Intraventricular Fibrinolysis Versus External Ventricular Drainage Alone in Intraventricular Hemorrhage
Background and Purpose—The purpose of this study was to analyze the effect of intraventricular fibrinolysis (IVF) compared with external ventricular drainage alone on mortality and functional outcome in the management of intraventricular hemorrhage secondary to spontaneous supratentorial intracerebral hemorrhage.
Methods—The authors conducted a systematic review and performed a meta-analysis. They reviewed the PubMed, Cochrane Library, and Liliacs databases. In addition, they conducted a manual review of article bibliographies.
Results—Using a prespecified search strategy, 4 randomized and 8 observational studies were included in a meta-analysis. These studies involved a total of 316 patients with intraventricular hemorrhage at baseline, of whom 167 had IVF (52·8%). Pooled odds ratios of the impact of IVF on patient mortality, functional outcomes, and complications were calculated. The overall mortality risk decreased from 46·7% in the external ventricular drainage alone group to 22.7% in the external ventricular drainage+IVF group, corresponding to an overall pooled Peto OR of 0.32 (95% CI, 0.19 to 0.52). This result was highly significant with urokinase, not with recombinant tissue-type plasminogen activator. IVF was also associated with an increase in good functional outcome. There was no difference between the 2 groups in terms of shunt dependence and complications.
Conclusions—The combination of IVF and external ventricular drainage in the management of severe intraventricular hemorrhage secondary to small intracerebral hemorrhage in young patients was associated with better survival and functional outcome results. Urokinase and recombinant tissue-type plasminogen activator could not have the same therapeutic effects. Well-designed randomized trials with special considerations to the fibrinolytic agents are needed.
- cerebral hemorrhage
- fibrinolytic agents
- tissue-type plasminogen activator
- urokinase-type plasminogen activator
Intracerebral hemorrhage (ICH) is an important public health problem leading to high rates of death and disability in adults with an annual incidence of 10 to 30 per 100 000 people. ICH accounts for 2 million (10% to 15%) out of approximately 15 million strokes each year worldwide.1 In 40% of the cases, there is an associated intraventricular hemorrhage (IVH), which can lead to the development of obstructive hydrocephalus, especially when both the third and fourth ventricles are involved.2 IVH and obstructive hydrocephalus are independent predictors of poor outcome in patients with spontaneous supratentorial ICH.3 External ventricular drainage (EVD) is the treatment of choice for obstructive hydrocephalus in the setting of an IVH and allows for the control of intracranial pressure, thus lowering the mortality rate in the acute phase.4,5 However, no study so far has proven either its efficacy in terms of improved functional outcome or in preventing the development of communicating hydrocephalus.4 The latter is a result of inflammatory scarring of the subarachnoidal space induced by blood breakdown products and leads to increased morbidity.2,4,6 In the last 20 years, strategies have been developed in response to consistent proof of the deleterious effect on outcome by the presence of blood in the ventricular system. Such strategies include the removal of IVH by means of clot lysis through a ventricular catheter using fibrinolytic agents (urokinase-type plasminogen activator [urokinase], recombinant tissue-type plasminogen activator [rtPA]). Intraventricular fibrinolysis (IVF) has been proven to hasten the clearance of blood in the ventricles, thereby improving the cerebrospinal fluid circulation and providing faster control of intracranial pressure.7–12 It has been suggested that EVD with IVF might also improve outcome by limiting the toxicity induced by blood breakdown products acting on the adjacent parenchyma and the arachnoid membranes. This could possibly lower the incidence of communicating hydrocephalus.2 However, the potential advantages of IVF might be counteracted by complications such as secondary hemorrhage, infectious, or chemical ventriculitis and direct toxic effects of the fibrinolytic drugs on brain tissue.13 In the present study, we analyzed whether there was an advantage in terms of mortality and functional outcome using EVD with IVF versus EVD alone in the treatment of IVH secondary to a spontaneous supratentorial ICH. We conducted a systematic review of the literature and performed a meta-analysis.
Study Selection and Data Collection
Relevant studies were identified by systematic searches of the scientific literature for all reported studies comparing IVF and EVD alone in adult IVH (Supplemental Methods, http://stroke.ahajournals.org). Because a well-designed previous meta-analysis did not found randomized controlled trial of sufficient quality, we decided to include both randomized trials and observational studies in our present meta-analysis.14 So, any study, either prospective or retrospective, comparing EVD and EVD associated with IVF was included if: (1) patients enrolled were adult and presented with IVH associated with obstructive hydrocephalus that required EVD; (2) IVH was secondary to spontaneous supratentorial ICH; (3) in-hospital and/or other follow-up was ensured; and (4) the study reported all-cause mortality as outcome measures.
The primary outcome chosen was all-cause mortality at the end of follow-up. Secondary outcomes were: (1) good functional outcome (GFO), defined as a patient being able to care for him- or herself at 1 month and at 3 months; (2) chronic hydrocephalus requiring shunting; (3) occurrence of rebleeding; and (4) occurrence of ventriculitis.
Criteria for Assessing Quality
We assessed the quality of the randomized controlled trials (RCTs) using a checklist criteria: a study was included in a strict RCT meta-analysis if it met all points in the checklist. The studies that did not meet the criteria were considered for inclusion in a meta-analysis that also could include observational studies. Specific quality aspects were used to assess the studies such as control of confounding factors, minimization of selection bias, method used for IVH quantification, clear IVF protocol, and the sample size.
Data Synthesis and Statistical Analysis
In the absence of significant heterogeneity, studies were pooled using the Peto fixed-effects model.15 Otherwise, the studies were pooled using a Mantel-Haenszel random-effects model. Pooled OR were reported with 95% CIs. The weighted mean difference and 95% CI were used for continuous variables. Publication bias was explored visually by using funnel plot constructions. Three sensitivity analyses were carried out using prespecified criteria. All analyses were carried out with RevMan 5.0 software.16
Of the 704 articles screened, 20 were considered appropriate for inclusion, but 8 were excluded because they did not meet inclusion criteria (Figure 1).17–24 Twelve studies met all inclusion criteria and had sufficient data available (Supplemental Tables I and II).7,9,12,13,25–32 Ten articles were in English, 1 was in Hungarian, and 1 medical thesis was in French. These studies had a total sample size of 325 patients with IVH at baseline. Nine patients were excluded for analysis due to etiologies of IVH other than spontaneous supratentorial ICH or due to early exclusion in the original study.7,9,12,29 Finally, 316 patients were included from which 167 (52·8%) had IVF. The number of patients at baseline in the primary studies varied from 11 to 48. Four studies were described as RCT. Three studies did not have an optimal randomization method.7,26,31 One was not designed on an intent-to-treat basis.29 We included these 4 studies along with 8 others in an observational study meta-analysis. These 4 RCTs included a total of 73 patients with IVH, of whom 39 had IVF (53.4%). Eight studies were observational studies with a total of 243 IVHs with 128 IVFs (52.6%). One was prospective,28 3 were retrospective,13,17,32 and 4 had a prospective IVF group, and a retrospective control group.9,12,25,30 One study used a series published in another article as a control group and compared it with IVF associated with lumbar drainage.30 We in turn used a group of patients having EVD+lumbar drainage as a control.33 In 3 studies, the results on mortality and outcome were presented in such a way that we could not extract the data according to our criteria of data inclusion.13,26,30 Authors clarified the data in 2 studies.13,30 For complete data concerning patients characteristics and their treatments, refer to Supplemental Tables I and II. In summary, clear quantification of IVH was reported in 11 studies (91.6%); 8 used the Graeb score (mean Graeb score, 8.35). Seven studies reported the ICH volume before IVF (mean, 24.3 cm3; range, 8.5 to 40). Four studies used rtPA as a fibrinolytic agent (33.3%) and 8 used urokinase (66.6%). Eight studies used a clearly defined fibrinolysis protocol, which means that fibrinolytic agent, doses, and duration of IVF were predefined and kept the same. Control of confounding factors was poor in none of the studies, adequate in 4, and good in 8.
Primary Outcome: All-Cause Mortality at Follow-Up
Twelve studies presented information on all-cause mortality at follow-up (Figure 2). A fixed-effects meta-analysis was carried out, revealing a decrease in absolute mortality rate of 46.7% without IVF to 22.7% with IVF corresponding to an overall pooled OR for mortality of 0.32 (95% CI, 0.19 to 0.52). There was no heterogeneity (χ2 for heterogeneity, P=0.45). Moreover, the overall effect was highly significant (Z=4.52; P≤0.00001). Surprisingly, the results significantly differed when the analysis was limited to RCT or observational studies. The pooled OR was 0.12 (95% CI, 0.05 to 0.33) for the 4 RCTs and 0.44 (95% CI, 0.25 to 0.79) for the 8 observational studies (test for subgroup differences, χ2=4.84, P=0.03, I2=79.3%) Funnel plots were created to compare the SE of the log OR with the log OR for the RCT and observational studies, revealing asymmetrical plots for observational studies (smaller observational studies showing no statistically significant effects are missing), revealing possible publication bias (Supplemental Figure I).
The sensitivity analysis used the following criteria: (1) a sample size of at least 25 patients at baseline; (2) good or adequate quality of control of confounding factors; and (3) with a clear IVF protocol. Only 4 studies were included that met all these criteria (Supplemental Figure II).25–27,30 However, the pooled OR for these 4 studies (OR, 0.30; 95% CI, 0.15 to 0.62, test for heterogeneity: χ2=2.41, df=3, P=0.49, I2=0%) was essentially the same as for all 12 studies (test for subgroup differences P=0.90). A sensitivity analysis between studies using urokinase (167 patients) and those using rtPA (149 patients) shows a significant difference (test for subgroup differences, P=0.005); there is an obvious beneficial effect on survival in the urokinase group (OR, 0.17; 95% CI, 0.09 to 0.33), whereas this effect is less apparent in the rtPA group (OR, 0.73; 95% CI, 0.34 to 1.55; Figure 3). Finally, according to the control of confounding factors, pooled OR of mortality was also similar between the groups delineated (P=0.74), revealing no impact of the quality of the included studies on our results (Supplemental Figure III).
GFO at Hospital Discharge or at 1 Month
Six studies presented information on early GFO, including 2 randomized studies (Figure 4). A GFO of 31.4% was noticed in the EVD+IVF group compared with 7.1% in the EVD group. The corresponding pooled OR was 5.02 (95% CI, 2.07 to 12.20). There was no difference between randomized and observational studies (P=0.43).
GFO at ≥3 Months
Six studies presented data on long-term outcome (187 patients) with only 1 randomized study (Figure 5). A GFO of 54.5% was noted in the EVD+IVF group compared with 34% in the EVD group. By using the Peto fixed-effects model, we found heterogeneity (P=0.06, I2=53%). Then we used the Mantel-Haenszel random-effects model. The pooled OR was 2.35 (95% CI, 0.97 to 5.69) with an overall effect of 1.9 (P=0.06).
Chronic Hydrocephalus Requiring Shunting at Follow-Up
Eleven studies presented data on the occurrence of hydrocephalus requiring internal shunting (290 patients; Supplemental Figure IV). The incidence of shunting procedures decreased from 18.2% in patients treated with EVD alone to 13% in patients treated with EVD+IVF, corresponding to a pooled OR of 0.68 (95% CI, 0.35 to 1.30).
Occurrence of Rebleeding at Follow-Up
Seven studies addressed the issue of rebleeding at follow-up (190 patients; Supplemental Figure V). The rebleeding rate was greater in the IVF group (11·4%) than in the EVD group (6·4%) but did not reach significant values with a pooled OR of 2.39 (95% CI, 0.82 to 6.87).
Occurrence of Ventriculitis at Follow-Up
Ten studies addressed this issue and the results were similar for the 2 groups with a pooled OR of 1.02 (95% CI, 0.46 to 2.23; Supplemental Figure VI).
The results of our meta-analysis comparing EVD+IVF versus EVD alone in the management of IVH secondary to spontaneous supratentorial ICH suggests the superiority of the first therapeutic type (ie, EVD+IVF) in terms of survival and short-term functional outcome. The results on long-term functional outcome also suggest the efficacy of IVF, but data did not reach statistical significance, possibly due to lack of power. Moreover, we did not find any beneficial effect on preventing chronic hydrocephalus and the need for a subsequent internal ventricular shunt, yet neither did we find a statistically significant difference in terms of complications between the 2 therapeutic types although we expected a rise in the incidence of rebleeding in the EVD+IVF group. Interestingly, the results also suggest the superiority of urokinase to rtPA in terms of survival.
IVF exerts its beneficial effect by acting on the 2 physiopathological mechanisms that account for the morbidity and mortality of IVH: obstruction of cerebrospinal fluid outflow and the toxic effect of blood breakdown products on ependyma and subarachnoidal membranes.2 In the acute phase, IVF maintains the patency of cerebrospinal fluid pathways and the ventricular catheter, thus allowing better control of the intracranial pressure and decreasing the mortality rate.4,18,23,25 The faster clearance of blood from ventricles also shortens the contact of blood breakdown products with ependyma and subarachnoidal membranes, limiting the inflammatory scarring that this contact induces. This in turn allows a faster resumption of normal cerebrospinal fluid circulation.2 In this respect, we were surprised by the results on the incidence of chronic hydrocephalus where IVF did not show any beneficial effect. The disappointing results on chronic hydrocephalus and the hiatus between a substantial improvement in survival and a moderate improvement in functional outcome raises the question of the extent of the proven toxic effects of the fibrinolytic agents on cerebral tissue, which could counterbalance their beneficial effects.13,34,35 The neurotoxicity of rtPA is now well established unlike urokinase, for which such toxicity has never been reported.36 This could be a possible explanation for the less convincing results in the rtPA group. Nevertheless, it is important to note that, in some of these studies, the dose regimen used was quite high when compared with the dose used in the Clot Lysis Evaluating Accelerated Resolution of Intraventricular Hemorrhage Clinical Trial (CLEAR-IVH) 3 trial (1 mg 3 times a day). Using lower amounts of rtPA may preserve beneficial effects and prevent putative neurotoxicity.
Our study has limitations. The most important is that the results are issued from an observational study meta-analysis, thus limiting the strength of their argument. Our predefined strict inclusion criteria concerning RCT did not permit the inclusion of the 4 studies that were presented as such in a RCT meta-analysis. Therefore, we decided to include these 4 studies along with 8 other observational studies in an observational study meta-analysis. Furthermore, from the 8 observational studies, only 1 was prospective; all the others had an IVF group and/or an EVD alone retrospective group. On the other hand, even if the methodology in the RCT was not optimal, there was still randomization, all the studies (observational included) had a good or adequate control of confounding factors (12 of 12), and most of them did not have any major selection bias (Supplemental Table II). Interestingly, the effect size of IVF was significantly lower among observational studies compared with RCT (Figure 2; P=0.03, I2=79.3%). Channeling bias, in which more severe patients received more aggressive therapy, usually associated with observational studies could contribute to explain this quantitative discrepancy. It is noteworthy that the trends always favored significantly EVD+IVF against EVD alone regardless of study design (Figure 2). The results may also be explained in part by the “small study effect” phenomenon, which is the tendency for the smaller studies in a meta-analysis to show larger treatment effects. Small-study effects may arise because of publication bias or if smaller studies are of lower methodological quality. Indeed, the RCT had a smaller mean patient sample size than the observational studies (18.2 versus 30.3, respectively).37 However, this hypothesis appears no more consistent because the funnel plot revealed no asymmetry for the RCT, and the 4 RCTs were of good quality for the control of confounding factors (Supplemental Table II). When interpreting the results of a meta-analysis, we must also take into account the heterogeneity of the patient population studied, which in our case was limited, because most studies dealt with relatively young patients (mean age, 58.6 years) with a small ICH (mean, 24.3 cm3) and severe intraventricular involvement (mean Graeb score, 8.3). Therefore, we believe that the results are applicable to this particular set of patients.
There were 2 types of fibrinolytic agents used and we observed a substantial difference in the results between the 2. This could be due to a different toxicity potential, as stated earlier, or the result of the “small study effect” (and the possible occurrence of publication bias), because the mean patient sample size is not similar between the rtPA and urokinase groups (37.2 versus 20.9, respectively).37
Like in all systematic reviews, there is always a possibility of publication bias, as suggested by the funnel plot (Supplemental Figure I). The publication bias appears effective for the observational studies for which the treatment effect could be overestimated. However, there was no argument for publication bias in RCT or the effect of IVF in this subgroup appears highly significant, providing proof of its possible efficacy.
Moreover, we consider it important to point out our exhaustive search strategy in that we did not limit our search to English language articles and we also considered the gray literature for inclusion.
Our results are in favor of the use of IVF as an adjunct to EVD in the management of IVH secondary to spontaneous supratentorial ICH by underscoring a definite beneficial effect on survival and a moderate beneficial effect on functional outcome. Our meta-analysis also points out the differential effects between the 2 fibrinolytic agents and the potential superiority of urokinase among rtPA. However, because our data are not extracted from well-designed RCTs, we cannot recommend IVF actually, and results of the CLEAR-IVH trial are needed to conclude on the efficacy of rtPA. We also consider that a trial using urokinase is needed.
We express our gratitude to Dr Andrew F. Ducruet from Columbia University, NY, for his prompt response to our data requests.
The online-only Data Supplement is available at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.111.615724/-/DC1.
- Received February 2, 2011.
- Revision received April 19, 2011.
- Accepted April 20, 2011.
- © 2011 American Heart Association, Inc.
- Hanley DF
- Diringer MN,
- Edwards DF,
- Zazulia AR
- Adams RE,
- Diringer MN
- Rohde V,
- Schaller C,
- Hassler WE
Review Manager (RevMan) [computer program]. Version 5.0. Copenhagen: The Nordic Cochrane Centre, The Cochrane Collaboration; 2008.
- Coplin WM,
- Agris JM,
- Vinas FC,
- Buciuc R,
- Muizelaar JP,
- Diaz FG
- Naff NJ,
- Carhuapoma JR,
- Williams MA
- Tush GM,
- Tesoro EP,
- Prabhu S,
- Brown K,
- Bullock MR
- Coplin WM,
- Vinas FC,
- Agris JM,
- Buciuc R,
- Michael DB,
- Diaz FG,
- et al
- Huttner HB,
- Tognoni E,
- Bardutzky J,
- Hartmann M,
- Kohrmann M,
- Kanter IC,
- et al
- Staykov D,
- Huttner HB,
- Struffert T
- Porhiel V
- Huttner HB,
- Nagel S,
- Tognoni E,
- Kohrmann M,
- Juttler E,
- Orakcioglu B,
- et al
- Tsirka SE,
- Rogove AD,
- Bugge TH,
- Degen JL,
- Strickland S