The two most-discussed newer options have not beaten placebo in their larger recent trials. In a 2026 trial of 98 women, low-dose naltrexone showed a 0.49-point pain difference favoring placebo at 3 months (Eur J Pain). In 200 patients, cannabidiol reduced pain by 0.4 points versus 1.1 with placebo (Ann Rheum Dis). The NEJM review says low-dose naltrexone “may be helpful” (NEJM). That is a stretch.

Brain Disease or Autoimmune Disease?

Where does fibromyalgia pain come from? The standard model in the 2026 NEJM review is that it is nociplastic pain, driven by central sensitization, an amplified response to painful and nonpainful input in the central nervous system, with little or no input from peripheral tissue (NEJM). Some researchers argue the periphery or the immune system plays a bigger role. Call them the central camp and the peripheral and immune camp.

The central camp received new support in 2026. A genome-wide meta-analysis of 2,563,755 people identified 26 risk loci, with heritability enriched in brain tissues and neural cell types (Nature Medicine).

The peripheral and immune camp has data too. Immunoglobulin (Ig)G from patients, transferred to mice, produced hypersensitivity and loss of skin nerve fibers (J Clin Invest). A 2026 meta-analysis found lower distal-leg nerve fiber density across non-neuropathic chronic pain conditions (16 studies, 1123 participants), with unclear clinical significance (Pain).

Both sides are working with small samples and no validated test. The NEJM review calls claims of autoimmune disease or small-fiber neuropathy unsubstantiated (NEJM). Note that several central-model authors, including Daniel Clauw, co-wrote both the NEJM review and the genetics study. A conditional from the task force that defined nociplastic pain is useful: if the antibody findings replicate and antigens are found, antibodies could mark a treatable subgroup (Pain).

Is Fibromyalgia a Diagnosis of Exclusion?

Two camps disagree, and the data show that a clinician’s diagnosis and meeting the formal criteria often identify different patients. 

Positive diagnosis. The 2016 criteria are valid irrespective of other diagnoses (Semin Arthritis Rheum). The NEJM review says childhood-onset, slowly evolving widespread pain needs very little workup (NEJM).

Evaluate first. The same 2016 paper says the criteria should never be applied without gathering essential medical information (Semin Arthritis Rheum). Clinician diagnoses and criteria diverge in both directions:

1. In the UK Biobank, 72.6% of 4197 people meeting the 2016 criteria had no clinician diagnosis. Women were far more likely than men to have one (32.0% vs 13.9%) (Arthritis Care Res).
2. In a US survey, 73.5% of adults reporting a clinician’s diagnosis did not meet surrogate criteria (PLoS One).
3. In a rheumatology clinic, clinicians had missed 49.6% of criteria-positive patients and diagnosed 11.4% of those who did not meet criteria (Arthritis Care Res).

Nothing here can be checked against a gold standard, because none exists (Semin Arthritis Rheum). For rheumatologists, the practical rule is restraint with serology. Antinuclear antibodies (ANA) are positive in 3% to 15% of healthy people and 8% to 11% of patients with fibromyalgia. Skip subserologies without suggestive signs (JAMA Intern Med). Keep lupus and axial spondyloarthritis enthesitis on the differential (NEJM).

The Opioid Warning Rests on Other Pain Conditions

The NEJM review says opioids should be avoided and that all-cause mortality “nearly doubles” (NEJM). That figure traces to a Tennessee Medicaid cohort of chronic noncancer pain, where long-acting opioids carried a mortality hazard ratio of 1.64 (JAMA). Back pain made up 75% of that cohort, and fibromyalgia was not a reported category. The excess was concentrated early, fading to a hazard ratio of 1.03 after 180 days (JAMA).

No randomized trial supports a conventional opioid in fibromyalgia (Mayo Clin Proc). The CDC’s 2022 guideline names fibromyalgia as a setting where opioid benefits are unlikely to outweigh risks (MMWR). So the advice stands. The evidence behind it lacks any data on its benefit and centers on general chronic-pain harm data, not fibromyalgia-specific harm data.

Exercise Is the Only Strong Recommendation, and Its Effect Shrinks Under Scrutiny

In a 167-trial meta-analysis, exercise lowered pain by a standardized mean difference of −0.84, against usual care, waitlist, placebo, or sham (Pain). In a separate, older analysis restricted to trials with at least 50 patients per group, aerobic exercise fell to −0.37 and CBT to −0.32 (Ann Rheum Dis). Participants cannot be blinded, and large trials with active comparators are few.

A 2026 cluster-randomized trial found that adding transcutaneous electrical nerve stimulation to physical therapy lowered movement-evoked pain by 1.2 points. The comparator was physical therapy alone, with no sham (JAMA Netw Open). A self-guided digital acceptance and commitment therapy program improved global impression in 71% vs 22%, but pain fell only 0.7 points more, research staff were unmasked, and the sponsor’s employees were among the authors (Lancet).

What to Do Monday

1. Diagnose from a published criteria set. Then order the small panel the patient’s onset and exam justify, not everything (NEJM).
2. Hold the serology. Skip ANA and subserologies without features of a rheumatologic disease (JAMA Intern Med).
3. Start with exercise. Begin at low intensity and build slowly. Reinforce the program before effects fade (NEJM).
4. Set a stop date for drugs. Try duloxetine, milnacipran, or pregabalin for 4-6 weeks and stop if relief is not substantial (Rheumatology).
5. Don’t start opioids. Watch gabapentinoid and opioid combinations for respiratory depression (FDA).

A Note on the Evidence

The pivotal trials for the approved drugs were manufacturer-sponsored, and the cyclobenzaprine trials were run by the sponsor (Pain Med). The Cochrane populations excluded inflammatory rheumatic disease (Rheumatology). And the 2026 NEJM review, the genetics study, and several central-model papers share authors.

The Bottom Line

• Drugs help about 1 patient in 10 by a small average margin. Try one for 4-6 weeks and stop if it fails (Rheumatology).
• Regulators and guidelines disagree. The FDA approved four drugs, the EMA none, and EULAR and NICE split on pregabalin (Arch Womens Ment Health).
• The newest approval, sublingual cyclobenzaprine, rests on two positive trials and several that failed (Clin Drug Investig).
• Most evidence points to central nervous system amplification of pain, and new genetics support that. Peripheral and immune contributions remain unsettled (Nature Medicine).
• The diagnosis is clinical and the formal criteria often identify different patients. Among people who meet the criteria, women are more likely than men to have a diagnosis (Arthritis Care Res).
• Exercise stays first line, with a smaller effect than headline numbers suggest. Opioids stay out (Ann Rheum Dis).