Feasibility of Minocycline and Doxycycline Use as Potential Vasculostatic Therapy for Brain Vascular Malformations: Pilot Study of Adverse Events and Tolerance

Tetracycline class drugs are emerging as clinically applicable nonspecific MMP inhibitors that have the potential to enhance vascular stability, thus reducing the risk of spontaneous hemorrhage. They also possess neuroprotective properties. Settings investigated to date include cerebral ischemia [1], intracranial hemorrhage (ICH) [2, 3], neurodegenerative disorders [4, 5], traumatic brain injury [6], and atherosclerotic disease [7]. In clinical trials, tetracyclines decreased MMP in abdominal aortic aneurysms (AAA) and carotid plaques [7,8,9]. We have preliminary data that doxycycline can reduce MMP levels in brain AVM (BAVM) tissue [10]. Further, both AVMs [11] and intracranial aneurysms [12] display inflammation as part of their lesional phenotype. Tetracyclines appear to generally display anti-inflammatory properties, beyond their anti-MMP effects [13].

Unlike cancer-related chemotherapy that aims to shrink abnormal tumor tissue as cytotoxic therapy [14], MMP inhibition is not necessarily intended to obliterate the brain vascular malformation. Rather, the concept would be to stabilize vascular tissue and thereby decrease the risk of spontaneous ICH, i.e., a vasculostatic approach.

Although tetracyclines have been studied extensively, long-term studies are rare. Further, none of the studies have targeted patients with intracranial vascular diseases. We performed this phase I study to document the expected tolerance and adverse events in the same population, on whom phase II trials would be conducted. This study was designed to test the feasibility of using minocycline and doxycycline as potential long-term vasculostatic therapy for brain vascular malformations, and to estimate frequency and nature of adverse events that might be expected from a larger phase II or phase III trial effort. We hypothesized that there would be no difference in tolerability to minocycline and doxycycline treatment in patients with cerebrovascular diseases.

This was a prospective open-label safety pilot trial, using minocycline or doxycycline treatment for patients with BAVM or intracranial aneurysms for up to 2 years. The primary outcome was to estimate dose-limiting intolerance (DLI) in this patient cohort and compare DLI between the two agents; DLI was defined as treatment-related dose reduction or study withdrawal. The secondary outcome was to examine laboratory values indicative of potential tetracycline-related hemological, hepatical and renal toxicity. The Institutional Committee on Human Research approved the study protocol, and all patients gave written consent before enrollment. The study was registered with ClinicalTrials.gov (NCT00243893).

Patients with untreated BAVMs or aneurysms were recruited at the University of California, San Francisco. Patients had elected not to undergo invasive therapy, usually as a result of being deemed untreatable by conventional means. Other inclusion criteria were patients aged 13 and older; BMI within 30% of normal; white blood cell (WBC) > 3,800/mm³, creatinine ≤2.0 mg/dl and alanin aminotransferase (ALT) ≤2 times upper limit of normal. Exclusion criteria included patients with unstable medical illness, e.g., unstable angina or advanced cancer over the last 30 days; contraindications to tetracyclines, including allergy or other intolerance; prior tetracycline use within 2 months of baseline visit; history of systemic lupus erythmatosis or vestibular disease, except benign positional vertigo; history of noncompliance with treatment or other experimental protocols.

Patients received either oral doxycycline or minocycline (100 mg twice a day). This dose was based on prior clinical trials that demonstrated a significant reduction of plasma MMP-9 [9] and tissue MMP-9 in AAA patients [7, 9]. Our previous pilot experiments with AVM tissue showed similar effects [10]. Besides the study drug and the experimental procedures, there was no change in patient management.

Follow-up was conducted through telephone contact (weekly in the first 3 months, monthly for the next 3 months, and quarterly thereafter), laboratory evaluations, and patient reports of suspected adverse events. Adverse events and intercurrent neurological events were monitored. Compliance was monitored by patient self-report, phone contacts and monthly assessment of pill counts by returned blister packages. Levels of creatinine and ALT, hemoglobin, platelets and WBC were measured at baseline, then 3 months, 6 months, and every 6 months thereafter.

The general algorithm for treating side effects was a stepwise progression, based in part on the initial severity of the side effect: (a) counsel patient on preventative measures; if unsuccessful, then (b) temporarily interrupt medication for 1 week or reduce to once a day dosing. If (b) was successful, the patient resumed normal dosage. If not, (c) discussion took place with primary physician to either try dose reduction again or, as a last resort, consider withdrawal from trial.

Serious adverse events (SAEs) were defined according to the FDA MedWatch definition (death, life-threatening, hospitalization, disability, or required intervention to prevent permanent impairment or damage).

If a patient could not resume the protocol-specified dose after the temporary interruption or dose reduction, then this was counted as the endpoint, DLI. Comparison of the study drugs was analyzed using Kaplan-Meier survival curves with a log-rank test, and a Cox regression to adjust for patient’s age, gender and disease type.

Differences in baseline group characteristics were evaluated using Mann-Whitney U test, and categorical variables were analyzed using two-sided Fisher’s exact test. Differences in lab values between groups over time were tested using repeated measures with a general linear model. All reported values were mean ± SD, unless stated otherwise. A two-sided p value <0.05 was considered statistically significant.

Twenty-six untreated patients with BAVMs (n = 12) or aneurysms (n = 14) were enrolled in the study (table 1). To date, 10 patients have completed 2 years of treatment. The mean follow-up duration was 17.6 ± 7.7 months for doxycycline and 18.7 ± 7.2 months for minocycline. No significant differences were found between the study groups regarding baseline clinical characteristics. Nine patients are still actively receiving drug with remaining study time ranging from 1 to 15 months.

Treatment-related adverse events occurred in 13 of the 26 patients enrolled, resulting in 3 (11.5%) dose reductions and 5 (19.2%) withdrawals. Two of 3 patients with dose reduction tolerated the 100-mg daily dose subsequently and finished the 2-year course, while 1 later withdrew from the study. In addition, 3 events were judged as not treatment related, including 2 SAEs (1 ICH leading to death after taking doxycycline for 22 months and 1 ICH after taking minocycline for 18 months), and 1 dropout due to lack of desire to continue being in the study. Overall tolerability as indicated by DLI was 69% for minocycline and 77% for doxycycline. Kaplan-Meier survival analysis (fig. 1) showed no difference of DLI between treatment groups (log rank p = 0.70), but DLI showed a trend that was greater in the minocycline group compared to doxycycline [HR (95% CI) = 3.06 (0.52–18.11), p = 0.217], after controlling for the effects of female gender [HR = 13.4 (1.4–129.4), p = 0.03], aneurysms [HR = 4.0 (0.6–25.6), p = 0.15], and age [HR = 0.99 (0.95–1.03), p = 0.61]. As a sensitivity analysis, we included all events (composite of DLI, 2 hemorrhages and 1 dropout), and found similar results (log rank p = 0.49).

The most reported side effect was photosensitivity (6/26; 23%), followed by vertigo (3/26; 12%), yeast infection (2/26; 8%), gastrointestinal symptoms (2/26; 8%), hyperpigmentation (1/26; 4%), and allergic reaction (1/26; 4%; table 2). Most of the adverse events were mild and were managed according to the general algorithm. Specifically, photosensitivity and hyperpigmentation were managed by increased skin protection from exposure to ultraviolet light. Vertigo was managed by temporary interruption of the drugs. Yeast infection was treated with antifungal medications. Patients who developed gastrointestinal symptoms were recounseled to take the drugs with food and not to take them near bedtime. One patient who experienced an allergic reaction withdrew from the study. Yeast infection, vertigo and hyperpigmentation only occurred in the minocycline group, while the allergic reaction only occurred in the doxycycline group. The incidence of side effects was reported more frequently in the minocycline group (61%) than in the doxycycline group (39%), although the difference did not reach statistical significance with our sample size. The 2 incidences of SAEs, 1 hemorrhage each in AVM and aneurysm patients, rendered the annual bleeding rates for them to 5.6% (95% CI = 0.1–31.2%)and 5.8% (95% CI = 0.1–32.4%), respectively.

Since thrombocytopenia, azotemia and autoimmune hepatitis, although rare, could be potential side effects of tetracyclines, periodical laboratory values of hemological, hepatical and renal function were examined (table 3). While the WBC decreased over time in the minocycline group compared to the doxycycline (p < 0.01), creatinine increased in the minocycline group compared to the doxycycline group (p = 0.04). Hemoglobin (p = 0.12), ALT (p = 0.69) and platelets (p = 0.45) showed no difference between the groups. The values for WBC and creatinine remained within the range of normal values and thus were considered not clinically significant.

This study has shown a trend toward more frequent side effects and increased DLI from the minocycline treatment, although the difference between minocycline and doxycycline did not reach statistical significance in this small pilot study. There were no trends showing an increase in adverse events compared to other tetracycline clinical trials [9, 15].

The percentage of side effects observed in this study (50%) was similar to, or even slightly lower than, previously reported clinical trials in other cardiovascular or neurological diseases. For instance, 61% of the subjects reported side effects in a multicenter study with 6 months of doxycycline at the same dosage in AAA patients [9]. The use of broad spectrum antibiotics for extended periods (1–2 years) to influence cardiovascular disease, e.g., through treatment of a potential chlamydial infection, has not revealed a significant issue with side effects regarding safety or compliance [16]. In recent trials examining the role of minocycline in neurodegenerative diseases, a futility trial in Parkinson’s disease revealed 77% tolerability and reported upper respiratory symptoms, joint pain and nausea as the most common side effects [15].

While the range of adverse events was similar in both drugs, side effects were less frequent in the doxycycline group. In our study population, the rate of gastrointestinal side effects was very low in both groups, in contrast to others showing that doxycycline was associated with a much higher risk of upper gastrointestinal disorders [17]. Vertigo, consistent with previous reports, occurred only in the minocycline group in our study. In contrast to reported hypersensitivity reactions to minocycline [18], we had only one case in the doxycycline group, and none in the minocycline group. Our monitoring of laboratory values did not reveal signs of thrombocytopenia, hepatal or renal dysfunction, although they have been documented as side effects by others [19]. None of our patients reported musculoskeletal toxicity, a common side effect in broad-band MMP inhibitors [20] or lupus-like syndrome, to minocycline [21, 22]. Although gender difference related to drug tolerability has not been reported in tetracyclines, our results suggested female gender as a possible predictor for increased intolerance. On the other hand, age did not seem to have an effect on DLI.

The two occurrences of hemorrhage in our study were not unexpected considering prior natural history studies. Increased age, initial hemorrhagic presentation, deep brain location, and exclusive deep venous drainage have been found as independent predictors of AVM hemorrhage in the untreated course after diagnosis [23, 24]. Thus, the 82-year-old AVM patient with ICH in our study, who was at an advanced age and had a hemorrhagic presentation, could be at elevated risk. The ISUIA study found that aneurysm size and location are predictors of bleeding, with a first year rupture risk of 17% for giant aneurysms, reaching 50% at 5 years for posterior circulation lesions [25]. The aneurysm hemorrhage in our study occurred in a patient harboring a giant basilar tip aneurysm. Our point estimates of the observed hemorrhage rate were 5.6%/year for AVMs and 5.8%/year for aneurysms, both of which were well within those reported in previous studies.

We feel that this is an appropriate time to report the overall tolerability and feasibility of the study, since most of the side effect-related withdrawals experienced by the patients occurred within 18 months of the study. To date, the majority of our patients have been enrolled for more than 18 months, making this the longest reported follow-up of tetracycline treatment in neurological and vascular diseases compared to previous trials [7,8,9, 15]. The retention of patients in this long-term study has been encouraging, as 69% have remained in the 2-year treatment course. Other shorter-term tetracycline clinical trials have reported higher retention rates, including 92% in the 6-month doxycycline treatment of AAAs [9], and 89% in 12-month minocycline administration with Parkinson’s disease. The length of the study appears to be a key factor, given 12 months as the mean time for all withdrawal events in our study. Allowing patients to take the highest tolerated dose, instead of fixed doses, may have reduced withdrawal rates.

This pilot study suggests that long-term tolerability of minocycline and doxycycline in patients with brain vascular malformations is sufficient to support a larger phase II or III trial to examine efficacy endpoints.

The following members of the UCSF AVM study project also contributed to this work: S. Claiborne Johnston, MD, PhD; Nerissa Ko, MD; Vineeta Singh, MD; Wade Smith, MD, PhD; Michael McDermott, MD; Nalin Gupta, MD; Heather J. Fullerton, MD; David Saloner, PhD.

The authors would like to thank Trudy Poon for statistical analysis, Brad Dispensa and Philippe Jolivalt for database management, and Voltaire Gungab for editorial assistance. This work was supported in part by the following grants: NIH NS34949 (W.L.Y.), and AHA 0730360N (C.Z.L.).

T. Frenzel and C.Z. Lee contributed equally to this work.