Objectives: To determine if intravenous or oral methylxanthines (pentoxifylline, propentofylline and pentifylline) are safe to administer and reduce early ( <= 28 days) and late ( < 1 year) case fatality and morbidity in patients with acute ischaemic stroke.Acute stroke is the third most common cause of death in the Western world and the most common cause of long-term adult disability. Although there are no proven safe acute treatments for ischaemic stroke, potential therapeutic strategies include antiplatelet, antileucocyte, vasodilator or neuroprotective agents (Bath, 1995).An alternative approach involves the use of agents which combine these properties such as the methylxanthine derivatives, pentoxifylline (oxpentifylline), pentifylline and propentofylline. These agents have mixed actions; they (i) inhibit cAMP- and cGMP-phosphodiesterases (thereby increasing the intracellular concentrations of these calcium-regulating cyclic nucleotides), (ii) block the adenosine transporter thereby preventing adenosine uptake and increasing adenosine levels (Andine, 1990), and (iii) block phospholipase (Ward, 1987). Additionally, propentofylline has been shown to increase brain GABA concentrations (Vukadinovic, 1986) and reduce glutamate release under ischaemic conditions (Andine, 1990; Miyashita, 1992).As a result of the above, methylxanthine derivatives are vasodilators, inhibit platelet aggregation and thromboxane A2 synthesis, inhibit phagocyte function (e.g. decrease the release of free radicals), increase prostacyclin synthesis from endoperoxides, reduce blood viscosity through increasing erythrocyte flexibility, decreasing erythrocyte aggregation and reducing plasma fibrinogen concentration (Ott, 1983), and may be neuroprotective (Ward, 1987) since adenosine inhibits glutamate release. Pentoxifylline has been shown to increase cerebral blood flow in patients with acute stroke (Hartmann, 1986). Ischaemic stroke is characterised by thrombosis, red cell, platelet and leucocyte aggregation, leucocyte infiltration, free radical generation, excess glutamate release, and reduced GABA release. Hence, pentoxifylline and analogues are candidate treatments for acute ischaemic stroke.Preclinical studies in gerbils and cats suggest that pentoxifylline and propentofylline reduce neural damage following ischaemia (Ganser, 1974; Hartmann, 1977; Bluhm, 1985; DeLeo, 1988). Similarly, uncontrolled open trials have suggested that pentoxifylline may improve stroke outcome (Schneider, 1983). As a result, a number of randomised placebo-controlled trials of pentoxifylline and propentofylline in cerebral ischaemia have been performed and are the subject of this Cochrane systematic review.Search strategy: Cochrane Stroke Review Group search strategy plus other searches of MEDLINE, EMBASE, ISI and Pharmaceutical Company records.
Selection criteria: All completed randomised controlled trials, published and unpublished, comparing pentoxifylline, propentofylline or pentifylline with placebo or control initiated within one week of stroke onset.
Data collection and analysis: Data on the following outcomes were extracted: early and late case fatality, neurological impairment, disability, quality of life, stroke recurrence, venous thromboembolism, and bleeding.
Main results: Six trials were identified, data is awaited from one. The five trials included 793 patients (4 trials of pentoxifylline, 393 drug versus 370 placebo, and one trial of propentofylline, 15 drug versus 15 placebo); no completed trials of pentifylline were identified. 34 early deaths occurred in those 408 subjects receiving a methylxanthine and 49 in the placebo group of 385 subjects, odds ratio (OR) 0.64, 95% confidence interval (CI) 0.41 to 1.02. Analysis of pentoxifylline alone similarly showed a non-significant reduction in the odds of early death, OR 0.65, 95% CI 0.41 to 1.04; this reduction was mainly due to the findings of one trial which found a highly significant reduction in early deaths, OR 0.10, 95% CI 0.02 to 0.44. Two trials assessed death and deterioration and found a non-significant reduction, OR 0.49, 95% CI 0.20 to 1.20. Only one trial, which used propentofylline, assessed late mortality, OR 0.70 (95% CI, 0.13 to 3.68). Published data relating to neurological impairment and disability were not in a form suitable for analysis. No data on quality of life, stroke recurrence, thromboembolism, or bleeding was available.
Conclusions: Insufficient patients have been studied and it is currently unclear whether methylxanthines reduce case fatality or morbidity. A large randomised controlled trial may be required to assess the safety and efficacy of pentoxifylline (or propentofylline). Routine administration of methylxanthines in patients with acute ischaemic stroke cannot be recommended at present.
Deterioration was defined as a reduction in neurological impairment score; disability was defined as dependency in activities in daily living; drug safety was defined as major systemic or intracranial bleeding; quality of life was measured by a recognised scale.
See: Collaborative Review Group search strategy
Relevant randomised trials were identified using the strategy described under ""Search Strategy"" in the details of the Cochrane Stroke Review Group. In addition, further search measures included: computerised searches of MEDLINE (from 1965), EMBASE (from 1981), ISI (from 1981) and the Ottawa Stroke Trials Registry; assessment of reviews of xanthine derivatives in cerebrovascular disease (Ward, 1987; Martindale, 1993); contact with the manufacturers of pentoxifylline, propentofylline or pentifylline (Hoechst, and Hoechst Roussel Pharmaceuticals); and contact with the principal investigators of the identified trials.Electronic search terms included: ""(stroke or cerebr*) and (pentoxifylline or oxpentifylline or propentofylline or pentifylline or Trental or HWA 285)"".Two of us (PB, FB) independently selected trials for inclusion in the review and sought additional information from the principal investigators of these trials. Information on the methods of randomisation, concealment of allocation, blinding, exclusion of cerebral haemorrhage, analysis (intention-to-treat versus efficacy analysis), drug dose, route and timing, the numbers of deaths, deterioration, and functional disability, were recorded.We assessed the methodological quality of trials using standard Cochrane criteria. We analysed by ""intention-to-treat"", tested for heterogeneity, and calculated a weighted estimate of the typical treatment effect across trials (odds ratio) using the Peto fixed effects model in the Cochrane Statistical package, REVMAN (Mac version 1.0). The effects of pentoxifylline and propentofylline were analysed separately.None of the trials reported quality of life, bleeding episodes or venous thromboembolism; the trials of Hsu and Huber gave disability results (Barthel Index) at two weeks but these were not dichotomised and cannot be analysed without the original patient data (Hsu, 1988; Huber, 1993).
The total number of subjects in the five analysed trials numbered 793. The average age of patients in the trials was 68 years and almost two-thirds were male. There was some evidence of statistical heterogeneity in the early case fatality analysis (chi sq 8.39, df=4, 0.05 <p <0.10), mainly due to the findings of one trial (Chan, 1993) (see below).
Early case fatality (within four weeks) was assessed in four trials of pentoxifylline (Hoechst, 1986; Wong, 1987; Hsu, 1988; Chan, 1993); a non significant reduction in early death of 35% (95% confidence interval, CI, 59% reduction to 4% excess) was observed. It is noteworthy that this result is almost completely dependent on the statistically significant mortality difference observed between the pentoxifylline and placebo groups in the trial of Chan which found a large reduction in the odds of early death of 90% (95% CI, 98% reduction to 56% reduction) (Chan, 1993). If the Chan trial is removed from this analysis, the reduction in early case fatality reduces to 21% (95% CI, 58% reduction to 50% excess). When combining both pentoxifylline and propentofylline, a non-significant 36% reduction (95% CI, 59% reduction to 2% excess) in the odds of early death was observed. Two trials assessed the combined outcome of early death or deterioration (Wong, 1987; Chan, 1993) and found a non significant reduction of 51% (95% CI, 80% reduction to 20% excess).
Only one trial studied late case fatality (less than one year); Huber and colleagues reported a non-significant reduction of 30% (95% CI, 87% reduction to 268% excess) (Huber, 1993). Hsu and Huber reported Barthel Indices at two weeks (Hsu, 1988; Huber, 1993); however, in the absence of the original data it is not possible to further analyse this data. None of the trials reported data relating to late disability, quality of life, stroke recurrence, or the incidence of venous thromboembolism or major bleeding.
These analyses are dependent on our intention to analyse early deaths as those that occurred within 28 days of stroke onset (as specified in the original protocol). However, Hsu and colleagues reported three extra deaths in the pentoxifylline group that occurred between days 29 and 35 although they had intended to report deaths up to four weeks (as detailed in their methods section); we have ignored these extra deaths since they are outside our four week cut-off whilst Hsu et al did not give details on their deaths for their control group beyond 28 days (Hsu, 1988).
Side-effects were reported in two trials. Both the Hsu and Hoechst trials found an excess of nausea and vomiting in those patients receiving pentoxifylline (Hoechst, 1986; Hsu, 1988).
This review summarises the available randomised controlled trials of pentoxifylline or propentofylline in acute ischaemic stroke. The overall results do not support the hypothesis that pentoxifylline or propentofylline definitely reduce early case fatality. However, the confidence interval was wide reflecting the low numbers of patients studied and pentoxifylline or propentofylline could equally reduce short-term case fatality by 59% or increase it by 2%.One trial compared pentoxifylline and aspirin with aspirin alone and found a significant (p <0.002) 90% decrease in short-term mortality (Chan, 1993); this same trial studied early death or deterioration and found this combined outcome to be non significantly reduced. The Huber trial of 27 patients studied late case fatality (less than one year) following propentofylline treatment and found no difference between the treatment groups (Huber, 1993).
Only one trial administered pentoxifylline within 12 hours (Hsu, 1988); the other trials delayed drug administration until later when ischaemic damage may be fixed and far less amenable to pharmacological intervention.
Two trials used stratification at the time of randomisation to prospectively assess the effect of drug intervention in sub-groups of patients. Chan and Kay stratified by cortical and lacunar infarction and analysed these subgroups separately. Whilst pentoxifylline appeared to offer no benefit to patients with lacunar infarcts, it significantly reduced (p <0.002) early (one week) case fatality in patients with cortical infarction (Chan, 1993). This trial gave aspirin to both patient groups and it is possible that aspirin may potentiate the antiplatelet effects of pentoxifylline thereby contributing to its apparent clinical efficacy (Chan, 1993). In contrast, the Hsu trial prospectively stratified patients by vascular territory (carotid versus vertebrobasilar); however, the subgroups did not differ in preliminary analysis and the results were published as a single group (Hsu, 1988). The Hsu trial found that patients with severe neurological deficits appeared to benefit most from pentoxifylline (Hsu, 1988); in contrast, Wong and colleagues found that pentoxifylline appeared to benefit milder patients (Wong, 1987). The findings that different subgroups of patients benefitted following treatment with pentoxifylline suggests that these were probably due to chance.
Three trials followed the intravenous pentoxifylline with oral pentoxifylline (Hoechst, 1986; Wong, 1987; Hsu, 1988). Hsu reported improved neurological deficit scores during the first three days of intravenous pentoxifylline but not during the subsequent oral treatment period between days four and 28 post-stroke; this difference was attributed to reduced pentoxifylline plasma levels during oral administration, as has been previously observed (Herskovits, 1981).
Pentoxifylline or propentofylline should not currently be used in the routine management of acute ischaemic stroke.
We propose to set up a collaborative group of the pentoxifylline and propentofylline trial investigators. If the original trial data are still available, particularly those relating to neurological impairment, disability, blood pressure and side effects, it may be possible to further assess the effects of pentoxifylline or propentofylline on morbidity and side-effect rates.
References to studies included in this review
Chan 1993 (published data only) Chan YW. Pentoxifylline in the treatment of acute ischaemic stroke - a reappraisal. Aust N Z J Med 1991; 21(4 suppl 2): 601-. Chan YW, Kay CS. Pentoxifylline in the treatment of acute ischaemic stroke - a reappraisal in Chinese stroke patients. Clin Exp Neurol 1993; 30: 110-6. Hoechst 1986 (unpublished data only) Hsu 1988 (published and unpublished data) Pentoxifylline Study Group. Pentoxifylline (PTX) in acute Ischemic stroke. Stroke 1987; 18: (1),298-. Hsu CY, Norris JW, Hogan EL, Bladin P, Dinsdale HB, Yatsu FM, Earnest MP, Scheinberg P, Caplan LR, Karp HR, Swanson PD, Feldman RG, Cohen MM, Mayman CI, Cobert B, Savitsky JP. Pentoxifylline in Acute Nonhemorrhagic Stroke. A randomised, placebo-controlled double-blind trial. Stroke 1988; 19: (6),716-722. Hoechst. Clinical/statistical report for pentoxifylline (Trental, BL 191) protocol 402A. Hoechst report 1988; : -. Huber 1993 (published data only) Huber M, Kittner B, Hojer C, Fink G R, Neveling M, Heiss W-D. Effect of Propentofylline on regional cerebral glucose metabolism in acute ischemic stroke. J Cereb Blood Flow and Metab 1993; 13: 526-530. Huber M, Hojer C, Fink G R, Neveling B, Kittner B, Heiss W D. Adenosine reuptake inhibition by propentofylline improves brain glucose metabolism as measured by FDG-PET in human acute ischemic stroke. A randomized, placebo-controlled, double-blind study. 2nd Int Conf Stroke, Geneva 1993; : 2-. Wong 1987 (published and unpublished data) Wong WJ, Hu HH, Lo YK, Luk YO, Chu FL. Treatment of acute ischemic stroke. Bull Neurol Soc (Taiwan) 1987; 12: 20-. * indicates the major publication for the study References to studies excluded from this review Apollonio 1989 Apollonio A, Castignani L, Magrini L, Angeletti R. Ticlopidine-Pentoxifylline combination in the treatment of atherosclerosis and the prevention of cerebrovascular accidents. J Int Med Res 1989; 17: 28-35. Beyreder 1983 Beyreder J. Use of Pentoxifylline in the treatment of acute cerebrovascular insufficiency. Eur. Neurol 1983; 22: (supp 1),1-123. Hoechst 1993 Kittner B. Propentofylline acute stroke study (PASS). HWA 285/6/MN/301/ST. Hoechst Draft Study Protocol 1993; : -. Janaki 1980 Janaki S. Pentoxifylline in Strokes: A Clinical Study. J Int Med Res 1980; 8: 56-62. Santos 1993 Santos M-T, Valles J, Aznar J, Yaya R, Perez-Requejo J L. Effects of Dipyridamole, Pentoxifylline or Dipyridamole plus Pentoxifylline on platelet reactivity in patients with ischemic cerebrovascular insufficiency. Thrombosis Research 1993; 72: 219-229. Zorzon 1987 References to studies awaiting assessment Hoechst 1985 Hoechst. Etude multicentrique en double aveugle contre placebo dr l efficacite et de la tolerance du Torental par voie orale dans les infarctus sylviens a la phase aigue. Hoechst internal report 1985; : -. Additional references Andine 1990 Andine P, Rudolphi KA, Fredholm BB, Hagberg H. Effect of propentofylline (HWA 285) on extracellular purines and excitatory amino acids in CA1 rat hippocampus during transient ischaemia. Br J Pharmacol 1990; 100: 814-8. Bath 1995 Bath PMW. Treating acute ischaemic stroke. Br Med J 1995; 311: 139-40. Bluhm 1985 Bluhm RE, Molnar J, Cohen MM. The effect of pentoxifylline on the energy metabolism of ischemic gerbil brain. Clin Neuropharm 1985; 8: 280-5. DeLeo 1988 DeLeo J, Schubert P, Kreutzberg GW. Protection against ischemic brain damage using propentofylline in gerbils. Stroke 1988; 19: 1535-9. Ganser 1974 Ganser V, Boksay I. Effect of pentoxifylline on cerebral edema in cats. Neurology 1974; 24: 487-93. Hartmann 1977 Hartmann JF, Becker RA, Cohen MM. Effects of pentoxifylline on cerebral ultrastructure of normal and ischemic gerbils. Neurology 1977; 27: 77-84. Hartmann 1986 Herskovits 1981 Herskovits E, Vazquez A, Famulari A, et al. Randomized trial of pentoxifylline versus acetylsalicylic acid plus dipyridamole in preventing transient ischaemic attacks. Lancet 1981; i: 966-8. Martindale 1993 Miyashita 1992 Miyashita K, Nakajima T, Ishikawa A, Miyatake T. An adenosine uptake blocker, propentofylline, reduces glutamate release in gerbil hippocampus following transient forebrain ischaemia. Neurochem Res 1992; 17: 147-50. Ott 1983 Ott E, Fazekas F, Lechner H. Haemorheological effects of pentoxifylline in disturbed blood flow behaviour in patients with cerebrovascular disease. Europ Neurol 1983; 22: (Suppl 1),-7. Schneider 1983 Schneider R, Schmid-Schoenbein H, Kiesewetter H. The rheological efficiency of parenteral pentoxifylline (Trental) in patients with ischemic brain lesions. Europ Neurol 1983; 22: (Suppl 1),-104. Vukadinovic 1986 Vukadinovic V, Stefanovich V, Rakic L. Effect of propentofylline on the GABA system of a rat brain. Drug Dev Res 1986; 7: 87-94. Ward 1987 Ward A, Clissold SP. Pentoxifylline. A review of its pharmacodynamic and pharmacokinetic properties, and its therapeutic efficacy. Drugs 1987; 34: 50-97. Extramural sources of support to the review Intramural sources of support to the reviewHentschel B, Forthofer R. Efficacy, safety and tolerance of Trental (pentoxifylline) in the treatment of acute non-hemorrhagic focal cerebral infarction (12-36 hours after onset). HRPI protocol 402 B - modified. Final report. Hoechst 1986.Aschenbrenner KM. Addendum to the final report. Efficacy, safety and tolerance of Trental (pentoxifylline) in the treatment of acute non-hemorrhagic focal cerebral infarction (12 - 36 hours after onset). HRPI protocol 402 B - modified, Hoechst AG 1987.
Zorzon M, Monti F, Luogo TP, Cazzato G. La Ticlopidina e la pentosifillina nell infarto cerebrale acuto. G. Clin. Med;LXVIII,11:569-572.
Hartmann A. Haemorheological treatment of acute cerebral ischaemia. In: Gotoh & Lechner (Eds). Clinical haemorheology - a new approach to cerebrovascular disease, pp. 65-73, Royal Society of Medicine Services, London, 1986.
Martindale. Oxpentifylline. London: Edition 30. 1993.
Short title Pentoxifylline in acute stroke. Reviewer(s) Bath PMW, Bath FJ, Asplund K Date of most recent amendment 19 August 1997 Date of most recent substantive amendment 11 June 1996 Contact address Prof Philip Bath
Division of Stroke Medicine
City Hospital
Hucknall Road
Nottingham
UK
NG5 1PB
Telephone: + 44 115 962 7786
Facsimile: + 44 115 962 7992
E-mail: philip.bath@nottingham.ac.ukEditorial group Cochrane Cochrane Stroke Group Editorial group code HM-STROKE This review should be cited as :
Bath PMW, Bath FJ, Asplund K. Pentoxifylline, propentofylline and pentifylline in acute ischaemic stroke (Cochrane Review). In: The Cochrane Library, Issue 4, 1998. Oxford: Update Software.Sources of support
Keywords
CEREBRAL-ISCHEMIA / drug-therapy; PENTOXIFYLLINE / therapeutic-use; XANTHINES / therapeutic-use; THEOBROMINE / analogs- &-derivatives; VASODILATOR-AGENTS / therapeutic-use; THEOBROMINE / therapeutic-use; CEREBRAL-ISCHEMIA / mortality; RANDOMIZED-CONTROLLED-TRIALS; HUMAN; ODDS-RATIO; SAMPLE-SIZETable of comparisons
Fig 01 PENTOXIFYLLINE OR ANALOGUE VS CONTROL IN ISCHAEMIC STROKE: ALL TRIALS
Tables of other data
Tables of other data are not available for this review
| Study | Method | Participants | Interventions | Outcomes | Notes |
|---|---|---|---|---|---|
| Chan 1993 | Double-blind placebo-controlled. Single centre. Intention-to-treat. No loss to FU. Randomisation by sealed numbered opaque envelopes. | Hong Kong. Males 73, females 37. Mean age 68 years. Enrollment between 36 to 48 hours. 100% CT. Stratified by cortical and lacunar infarction. | Rx: Continuous intravenous PTX (Hoechst, China) 600mg daily for 5 days with oral aspirin 150mg daily. PL isotonic saline/matching PL injections with oral aspirin 150mg daily. | Death at one week. Neurological impairment (modified Scandinavian Stroke Scale) at entry and 1 week. | Ex: previous stroke, other neurological disease, TIA, cerebral haemorrhage, systemic disease. FU: 1 week. |
| Hoechst 1986 | Double-blind placebo-controlled. Multicentre (14 sites). Intention-to-treat (efficacy data also available). 8 patients incompletely followed-up. Randomisation method not detailed. | Austria, Belgium, Mexico, Sweden. 153 male, 111 female, 2 unknown. Mean age 67 years. Enrollment between 12-36 hours. 100% CT (optional scan at 7 days). | Rx: PTX (Trental) loading with 50mg i.v. then 16mg/Kg/day (max. daily dose 1200mg) for 72 hours (or 7 days if dysphagia). After infusion, 400mg PTX p.o. t.d.s. (or b.d.) for 4 weeks. | Case fatality at day 28. Neurological impairment (level of consciousness, motor, cranial nerve, higher cortical, and sensory function) on days 1, 2, 3, 4, 7, 14, 21, 28. Functional impairment (National Survey of Stroke) at days 4, 7, 14, 21, 28. | Ex: recurrent stroke, intracranial haemorrhage, space-occupying lesion (tumour, vascular malformation), cranial trauma, multiple cerebral emboli if heart disease or TIA, severe chronic disease, drug hypersensitivity, anticoagulant therapy. FU: 4 weeks. |
| Hsu 1988 | Double-blind, placebo-controlled. Multicentre (13 sites). Intention to treat. No loss to FU. Randomisation by computer. | USA, Canada, Australia. 172 male, 125 female. Age >35 years, mean age 69 years. Enrollment within 12 hours. Stratified by carotid and vertebrobasilar territories. 100% CT. | Rx: PTX iv loading 50 mg over 1 min, then 16mg/Kg/day (up to 1200 mg/day) for 72 hours (or 7 days if dysphagia). After infusion 400 mg PTX p.o. t.d.s for 25 days. PL: isotonic saline and matching tablets. | Death at 28 days. Neurological impairment (level of consiousness, motor, cranial nerve, higher cortical, and sensory functions) at baseline, days 1-4, 7, 14, 21, 28. Functional impairment (Barthel Index) at days 7, 14, 21, 28. | Ex: pregnancy, previous stroke, TIA, coma, cerebral haemorrhage, systemic diseases, intracranial lesion other than stroke, anticoagulation. FU: 4 weeks. |
| Huber 1993 | Double-blinded placebo controlled. Intention to treat. No loss to follow up. Randomisation by sealed numbered opaque envelopes. | Germany. 17 male, 10 female. Mean age 65 years. Enrollment <48 hrs. 100% CT. | Rx: PPF (Hoechst) i.v. infusion 1,200 mg within 72 hours of ictus for 7 days. PL: carrier soln. Both groups received 6% hydroxyethylstarch (500 ml/day i.v. for 14 days) and 10% glycerol i.v. for 5 days. | Death at 2 weeks and 3 months. Neurological impairment (Mathews score) at admission and 2 weeks. Functional impairment (Barthel index) at 14 days and 3 months. FDG PET (rCMRglu) at admission and 14 days. | Ex: Mathew score >65, subarachnoid haemorrhage, intracerebral haematoma, recent myocardial or cerebral infarction, cardiac arrhythmias, circulatory insufficiency, renal or hepatic failure, systemic infection, nonstabilised diabetes mellitus, coma. FU: 3 months. |
| Wong 1987 | Single-blind (patients) placebo-controlled. Single centre. Intention to treat. No loss to follow up. Randomisation by identical numbered vials. | Taiwan. 76 males, 14 females. Mean age not stated. Enrollment within 36 hrs. 100% CT. | Rx: PTX 900 mg/day iv and glycerine 1000 ml/day for 3 days, then 600-1200mg/day po (length of oral therapy not stated). Pl: glycerine 1000 ml/day iv. | Death at 28 days. Neurological impairment (Hoechst score, total score 100) and Barthel Index at 28 days. | Ex: ICH, TIA, ventrebrobasilar stroke, recent MI, malignancy. FU: 4 weeks. |
| Study | Reason for exclusion |
|---|---|
| Apollonio 1989 | Randomised trial of (PTX and ticlopidine) versus (aspirin, dipyridamole and buflomedil) versus (buflomedil and placebo). Excluded because treatment confounded and no placebo/open control. |
| Beyreder 1983 | Randomised trial of PTX versus hexobendine, ethamivan and etofylline. Excluded because no placebo/open control. |
| Hoechst 1993 | Proposed multinational multicentre phase III trial of PPF (1200 mg/day iv for 3 to 14 days then 300 mg tds po for 3 months) within 12 hours of acute ischaemic stroke. Trial cancelled by Hoechst; 0 of 1600 patients enrolled. Excluded because trial aborted prior to recruitment. |
| Janaki 1980 | Randomised trial of PTX versus xanthinol nicotinate. Excluded because no placebo/open control. |
| Santos 1993 | Patients with stroke 6-8 months before study. Excluded because not acute trial. |
| Zorzon 1987 | Randomised trial of PTX versus ticlopidine. Excluded because no placebo/open control. |