Ketamine for Refractory Chronic Migraine: an Observational Pilot Study and Metabolite Analysis

Patients with refractory chronic migraine commonly experience near-constant pain, marked disability and failure of multiple preventive and acute treatments. Previous retrospective reports suggest that 5-day continuous infusions of racemic (R,S)-ketamine can produce short-term analgesic benefit in such patients, with a subset maintaining response for weeks to months. Ketamine is extensively metabolised by cytochrome P450 enzymes into several metabolites, notably hydroxynorketamine (HNK). Preclinical work in rodents has shown that (2R,6R)-HNK can have antidepressant and analgesic properties and that HNK may reach high plasma concentrations during prolonged ketamine infusions, but the enantiomeric distribution and its relation to clinical response in refractory migraine have not been characterised prospectively. Schwenk and colleagues therefore conducted a prospective, open-label observational pilot study to compare the short-term analgesic effect of a 5-day continuous (R,S)-ketamine infusion with historical data from a prior inpatient lidocaine infusion in the same patients. A secondary objective was to measure plasma concentrations of ketamine and its metabolites, including the enantiomers (2R,6R)-HNK and (2S,6S)-HNK, at predefined time points and to explore associations between metabolite levels and pain outcomes.

This single-centre, prospective, observational pilot enrolled adults (≥18 years) with chronic migraine meeting predefined refractory criteria (class IV). The study was approved by the institutional review board and registered on ClinicalTrials.gov. Recruitment identified patients who had previously completed a 5-day lidocaine admission and were scheduled for a subsequent 5-day admission for ketamine; historical data from the lidocaine admission were extracted retrospectively from the electronic medical record. Exclusion criteria included contraindications to ketamine (for example schizophrenia, active psychosis, pregnancy, poorly controlled cardiovascular disease, cirrhosis) and prior intravenous ketamine treatment. The study was open-label and neither patients nor assessors were blinded. During the lidocaine admission (retrospective comparator), patients received a 5-day infusion starting at 1 mg/min and titrated according to pain, adverse effects and plasma lidocaine levels, with a target not to exceed 5 mcg/mL or 4 mg/min. Preventive migraine medications were continued during hospitalisations. For the prospective ketamine admission, (R,S)-ketamine (Ketalar) was started at 10 mg/h and increased every 4–6 hours by 5–10 mg/h up to intolerable adverse effects or 1 mg/kg/h. Pain was assessed using a 0–10 numerical rating scale (0=no pain, 10=worst pain imaginable) at baseline, throughout hospitalisation and at the first post-discharge visit; follow-up phone calls occurred 8–18 days after discharge in most patients. Blood for metabolite analysis was collected at baseline, 24 h, 72 h and ≈120 h after starting the ketamine infusion. Samples were processed and stored at −70°C prior to shipping to an external laboratory. Plasma concentrations of ketamine and metabolites—(R,S)-ketamine, (R)- and (S)-norketamine, (R,S)-dehydronorketamine (DHNK), and (2R,6R)-HNK and (2S,6S)-HNK—were measured using high-performance liquid chromatography coupled to tandem mass spectrometry with enantioselective separation where indicated. Statistical planning assumed a conservatively estimated 50% pain reduction with ketamine versus baseline; with paired comparisons and the chosen parameters, a sample of 6 pairs was calculated to provide approximately 87.5% power. Comparisons used Kruskal–Wallis tests for between-hospitalisation pain ratings, one- and two-way repeated-measures ANOVA (with Bonferroni correction) for within- and between-time analyses, and Pearson correlation coefficients for relationships between mean daily pain and metabolite concentrations. Continuous data are reported as mean ± SD.

Seven patients consented and underwent study procedures; one was excluded after enrolment because their diagnosis was not chronic migraine, leaving six patients for analysis. Ages ranged from 20 to 55 years, three were female, and two had comorbid depression. Five of the six patients had received dihydroergotamine during the initial lidocaine admission. All six patients had Migraine Disability Assessment (MIDAS) scores consistent with grade 4 severe disability; five had scores ≥140 indicating near-constant pain. No patients were taking opioids at the time of the ketamine admission. Pain outcomes during the lidocaine admission showed a mean baseline pain of 7.5 ± 2.2 (0–10 scale) that decreased to 4.7 ± 2.8 by the end of the hospitalisation (p < 0.05). Pain at the first post-discharge office visit (mean 28 ± 8 days after treatment) did not differ significantly from baseline (7.5 ± 2.2 vs 7.0 ± 1.4, p > 0.05). During the ketamine admission the mean baseline pain was 7.4 ± 1.4 and decreased to 3.7 ± 2.3 by the end of the 5-day infusion (p < 0.05); pain at the first post-discharge visit (mean 41 ± 7 days) again returned to near baseline (7.4 ± 1.4 vs 7.2 ± 1.7, p > 0.05). Comparing the two treatments directly, the mean change in pain was greater after ketamine than lidocaine: −3.7 for ketamine versus −2.8 for lidocaine (p < 0.05). The temporal profile differed between treatments: pain relief with ketamine began around day 3 and remained greater through the end of hospitalisation, whereas lidocaine-associated pain reduction did not appear until the end of the 5-day admission. Ketamine dosing began at 10 mg/h for all patients; the mean infusion rate by the end of day 2 was 46.7 ± 9.8 mg/h and the mean maximum infusion rate achieved was 72.5 ± 10.4 mg/h (occurring on day 5). For lidocaine, mean infusion rates reported were 1.6 ± 0.8 mg/min on day 1 and 2.5 ± 0.7 mg/min on day 4. Adverse effects were common but not serious. Four of six patients experienced transient adverse events during lidocaine treatment (including bradycardia/junctional rhythm, hallucinations, blurred vision, nausea and insomnia). All six patients experienced adverse events during ketamine infusion—hallucinations, nightmares or vivid dreams, blurred vision and nausea/vomiting were reported—but none required permanent discontinuation of ketamine and no serious adverse events occurred. Metabolite analyses were complete for all six patients. Plasma (R,S)-ketamine concentrations increased throughout the 5-day infusion and were highest at the end of infusion. (R,S)-norketamine and (R,S)-DHNK concentrations stayed relatively stable from 24 h until the end of infusion, while total HNK concentrations peaked at 72 h. When enantiomers were resolved, mean concentrations of (R)-norketamine exceeded (S)-norketamine at 24 h, 72 h and end of infusion (p < 0.05). Similarly, (2R,6R)-HNK concentrations exceeded (2S,6S)-HNK at 72 h and at the end of infusion (p < 0.05). The ratio of (2R,6R)-HNK to (2S,6S)-HNK increased from 24 h to the end of infusion (p < 0.05), whereas the ratio of (R)-norketamine to (S)-norketamine did not change significantly. No statistically significant correlations were observed between mean daily pain ratings and plasma concentrations of (R,S)-ketamine, (R,S)-norketamine or (2R,6R;2S,6S)-HNK; nonetheless the authors report strong but non-significant negative Pearson correlations (R = −0.822 for (R,S)-ketamine vs pain and R = −0.887 for total HNK vs pain, p > 0.05), recognising the limited sample size.

Schwenk and colleagues interpret their findings as indicating that both 5-day lidocaine and (R,S)-ketamine infusions can produce short-term pain reduction in patients with refractory chronic migraine, with ketamine producing a statistically greater reduction in pain scores than lidocaine. They note, however, that the clinical significance of the between-treatment difference is uncertain because a minimum clinically important difference on the 0–10 numeric rating scale has been reported as approximately 2 points. Temporal patterns differed: ketamine-associated relief began around day 3 and coincided with the period when plasma HNK concentrations peaked, whereas lidocaine-related improvement was more apparent only at the end of its 5-day course. By the first post-discharge visit (several weeks later) pain had generally returned to baseline, underscoring the difficulty of achieving sustained benefit in this highly refractory population. The metabolite data showed that (2R,6R)-HNK was the predominant circulating compound at 72 h and at the end of infusion in most patients, and that its concentration exceeded both parent ketamine and norketamine at those time points. The study therefore provides the first prospective enantioselective characterisation of ketamine metabolites during a 5-day infusion in refractory migraine, showing stereoselective accumulation favouring (2R,6R)-HNK. The investigators discuss possible biological mechanisms: prior work implicates CYP2A6 in stereoselective conversion of norketamine to HNK and cell-based studies show that ketamine and HNK enantiomers can induce CYP2A6 and CYP2B6 expression. They also cite data suggesting that these compounds can interact with oestrogen receptor α (ERα) and modulate CYP expression, raising the possibility that prolonged infusions could induce metabolic pathways not seen after single infusions. Important limitations acknowledged by the study team include the small sample size, the retrospective and sometimes inconsistent timing of pain assessments from the lidocaine hospitalisations, potential recall bias in some follow-up calls, the open-label design, and the inability to control for concomitant inpatient medications (for example dihydroergotamine, ketorolac). Urinary excretion was not measured, limiting interpretation of clearance pathways, and the findings may not generalise to settings using lower infusion doses or to less refractory patient populations. The authors recommend larger, controlled studies that incorporate pre- and post-treatment metabolic phenotyping (for example CYP2A6 and CYP2B6 activity) and urine analysis to clarify whether stereoselective metabolism or renal clearance accounts for the observed HNK accumulation and whether (2R,6R)-HNK contributes to analgesic effects.

In this small observational pilot, both 5-day lidocaine and (R,S)-ketamine infusions were associated with short-term reductions in pain among patients with refractory chronic migraine, with a larger mean reduction after ketamine. Circulating (2R,6R)-HNK concentrations peaked between Days 3 and 5—coinciding with the interval of greatest pain relief—and the (2R,6R) enantiomer accumulated relative to (2S,6S)-HNK over the infusion period. No statistically significant correlations were found between pain scores and measured plasma concentrations, but the timing and stereoselective profile of HNK suggest hypotheses worthy of further investigation. The study team concludes that larger, controlled studies with metabolic phenotyping and urine collection are needed to determine whether (2R,6R)-HNK or related mechanisms contribute to ketamine's short-term analgesic effects.