Intravenous Minocycline in Acute Stroke
A Randomized, Controlled Pilot Study and Meta-analysis
Background and Purpose—Minocycline, in animal models and 2 small randomized controlled human trials, is a promising neuroprotective agent in acute stroke. We analyzed the efficacy and safety of intravenous minocycline in acute ischemic and hemorrhagic stroke.
Methods—A multicenter prospective randomized open-label blinded end point evaluation pilot study of minocycline 100 mg administered intravenously, commenced within 24 hours of onset of stroke, and continued 12 hourly for a total of 5 doses, versus no minocycline. All participants received routine stroke care. Primary end point was survival free of handicap (modified Rankin Scale, ≤2) at day 90.
Results—Ninety-five participants were randomized; 47 to minocycline and 48 to no minocycline. In the intention-to-treat population, 29 of 47 (65.9%) allocated minocycline survived free of handicap compared with 33 of 48 (70.2%) allocated no minocycline (rate ratio, 0.94; 95% confidence interval, 0.71–1.25 and odds ratio, 0.73; 95% CI, 0.31–1.71). A meta-analysis of the 3 human trials suggests minocycline may increase the odds of handicap-free survival by 3-fold (odds ratio, 2.99; 95% CI, 1.74–5.16) but there was substantial heterogeneity among the trials.
Conclusions—In this pilot study of a small sample of acute stroke patients, intravenous minocycline was safe but not efficacious. The study was not powered to identify reliably or exclude a modest but clinically important treatment effect of minocycline. Larger trials would improve the precision of the estimates of any treatment effect of minocycline.
Clinical Trial Registration—URL: http://www.anzctr.org.au. Unique identifier: ACTRN12612000237886.
Stroke remains a leading cause of death and a significant cause of adult disability in Western countries.1 Treatments that effectively reduce disability after acute stroke include aspirin, intravenous recombinant tissue-type plasminogen activator (rtPA), care in a specialized stroke unit, and hemicraniectomy in the setting of malignant middle cerebral artery infarction.
To date, the search for a safe and effective form of neurovascular protection to complement current therapies has been unsuccessful. Minocycline, a semisynthetic tetracycline antibiotic, has shown promise as a neuroprotective agent in animal models with induced spinal cord injury, traumatic brain injury, and ischemic and hemorrhagic stroke.2–17 It is highly lipophilic, crosses the blood–brain barrier, and acts via several pathways to inhibit microglial activation, reduce migration of T-cells, attenuate neural apoptosis, suppress free radical production, decrease central nervous system expression of chemokines and their receptors, and inhibit matrix metalloproteinases, in particular matrix metalloproteinase-9. In experimental stroke studies, minocycline improved behavioral function and reduced infarction volume and hemorrhagic transformation.13–16 It did not impair the fibrinolytic effect of rtPA14 and shows potential for expanding the time window for thrombolysis.17
The first human trial of minocycline in acute ischemic stroke randomly allocated 152 patients within 6 to 24 hours of stroke onset to open-label placebo (n=77) or oral minocycline 200 mg daily (n=74) for 5 days. At day 90, the odds of survival free of handicap (modified Rankin Scale [mRS], ≤2) was higher among participants assigned minocycline compared with placebo (91% minocycline versus 47% placebo; odds ratio [OR]: 11.5; 95% confidence interval [CI], 4.7–28; P<0.0001).18 Similar outcomes for the mRS≤2 were observed in a subsequent randomized, single-blind, open-label trial of oral minocycline 200 mg daily for 5 days versus control (oral vitamin B) in 50 patients with acute ischemic stroke (72% minocycline versus 37% control; OR 4.4; 95% CI, 1.4–14; P<0.01).19
In view of the strongly positive, yet imprecise, results of the 2 previously published human trials,18,19 we undertook an initial pilot study to explore their external validity. We also chose to examine the effect of minocycline administered by the intravenous route (to ensure rapid administration and absorption of the medication), in patients with both acute ischemic and hemorrhagic stroke (because of promising results of minocycline in animal models of both pathological types of stroke),15–17 and according to prespecified clinical subgroups (eg, rtPA versus no rtPA).
The trial was conducted in 2 tertiary hospitals and 1 general hospital in Perth, Western Australia.
The study was approved by the hospitals’ ethics committees. In 1 of the 3 sites, next-of-kin consent was permitted in instances where the patient was not able to give personal consent.
Before randomization, informed consent was obtained from all participants to be randomly assigned into either of the study groups with follow-up at 90 days.
This was a multicenter, prospective, randomized, open-label, blinded, end point evaluation pilot study.
Patients were recruited from the emergency departments and stroke units of the 3 hospitals.
Inclusion criteria were: (1) age ≥18 years; (2) onset of symptoms of stroke (ischemic or hemorrhagic) within 24 hours of administration of the trial intervention; (3) any measurable neurological deficit on National Institutes of Health Stroke Scale (NIHSS); and (4) ability to provide informed consent.
Exclusion criteria were: (1) evidence of other neurological disease that could confound the neurological assessment (eg, tumor, multiple sclerosis); (2) allergy to tetracycline/intolerance for minocycline; (3) systemic lupus erythematosus; (4) idiopathic intracranial hypertension; (5) concurrent treatment with retinoids; (6) participation in another clinical drug trial; (7) creatinine clearance (<30 mL/min by the Crockoft–Gault equation; (8) abnormal liver function (alanine transaminase [ALT]; >3× upper limit of normal); (9) thrombocytopenia <100×109/L; (10) concurrent infection (at enrollment) requiring antibiotic therapy; (11) pregnancy; (12) severe stroke or comorbidities likely to result in the patient dying within a week. Thrombolytic therapy was not an exclusion criterion.
At baseline, medical history, physical examination, vital signs, NIHSS, computed tomography, or MRI brain scan, and blood tests for full blood count, urea, creatinine and electrolytes, and liver function tests were obtained as per the Stroke Unit protocols.
Randomization and Intervention
Participants were randomly assigned, by means of a random numbers table, to either minocycline 100 mg administered intravenously, and continued 12 hourly for a total of 5 doses, or no minocycline.
Minocycline was obtained from GlobalRx in Japan. It was administered intravenously as a 500-mL solution into a large vein over a one hour period.
All patients received routine organized multidisciplinary stroke care in a stroke unit.
The primary outcome was survival free of handicap (mRS≤2) at day 90. The secondary outcomes are listed in Table 1.
Clinical assessments including NIHSS score and vital signs were repeated by examiners blinded to treatment allocation on days 1, 2, and 7.
All patients undergoing thrombolysis had a follow-up noncontrast-enhanced computed tomographic head scan within 24 to 48 hours of presentation. A neuroradiologist blinded to treatment arm assessed the degree of hemorrhagic transformation using the European Cooperative Acute Stroke Study classification.20
On day 7, patients were asked specific questions regarding the occurrence of headache and vascular events, and blood tests including full blood count, urea, creatinine and electrolytes, and liver function tests were repeated.
On hospital discharge, stroke classification according to the Trial of Org 10172 in Acute Stroke Treatment and Oxfordshire classifications was recorded.
On days 30 and 90, mRS, Barthel Index and questions regarding occurrence of headache and vascular events were performed by means of a telephone interview with the patient (or primary carer if the patient was dysphasic or demented) by a blinded staff member from a different study site.
At the end of the study the hospital files of all participants were reviewed regarding adverse events. Urinary tract infections were diagnosed on clinical grounds and confirmed by microbiology.
Sample-Size Calculations and Statistical Analysis
We based our sample-size calculation on the estimates derived from the trial of Lampl et al18 and calculated that, with 80% power and an α of 0.05, 39 patients per treatment group would be required.
The primary and secondary outcomes were assessed in the intention-to-treat and per protocol population. The proportions and rates were calculated for each group, and compared between the 2 groups using χ2 test. The effect of minocycline on the outcomes was expressed as a rate ratio, the rate for minocycline group divided by the rate for the routine stroke care group, with its 95% CI. A P value of ≤0.05 was considered as significant for all tests.
For the meta-analysis we pooled the results of our study and the published results of the other 2 human trials.18,19 Analysis was performed using Comprehensive Meta-analysis software (version 2.2; Biostat, Englewood, NJ). The effect of minocycline on the outcome (mRS≤2) was expressed as an OR with its 95% CI.
Ninety-five patients were enrolled in our study (Figure 1). Characteristics of the intention-to-treat population at baseline are shown on Table 2. The 2 groups were well balanced with regard to age, sex, medical history, medication use, and mean NIHSS score. Hemorrhagic stroke was present in 5 patients (11.4%) allocated minocycline and 6 (12.5%) allocated no minocycline. Thrombolysis with tPA was administered in 8 patients (18.2%) allocated minocycline and 6 patients (12.5%) allocated no minocycline.
Three patients allocated minocycline exhibited protocol violations and were withdrawn from the study within the first days of enrollment, 1 patient because of systemic lupus erythematosus and the other 2 because of a lack of measurable neurological deficit (Figure 1).
The mean time to treatment was 10.7 hours in the minocycline group: 14 patients received treatment between 0 and 6 hours, 15 between 6 and 12 hours, 6 between 12 and 18 hours, and 9 between 18 and 24 hours.
One death occurred in each group within the first 7 days and before the day 30 follow-up. One patient allocated no minocycline was lost to follow-up (Figure 1).
There was no significant difference in the primary and secondary outcomes between the 2 groups (Table 1). In the intention-to-treat population, 29 participants (65.9%) who had been allocated minocycline reached the primary outcome (mRS≤2) compared with 33 (70.2%) who were allocated no minocycline (rate ratio, 0.94; 95% CI, 0.71–1.25 and OR, 0.73; 95% CI, 0.31–1.71). Shift analysis showed a similar distribution of mRS score at 90 days in both groups (Table 1). Also, the 2 groups did not significantly differ in the subgroup analysis on primary outcome at day 90 (Table 3). With regard to the safety outcomes we detected similar numbers of death and hemorrhagic transformation in both groups. There was no significant difference with regard to headaches, rash, gastrointestinal symptoms, renal failure, and abnormal liver function (Table 4).
Urinary tract infections occurred in no patients allocated minocycline and in 7 (14.6%) allocated no minocycline (P=0.012; Table 4).
When we pooled the data from the three studies of minocycline (Figure 2), the odds of survival free of handicap (mRS≤2) increased by 3-fold (OR, 2.99; 95% CI, 1.74–5.16), but there was significant heterogeneity among the trials (I2=89.82).
The results of our small pilot study indicate that the use of intravenous minocycline for acute stroke within 24 hours of onset is safe and reduces urinary tract infection rates but does not reduce death or dependency at 90 days of follow-up. Analysis of prespecified subgroups in our trial did not reveal any improvement in stroke outcome in any group.
Our study adds to the efficacy data from the previous 2 randomized controlled trials,18,19 and the safety data from an intravenous dose-finding study.21 Our findings for the primary outcome measure (mRS≤2: OR, 0.7; 95% CI, 0.3–1.7) are discordant when compared with historical data from animal studies and the two other randomized controlled clinical trials in humans (Figure 2). A formal statistical test indicates significant heterogeneity among the clinical trials in humans (I2=89.82). There are several possibilities for this. The first is statistical heterogeneity caused by random error (chance) resulting from the small number of patients studied in all trials. The point estimates of the treatment effect in each trial differ substantially but the 95% CIs of the estimates of the treatment effects in each of the trials are wide and some overlap with each other. The second is clinical heterogeneity because of differences in the type of patients (ie, case mix) enrolled in each trial. Participants in our trial had more severe strokes (mean NIHSS score, 1.1–1.6 points higher) than those enrolled in the study by Lampl et al18 but less severe strokes than those enrolled in the Padma Srivastava19 study. The third is methodological heterogeneity caused by differences in the methods of the 3 trials. Our study differed from previous studies in the early initiation of minocycline therapy, the intravenous (versus oral) route of administration, and shorter duration of administration (2.5 versus 5 days). Although we assessed our primary outcome, the mRS score, at day 90 by means of a structured telephone interview, this method has been shown to have good agreement with face-to-face assessment and can thus be used reliably in the setting of a clinical trial.22 Finally, the animal studies were performed in rodents with large artery disease, whereas strokes caused by large-vessel atherosclerosis only accounted for about a third of patients in the three human trials.
The results of these randomized trials and a dose-finding study of intravenous minocycline collectively support the safety of minocycline in acute stroke.18,19,21 We also observed lower rates of urinary tract infection among participants treated with minocycline, which is plausible.
The main limitations of our trial, and of the 2 earlier human randomized controlled trials, are the potential for systematic error because of the open-label allocation of treatment and random error caused by the small number of patient outcomes evaluated. Although our trial was only a pilot study to explore the external validity of the results of 2 earlier trials,18,19 we acknowledge that our sample-size calculations, based on the strongly positive results of the study by Lampl et al,18 were unjustifiably optimistic. This is particularly so because we failed to account for potential beneficial effects on the primary outcome of additional rtPA in a minority of patients in both treatment groups, which could further reduce any difference in outcome between patients assigned minocycline versus control. Hence, the negative efficacy result of our study may be a false-negative (ie, a type II error) because of inadequate statistical power, and minocycline could potentially improve survival free of handicap by as much as 25% (ie, the upper 95% CI of the observed estimate of the rate ratio, 0.94; 95% CI, 0.71–1.25).
The strengths of our studies are the randomized design, which minimized imbalance among the treatment groups in prognostic factors, the near-complete patient follow-up that minimized attrition bias, and the outcome assessment by a blinded observer, which minimized observer bias. Unlike other studies of this agent we were able to ensure that the medication was bioavailable as we used the intravenous route.
Minocycline is inexpensive, easy to administer, and safe. It remains uncertain whether it is effective in reducing death and disability after stroke, although the meta-analysis (with the limitations of the included studies) is promising. Our group is currently undertaking a pilot study of intravenous minocycline in patients with ischemic stroke treated with intravenous tPA, as per usual clinical guidelines.23 The primary end point is the rate of hemorrhagic transformation measured by computed tomography and MRI in subjects concurrently treated with minocycline, compared with those treated with tPA alone. Additional clinical data will be collected at time points similar to the current and other studies. This combination approach with tPA is likely the best way to study any neuroprotective agent in acute ischemic stroke.
If minocycline is found to be effective, it could be administered in the prehospital setting, and therefore be particularly beneficial for patients in remote areas and for stroke patients in countries where specialized stroke unit care and intravenous thrombolysis with tPA is not available.24 It has the potential therefore, to become a widely available stroke treatment applicable to a broad range of settings.
We thank Sarah Claxton and Clare Williams from Stroke Research Office, Royal Perth Hospital, Australia, and Lawrence Wapnah, BMBS, FRACGP, Perth, Australia.
Sources of Funding
This study was supported by grants from the National Health and Medical Research Council, Canberra, Australia.
- Received January 21, 2013.
- Revision received January 21, 2013.
- Accepted May 29, 2013.
- © 2013 American Heart Association, Inc.
- Yrjänheikki J,
- Tikka T,
- Keinänen R,
- Goldsteins G,
- Chan PH,
- Koistinaho J
- Stirling DP,
- Khodarahmi K,
- Liu J,
- McPhail LT,
- McBride CB,
- Steeves JD,
- et al
- Teng YD,
- Choi H,
- Onario RC,
- Zhu S,
- Desilets FC,
- Lan S,
- et al
- Wells JE,
- Hurlbert RJ,
- Fehlings MG,
- Yong VW
- Machado LS,
- Sazonova IY,
- Kozak A,
- Wiley DC,
- El-Remessy AB,
- Ergul A,
- et al
- Lapchak PA,
- Chapman DF,
- Zivin JA
- Lee CZ,
- Xue Z,
- Zhu Y,
- Yang GY,
- Young WL
- Murata Y,
- Rosell A,
- Scannevin RH,
- Rhodes KJ,
- Wang X,
- Lo EH
- Fagan SC,
- Waller JL,
- Nichols FT,
- Edwards DJ,
- Pettigrew LC,
- Clark WM,
- et al