Nimodipine in aneurysmal subarachnoid hemorrhage: A randomized

study of intravenous or peroral administration

Aneurysmal SAH remains a stroke subtype with high morbidity and mortality rates despite
considerable advances in its diagnosis and management during the last decade.27 A major
reason for an unfavorable outcome following SAH is DIND caused by arterial cerebral
vasospasm typically developing between 4 and 10 days after ictus.4,12,21 Poor outcome
(death or dependence) has been reported in 70% of the patients who have suffered an SAH,
and it has been estimated that a DIND is implicated in ~ one-third of these cases.3,13
The pathophysiological mechanism underlying SAHrelated vasospasm is largely unknown,
and medical treatment has so far failed to prevent its occurrence. Treatment options include
hypervolemia, hemodilution, and induced hypertension (so-called triple-H therapy); calcium
antagonists; and angioplasty.16 Based on the hypothesis that vasospasm is at least partially
dependent on calcium influx in vascular smooth-muscle cells, the dihydropyridine calcium
antagonist nimodipine was introduced during the 1980s as a prophylactic treatment in
patients who experienced an SAH. Randomized trials have demonstrated that treatment
with nimodipine reduces the proportion of patients with DIND and poor outcome as well as
the number of cerebral infarctions3,19 and that this treatment is cost-effective.11 It is now
common practice in most centers to administer nimodipine to all patients who have had an
SAH. Because most trials have been conducted with peroral administration of the drug, this
route has become the most common. However, despite the fact that the intravenous
administration of nimodipine can induce arterial hypotension, is substantially more
expensive, and has not proven beneficial in randomized trials, several centers routinely
administer nimodipine intravenously. A major argument for a parenteral route has been the
demonstration of more reliable and stable plasma concentrations of nimodipine.14,24,28 A
direct comparison between nimodipine administered perorally and intravenously in regard to
clinically meaningful differences has never been published.

In the present study, patients who had an SAH were prospectively randomized to receive
either peroral or intravenous treatment with nimodipine. We evaluated the incidence of
clinical signs of vasospasm in the acute phase as well as the clinical and radiological outcome
3 months after SAH.

Methods
The study was undertaken in the Department of Neurosurgery, Lund University Hospital,
which is the only neurosurgical department in the southernmost part of Sweden, covering a
population of 1.6 million. Patients were recruited during a 30-month period between October
2002 and March 2005. Eligible for inclusion were patients with aneurysmal SAH in whom
randomization and nimodipine treatment could be initiated within 48 hours after ictus. The
study was approved by the ethics committee at the Lund University Medical Faculty and
Medical Products Agency in Sweden. Written informed consent was obtained from either the
patient or his or her next of kin. The diagnosis of SAH was made using CT or lumbar
puncture if the CT failed to detect any blood. Clinical status on admission was assessed using
the Hunt and Hess scale.9 Blood distribution on CT was graded according to the Fisher
scale.5 The presence of an intracranial aneurysm was confirmed with DS angiography within
24 hours after admission. Only patients with an angiographically verified aneurysm qualified
for the study. All patients were admitted to an ICU dedicated to neurosurgical patients where
the initial treatment and monitoring was performed. Aneurysms were occluded using either
an exovascular (clipping) or endovascular technique (coiling), usually within 48 hours of
SAH. Tranexamic acid was administered to prevent early rebleeding8 prior to permanent
occlusion (1 g intravenously, every 6 hours).

Randomization and Treatment Protocol

Patients were randomized to receive either peroral or intravenous nimodipine by drawing a
sealed envelope. One hundred slips with the text “per os” and 100 with “iv infusion” were
produced. All slips were put in identical envelopes, which were sealed and then shuffled.
After inclusion into the study, an envelope was drawn for each consecutive patient.
Nimodipine was administered as 2 30-mg tablets by mouth every 4 hours or as a continuous
intravenous infusion of 0.2 mg/ml at 10 ml/hour (Nimotop, Bayer AG). If the patient became
hypotensive (systolic blood pressure < 110 mm Hg), a temporary reduction in the infusion
rate was allowed (typically to 5 ml/hour). Patients who were unable to swallow were given
crushed tablets with water via a nasogastric tube. Nimodipine treatment was continued for at
least 10 days. Treatment exceeded 10 days if the patient showed signs of a DIND or
increased flow velocities as indicated by TCD ultrasonography (see below). Crossover of
administration could not be blinded to the clinician. If a patient assigned to peroral treatment
experienced significant gastric retention suggesting disturbed absorption of nimodipine, he or
she was withdrawn from the study.

Clinical Monitoring
Neurological status was monitored and documented 3–12 times every 24 hours, depending on
clinical status. Clinical deterioration with the onset of new focal neurological deficits or a
diminished state of consciousness (decline of > 1 point on the Glasgow Coma Scale) lasting
for at least 24 hours was interpreted as a DIND if other causes such as hydrocephalus,
rebleeding, or hyponatremia could be excluded. Sedated patients on ventilator treatment were
assessed neurologically through the temporary discontinuation of sedation. Measurements of
blood flow velocities in both MCAs according to TCD ultrasonography were obtained on a
daily basis. Blood flow velocities > 120 cm/second or increases of > 50 cm/second in 24
hours in the MCA were interpreted as indicative of vasospasm.1,29 Patients with a DIND or
increased blood flow velocities received, in addition to the nimodipine, hypervolemia-
hemodilution therapy. Induced hypertension was used very rarely and thus was not recorded
in the study. An external ventricular drain was inserted in comatose patients and in those with
clinical deterioration due to hydrocephalus. At 3 months after SAH, patients were assessed
according to the GOS.10 At the same time, MR imaging was performed to detect new areas
of cerebral infarction. Highsignal lesions on T2-weighted or FLAIR MR sequences
were compared with hypodense areas on the original CT scans. A neuroradiologist blinded to
the treatment groups performed the radiological evaluation.

Study End Points

The primary objective of this study was to determine whether the mode of nimodipine
administration (peroral or intravenous) influences the incidence of DINDs. Secondary
end points were blood flow velocities in the MCAs as measured by TCD ultrasonography,
clinical outcome based on the GOS, and new infarctions according to MR imaging studies 3
months after SAH. We also analyzed the need for hypervolemia-hemodilution therapy as well
as the time required in the ICU.
Statistical Analysis
Group comparisons were performed using the chisquare test for proportions and the Student
t-test for the number of days with elevated TCD ultrasonography values and the number of
days in the ICU. Probability values < 0.05 (2-tailed) were considered statistically significant.

Results
Of 171 patients admitted to our center with a diagnosis of SAH, 106 were randomized in the
study. Forty-nine patients were randomized to the peroral group and 57 to the intravenous
group. The main reasons for not participating in the study were an inability to give informed
consent (early death and/or no relatives available), an angiogram negative for aneurysm, and
a late arrival to our unit (> 48 hours after ictus). During the study process repeated interim
analyses were performed, and based on the results the inclusion process was interrupted after
106 patients. One patient in the intravenous group was lost at follow-up. Three patients in the
peroral group were excluded due to vomiting or gastrointestinal paralysis suggesting
impaired drug absorption. Clinical condition on admission did not differ between the
treatment groups, and neither did the amount of bleeding on CT (Fisher grade) nor age
distribution (Table 1). Sex distribution, type of aneurysm treatment (open surgery or
endovascular therapy), and number of patients with external ventricular drainage were also
similar in both groups. In contrast, aneurysm location differed between the 2 groups; there
were more internal carotid artery (ophthalmic, posterior communicating, anterior choroidal,
and internal carotid artery bifurcation) aneurysms as the cause of SAH in the peroral group
(chi-square = 8.50, p < 0.01) and more anterior cerebral artery (including anterior intravenous
group (chi-square = 5.23, p = 0.02). The proportion of patients with bacterial ventriculitis did
not differ significantly between the groups. There was no difference in the occurrence of
DINDs between treatment groups: 13 (28%) of 46 patients in the peroral and 17 (30%) of 57
in the intravenous group (chisquare = 0.03, p = 0.86; Table 2).

Measurements of blood flow velocity in the MCA according to TCD ultrasonography were
also similar in both groups (Table 2). In 4 patients (2 in each group), reliable TCD
ultrasonography examinations were not possible because of the cranial thickness. Twenty-
two (50%) of 44 patients in the peroral group and 25 (45%) of 55 in the intravenous group
had blood flow velocities > 120 cm/ second (chi-square = 0.20, p = 0.65). The average
number of days with blood flow velocities > 120 cm/second were also similar: 2.3 ± 3.1
(mean ± SD) in the peroral group and 2.1 ± 3.2 in the intravenous group (t = 0.19, p = 0.85).
Increases of > 50 cm/second in 24 hours were seen in 7 (16%) of 44 patients and 7 (13%) of
55 in the peroral and intravenous groups, respectively (chi-square = 0.20, p = 0.65).
Histograms of the daily blood flow velocities in both MCAs were strikingly similar (Fig. 1).
The number of patients treated with hemodilution and hypervolemia, an indirect sign of
vasospasm or impending vasospasm as judged by the clinician, was similar in both groups
(52 and 53%, chi-square = 0.00, p = 0.96). The mean time required in the ICU did not differ
between the groups: 10.9 ± 7.5 and 10.9 ± 7.1 days in the peroral and intravenous groups,
respectively (t = 0.02, p = 1.00).

The distribution of clinical outcomes 3 months after SAH was similar in both groups (Table
3). None of the patients ended up in a vegetative state (GOS Score 2). Five patients in the
peroral group and 8 in the intravenous group died. The number of patients with a favorable
clinical outcome (GOS Score 4 or 5) did not differ between therapeutic modalities: 35 (76%)
of 46 in the peroral group and 39 (70%) of 56 in the intravenous group (chi-square = 0.53, p
= 0.47). The number of patients with chronic hydrocephalus requiring cerebrospinal fluid
shunting was also analyzed. In the peroral group, 12 (26%) of 46 patients received a shunt, as
compared with 10 (19%) of 56 in the intravenous group (chi-square = 1.01, p = 0.31).
Magnetic resonance imaging follow-up was possible in 33 patients in the peroral group and
41 in the intravenous group 3 months after SAH. Reasons not to perform MR imaging
included death, claustrophobia, and pacemaker implant. The number of patients with new
infarctions did not differ significantly between the groups: 14 (42%) of 33 patients in the
peroral and 18 (44%) of 41 in the intravenous group (chi-square = 0.02, p = 0.90). An
analysis of only those who had clinical signs of vasospasm—that is, either a DIND or
pathological TCD ultrasonography measurements—did not reveal any significant differences
in the number of infarctions between treatment groups (Table 4). In surviving patients in
hom MR imaging was contraindicated, a follow-up CT was performed. No difference in new
infarctions (hypodense areas) was detected, but the groups were too small for statistical
analysis to be meaningful.

Discussion
The present study was designed to investigate whether the peroral administration of
nimodipine is as efficient as an intravenous route in preventing ischemic complications
following aneurysmal SAH. The comparison of peroral and intravenous administrations
revealed no differences in the occurrence of DINDs, flow velocities in the MCA according to
TCD ultrasonography recordings, the need for hypervolemia-hemodilution therapy, or the
length of stay in the ICU. In addition, there were no differences in clinical outcome or
number of new infarctions on imaging at the 3-month follow-up. The strikingly similar
results for the 2 treatment modalities were consistent for all studied variables, which suggest
that the findings are reliable despite the relatively small number of patients. The proportion of
patients that had a DIND (30%) or increased blood flow velocities in the MCA (50%) is in
agreement with values in several previous studies.3,6,12,17,25,27 A favorable clinical
outcome (GOS Score 4 or 5) in 70–75% of patients is also in keeping with data in the modern
literature,18,22,26 suggesting that our study population is, in general, representative of
patients who have had an SAH. A control group without nimodipine was not included in our
study because to do so could have been considered unethical and made patient recruitment
more difficult.
The gold standard for diagnosing cerebral vasospasm is DS angiography, which can detect
vasospasm in up to 70% of patients who have had an SAH.12 Thus, given its potentially
greater sensitivity, angiography might have detected differences in efficacy between peroral
and intravenous administrations in the present study. Nonetheless, because DS angiography
carries a low risk of vascular complications (thromboembolism or dissection) and because we
do not routinely use angiography to detect vasospasm in patients who have had an SAH, this
imaging method was not adopted in our study for practical and ethical reasons. Note,
however, that TCD ultrasonography is a safe and noninvasive technique and has been shown
to correspond well with radiological findings of cerebral vasospasm.15,23,29 A potential
statistical bias in the study groups might be the fact that 3 patients in the peroral group were
excluded because of gastric retention and vomiting, probably causing inferior drug
absorption. It is possible that these patients were in worse clinical condition and therefore
might have been more prone to vasospasm. Such bias is difficult to avoid with our study
design, and it would have been unethical not to offer such patients intravenous treatment
(thus leading to exclusion from the study as crossover was not allowed). A power analysis
was not conducted when designing the study given that our purpose was to determine
whether peroral and intravenous nimodipine treatment had similar efficacy in preventing
cerebral ischemia and not primarily to detect any actual differences. Instead, repeated interim
analyses were performed to determine the size of the study.
As in previous randomized studies on the clinical efficacy of nimodipine,3,19 neither plasma
nor cerebrospinal fluid concentrations were measured. Although the pharmacokinetics of
intravenous nimodipine have been thoroughly investigated,14 relatively little is known
concerning the pharmacokinetics of peroral nimodipine.24,28 Vinge and colleagues28 have
demonstrated considerable variations in nimodipine concentrations after peroral
dministration (range < 3–96 ng/ml, average 13.2 ng/ml prior to next oral dose at 45 mg every
4 hours) as compared with those after intravenous delivery when levels were more stable
(range 15–58 ng/ml, average 27 ng/ml at 2 mg/hour). The large variations in plasma
concentration with oral treatment probably reflect variability in first-pass elimination.28 In a
pilot study by Soppi et al.,24 plasma concentrations of nimodipine varied substantially after
peroral administration (area under the curve, range 0.1–21 μg × min/ml) as compared with
intravenous administration (range 2–18 μg × min/ml). In addition, the absorption of peroral
nimodipine was affected in patients with a decreased level of consciousness, and concomitant
phenytoin medication may reduce its bioavailability via enzyme induction (cytochrome P450)
in the liver. Other studies have shown a relationship between intravenous dosing of
nimodipine and blood flow velocities in the MCA as measured with TCD
ultrasonography.23,30 These data suggest that intravenous administration is more reliable in
terms of providing optimal bioavailability of nimodipine for the patient. In fact, parenteral
administration of nimodipine is the preferred treatment in several centers (particularly in
Europe, personal communications) for these reasons. However, results of the present study
suggest that such variation in plasma nimodipine concentrations has no significant clinical
relevance. Because intravenous administration is considerably more expensive (in Sweden ~
25-fold), carries a higher risk of induced hypotension,7,20 and requires invasive central
venous access, our results strongly suggest that nimodipine should be administered via the
oral route unless disturbed absorption or metabolism is suspected.

Conclusions
In summary, previous studies have demonstrated that calcium antagonists reduce the risk of
poor outcome and secondary ischemia after aneurysmal SAH.2,3 Although the peroral and
intravenous routes should have similar beneficial effects, benefits have been proven in
clinical trials only for peroral administration.3 Results in the present study show that the
efficacy of the peroral and intravenous administration of nimodipine is similar from a
practical clinical point of view. Thus, the incidence of DINDs, TCD ultrasonography
measurements, clinical outcomes, and number of infarctions on MR imaging were the same
for both types of administration. Because peroral treatment is safer and considerably less
expensive, we suggest that parenteral nimodipine treatment be restricted to patients with
derangement in enteral absorption or metabolism.

Disclaimer
The authors report no conflict of interest concerning the materials or methods used in this
study or the findings specified in this paper.

Acknowledgments
We thank Susanne Månsson for assistance in the collection of data, and Lennart Brandt,
M.D., Ph.D., for valuable discussions.