Pentoxifylline for treatment of venous leg ulcers: a systematic review

ARTICLES Pentoxifylline for treatment of venous leg ulcers: a systematic review Andrew Jull, Jill Waters, Bruce Arroll Summary Introduction Venous ...
Author: Sheryl Glenn
1 downloads 0 Views 88KB Size

Pentoxifylline for treatment of venous leg ulcers: a systematic review

Andrew Jull, Jill Waters, Bruce Arroll

Summary Introduction Venous ulcers are usually treated with compression therapy, but, because this treatment may not be effective for some people, adjuvant therapy could be beneficial. We did a systematic review of randomised controlled trials that compared pentoxifylline (with and without compression treatment) with placebo, or other treatments, in patients with venous leg ulcers. Methods We identified eight trials (547 adults), five of which compared pentoxifylline and compression with placebo and compression (n=445), and three of which compared pentoxifylline alone with placebo (102). Our main aim was to determine whether pentoxifylline, with or without compression, was effective in treatment of venous leg ulcers. Analysis was by intention to treat. Findings Pentoxifylline was more effective than placebo in complete healing or substantial improvement of venous leg ulcers (relative risk 1·49, 95% CI 1·11–2·01). Pentoxifylline with compression was also more effective than placebo and compression in complete healing (1·30, 1·10–1·54). Patients taking pentoxifylline had no more adverse events than those on placebo (1·25, 0·87–1·80). The most frequent adverse event was mild gastrointestinal disturbance (43%). Interpretation Our results suggest that pentoxifylline gives additional benefit to compression for venous leg ulcers, and is possibly effective for patients not receiving compression. Lancet 2002; 359: 1550–54

Centre for Evidence Based Nursing Aotearoa, Auckland Hospital, Auckland, New Zealand (A Jull RN), and Department of General Practice (B Arroll MBChB), University of Auckland, New Zealand (J Waters MBChB) Correspondence to: Andrew Jull, Centre for Evidence Based Nursing Aotearoa, Auckland Hospital, Private Bag 92024, Auckland, New Zealand (e-mail: [email protected])


Introduction Most leg ulcers are venous.1 High compression treatment (three-layer or four-layer short-stretch bandaging, or Unna boot) is an effective cure for venous ulcers,2 but despite compression, almost 30% of people remain unhealed after 1 year of treatment.3 Adjuvants to compression treatment could be beneficial. Pentoxifylline, a haemorrheological agent, reduces the viscosity of blood by increasing the flexibility of erythrocytes, encouraging migration of white cells, inhibition of aggregation of platelets, and lowering of the viscosity of plasma,4 actions that might correct microcirculatory disorders. Trials5,6 of the clinical effectiveness of pentoxifylline have had conflicting results, and the usefulness of the drug remains unclear.7 We therefore systematically reviewed randomised controlled trials to quantify the effect of pentoxifylline (Trental, Aventis Pharma, Lyon, France) on healing in venous leg ulcers, either as an adjuvant to compression, or compared with placebo.

Methods We searched the CENTRAL registers of the Cochrane Wounds Group and the Cochrane Peripheral Vascular Diseases Group to September, 2000. We identified no new controlled trials in our search of the Wound Review Group register to April, 2001. The CENTRAL registers differ from the Cochrane Controlled Trials Register in that they contain references that have not been confirmed as controlled trials, or trials that have not yet been forwarded for inclusion in the Cochrane register. The register of the Wounds Group contains citations from searches of 19 electronic databases, hand-searches of conference proceedings and five journals, and contacts with manufacturers. The register of the Peripheral Vascular Diseases Group contains citations from electronic searches of Medline and Embase, handsearching of conference proceedings and 38 journals, and contacts with manufacturers. We also reviewed the citations of the retrieved studies and review articles obtained from a search of Medline (1966–99) and Embase (1980–99). To identify unpublished or continuing trials, we contacted the Australasian medical director for the manufacturer of the drug (Aventis Pharma, Lyon, France). No language restrictions were placed on the search. Trial selection and data extraction We included trials if they had reported an objective outcome measure (percentage change in ulcer area, proportion of ulcers completely healed), or an operationalised measure (substantial improvement). Two reviewers (BA, AJ) independently assessed citations obtained from the search and selected trials for inclusion. We obtained reports of trials that satisfied the inclusion criteria, and those for which there was any doubt about exclusion. We translated only the methods and results

THE LANCET • Vol 359 • May 4, 2002 •

For personal use. Only reproduce with permission from The Lancet publishing Group.


Inclusion and exclusion criteria

A priori sample size calculation

Method of randomisation

Allocation concealment

Baseline comparability of groups


Intentionto-treat analysis

Exclusion only Yes

Not stated Not stated

Not stated Not stated


Not stated Yes

Double Double

Inclusion only

Not stated




Yes Yes Yes

Yes Yes Not stated

Block, sealed envelope* Sealed envelope Pharmacy* Not stated


Yes Double Yes Double Favoured control Not stated

Schürmann and Eberhardt12 Exclusion only

Not stated

Central list*


Favoured control Single


Not stated

Not stated



No Not stated no withdrawals Not stated no withdrawals Yes Yes† Not stated no withdrawals Not stated no withdrawals No

Arenas and Attoche8 Barbarino9 Colgan and colleagues5 6

Dale and colleagues Falanga and colleagues10 Herdy and colleagues11

Inclusion only


A=adequate concealment, B=concealment unclear. *Information obtained from investigators directly. †Includes two participants excluded from analysis after randomisation. We added these participants to the denominator in the treatment groups as treatment failures.

Table 1: Methodological characteristics of included studies

sections of reports that were not written in English. One reviewer (AJ), who was aware of the authorship and journal the trial was published in, extracted the data using a standard procedure. A second reviewer (JW) independently assessed the data for accuracy. All disagreements were resolved by discussion. We assessed the methodological characteristics of all included trials on variables required by the Cochrane Wounds Group (table 1). In reports in which the study details were unclear, we sought information from the investigators. Data analysis and statistical methods We used RevMan (Review Manager) version 4.1 statistical software to calculate relative risks and 95% CIs for every trial. Analysis was done by intention to treat. We used relative risk instead of odds ratios because the event rates were high. Heterogeneity between trials was assessed with the 2 test; heterogeneity was judged significant if p was less than 0·1. In cases in which heterogeneity was not significant, we used a fixed-effects model to calculate summary relative risk and 95% CIs, and applied a random-effects model to test the robustness of the result. The random-effects model was used for analyses with significant heterogeneity. Where clinical or methodological heterogeneity was present, we assessed sensitivity to determine the effect of variation. These analyses were done after the data had been extracted from individual studies, but before they were combined. Since increased Diagnostic Method


Colgan and colleagues5 Dale and colleagues6

Falanga and colleagues10

Herdy and colleagues11 Schürmann and Eberhardt12 Weitgasser13

Role of the funding source The sponsor of the study had no role in study design, data collection, data analysis, data interpretation, or in the writing of the report.

Results We included eight trials (table 2),5,6,8–13 two of which were not written in English.11,12 In most studies, the primary outcome was either complete healing of the patients’ reference ulcer (usually the largest ulcer) or of all ulcers on the reference leg,5,6,9 or data was provided for individual patients, from which we could calculate the number of patients healed.11,12 In one trial,10 data were presented in a life table, from which the number of patients healed was estimated independently by two reviewers. Investigators of two trials8,13 reported outcomes by operationally defined ratings such as good (complete healing or substantial reduction in ulcer area) and substantial improvement (60–94% reduction in ulcer area). In one trial,11 the area of every participant’s ulcer was reported at baseline and trial




Length of follow-up

Pentoxifylline 1200 mg daily


6 months

Clinical signs ABI>0·8 Clinical signs ABI>0·8 Clinical signs ABI>0·8 Doppler-determined venous reflux Clinical signs ABI>0·8 Doppler-determined venous reflux Clinical signs

Pentoxifylline 1200 mg daily plus 2-layer compression bandaging Pentoxifylline 1200 mg daily plus 2-layer compression bandaging Pentoxifylline 1200 mg daily plus either elastic single layer or 4layer compression bandaging (evenly balanced) Either pentoxifylline 1200 mg or 2400 mg daily plus compression by Unna boot and elastic bandage Pentoxifylline 1200 mg daily.

Placebo plus 2-layer compression bandaging Placebo plus 2-layer compression bandaging Placebo plus either elastic single layer or 4-layer compression bandaging (evenly balanced)

Healing and substantial improvement Complete healing

67 days

Complete healing

24 weeks

Complete healing

24 weeks

Placebo plus compression by Unna Complete healing boot and elastic bandage

24 weeks


Individual data

12 weeks

Clinical signs*

Pentoxifylline 1200 mg daily plus short-stretch compression bandages* Pentoxifylline 1200 mg daily

Placebo plus short-stretch compression bandages

Individual data

8 weeks


Complete healing or substantial reduction in ulcer area

8 weeks

Arenas and Attoche8 Not reported 9

ulcer healing is the desired benefit, the results are presented graphically with intervention on the right-hand side of the forest plot instead of the usual left-hand side. To assess how publication bias might have affected the findings, we prepared a funnel plot using RevMan, and assessed for significant asymmetry using Begg and Mazumdar’s rank-correlation test.

Not reported

ABI=ankle brachial index. *Data obtained from investigators directly.

Table 2: Summary of trials that met inclusion criteria

THE LANCET • Vol 359 • May 4, 2002 •


For personal use. Only reproduce with permission from The Lancet publishing Group.


Treatment Control Relative risk Weight Relative risk completion. Rather than report an n/N n/N (95% CI) (%) (95% CI) individual trial as continuous data, the operational definitions for healing 4·00 (0·61–26·12) Barbarino9 4/6 1/6 1·0 or substantial improvement from 5 Colgan 2·12 (1·23–3·65) 23/38 12/42 10·9 Arenas and Atoche8 were applied to 6 Dale 1·23 (0·97–1·55) 65/101 52/99 50·2 create categorical data for Herdy and 61/86 28/45 35·1 1·14 (0·87–1·49) Falanga10 colleagues.11 In one study,10 two different doses of pentoxifylline were 2/12 3/12 2·9 0·67 (0·13–3·30) Schürmann12 compared with placebo. We combined these two groups, and Total (95% CI) 1·30 (1·10–1·54) 155/243 96/204 100·0 analysed the sensitivity of the Test for heterogeneity p=0·17 combination. Investigators from three trials5,10,11 provided further 0·1 0·2 1·0 5·0 10·0 information to clarify their reports. Favours Favours control treatment In five of the eight trials,5,6,9,10,12 pentoxifylline was compared with placebo with a form of compression Figure 1: Relative risk of healing associated with pentoxifylline compared with placebo therapy as a standard treatment; in (with compression therapy as standard treatment) in patients with venous leg ulcers the other three trials,8,11,13 pentoxifylline on its own was compared with placebo. The allocation concealment were excluded (relative risk 1·30, relative risk of healing or substantial improvement for all 95% CI 1·09–1·54).9,12 studies combined was 1·49 (95% CI 1·11–2·01). This A second subgroup (three trials, n=102) compared result was robust to inclusion only of trials that used that pentoxifylline with placebo without compression used only complete healing as an outcome. Heterogeneity bandaging as a standard treatment.8,11,13 In this subgroup, between the groups of trials was significant (p=0·06). In patients in the treatment groups were almost 2·5 times six of eight trials,5,6,8–10 the number of adverse events did not more likely to heal completely than those in the placebo group (figure 2). Heterogeneity was not significant in this differ between patients in the treatment group and those subgroup (figure 2). However, Weitgasser13 judged on placebo (relative risk 1·25, 95% CI 0·87–1·80). Exclusion of the 2400-mg group in the trial by Falanga and substantial improvement as “considerable reduction” in colleaques10 had little effect on the risk of adverse events. ulcer sizes, without reporting a numerical criterion. The summary result was sensitive to exclusion of this trial In the studies in which adverse events were reported, the (relative risk 1·49, 95% CI 0·50–4·42). most frequent were minor gastrointestinal disturbances We created a funnel plot with all trials apart from that (loss of appetite, heartburn, nausea, vomiting, diarrhoea). which had zero events in both groups. The adjusted rankOther adverse events included headache, sleep correlation test suggests that the funnel plot is symmetrical disturbance, hot flushes, itching, and hypotension. Of the (Spearman’s =0·214, p=0·64). Similarly the funnel reported withdrawals, only 25% were for adverse events. plots of trials comparing compression and pentoxifylline Four of eight trials lasted for 8–12 weeks,9,11–13 and four to compression and placebo were not significantly lasted for 24 weeks.5,6,8,10 In the 108 participants of trials asymmetrical (Spearman’s =0·300, p=0·62). that lasted for less than 12 weeks (short duration), relative risk of healing or substantial improvement with pentoxifylline versus placebo was 2·30 (95% CI Discussion 1·30–4·10). However, the combination was sensitive to Our results showed that pentoxifylline improves healing exclusion of Weitgasser’s findings,13 which shared healing rates compared with placebo. Pentoxifylline was effective for treatment of leg ulcers, but the evidence for or substantial improvement as a single outcome. In the pentoxifylline on it’s own is not as strong as that for this 441 participants in trials lasting 12 weeks or longer, drug as an adjuvant to compression treatment. The finding relative risk of healing or substantial improvement was is stronger in trials that lasted longer than 12 weeks, 1·30 (1·10–1·54). This risk was not sensitive to exclusion although this result could be because the longer trials were of Arenas and Atoche,8 who reported healing or substantial of better quality than the shorter trials. improvement as a single outcome. Although which microcirculatory events link venous The heterogeneity recorded when all studies were hypertension with ulceration is not clear, one explanation combined was probably caused by differences in treatment between subgroups. One subgroup (five trials, n=441), which compared Treatment Control Relative risk Weight Relative risk pentoxifylline with placebo with n/N n/N (95% CI) (%) (95% CI) compression as standard therapy, all reported the outcome of complete 8 7/18 3/12 34·0 1·56 (0·50–4·86) Arenas healing, or this outcome could be Herdy11 0/6 0/6 0·0 Not estimable determined from individual patient Weitgasser13 20/30 7/30 66·0 2·86 (1·42–5·73) data. In a fixed-effect model, patients in the treatment groups of these trials Total (95% CI) 27/54 10/48 100·0 2·42 (1·34–4·35) combined were 30% more likely to heal Test for heterogeneity p=0·37 than those on placebo (figure 1). Heterogeneity was not significant (figure 1). This result was robust to 0·1 0·2 1·0 5·0 10·0 sensitivity analysis with the randomFavours Favours effects model, masking, dose, route of control treatment drug administration, and study setting. The results did not change by much Figure 2: Relative risk of healing or substantial improvement associated with when the two trials with unclear pentoxifylline compared with placebo in patients with venous leg ulcers


THE LANCET • Vol 359 • May 4, 2002 •

For personal use. Only reproduce with permission from The Lancet publishing Group.


could be that white cells aggregate and release proteolytic enzymes that damage tissue.14 Pentoxifylline, which has an inhibitory effect on expression of cellular adhesion molecules,15 could interfere in these processes. However, the inhibitory effect has not been investigated in venous ulceration. The inhibitory effect could be a consequence of the immunosuppressive activity of pentoxifylline on tumour necrosis factor- (TNF-).16 Expression of cellular adhesion molecules by endothelial cells is upregulated in early venous disease, remains upregulated as the disease progresses, and is accompanied by increased extravasation of leucocytes.17 Expression of these molecules is regulated by proinflammatory mediators such as TNF-.18 Pentoxifylline might also interfere with the production and action of oxygen metabolites.19,20 Although these are physiological studies, a double-blind randomised controlled trial comparing two scavengers of free radicals—allopurinol or dimethyl sulphoxide—with placebo (with compression as a standard treatment) suggested that free radicals derived from oxygen are implicated in the mechanism of ulceration, and that management of these free radicals significantly increases healing of venous ulcers at 12 weeks.21 In our study, heterogeneity could have arisen from differences in study populations, design, and treatment; the diagnostic criteria for inclusion of participants ranged from extensive non-invasive testing to simple clinical examination. The small number of studies meant that the effect of the different methods could not be explored. Venous disease accounts for about 70% of all ulcers.1 Only one participant in the five trials was reported as being misdiagnosed, with a pemphigoid.5 If more ulcers had been misdiagnosed, they would probably have been arterial ulcers, this being the next most common cause after venous disease. Pentoxifylline has been used in treatment of claudication, although its efficacy has not been broadly established.22 Thus, pentoxifylline might have an effect on arterial ulcer healing. However, in the studies in which compression was used, no participants were reported as withdrawing for ischaemia related to compression, a complication that is associated with undiagnosed arterial insufficiency.23 For this reason, we were satisfied that the participants in the included studies had venous ulcers. Differences in study design can contribute to heterogeneity, and inclusion of controlled trials that used poor or unclear methods of allocation concealment can overestimate treatment effects by 30–41%.24 The trials describing adequate allocation concealment were all in the compression subgroup. Exclusion of trials with unclear concealment did not affect the results by much. The trials in the pentoxifylline-only subgroup all had unclear allocation concealment, and in one,13 investigators used a subjective outcome assessment. In that study, doubleblinding was used, making systematic bias unlikely. Little investigation has been done into patients who cannot tolerate compression bandaging, and a large well-designed trial investigating the effect of pentoxifylline on ulcer healing in this group is needed. The strong association between pentoxifylline and healing in two of the trials5,9 in which compression and pentoxifylline were compared with compression and placebo could have been because the ulcers of selected patients were unusually slow to heal. In one trial,9 patients had to have been receiving compression for at least 2 months with no signs of healing, and compression before the trial was not standardised and varied between centres (personal communication, MP Colgan). In the second trial,9 patients must have had a leg ulcer for

THE LANCET • Vol 359 • May 4, 2002 •

2 years before trial entry, but previous treatment was not specified, and attempts to contact the investigator have been unsuccessful. In two other trials,6,10 participants might also have received compression for at least 2 months before trial entry. Thus, we could not conclude whether Colgan and colleagues5 or Barbarino9 did select patients from a slow-to-heal population, or whether they merely selected from populations with longstanding poorly treated ulceration. Patients who receive adequate compression, but are slow to heal are also an underinvestigated group, and would benefit from a large well designed trial investigating the effect of pentoxifylline on ulcer healing. The symmetrical funnel plot suggests that publication bias in this review is unlikely. This bias, however, cannot be ruled out on the basis of the shape of the funnel plots alone. The small number of studies analysed suggests that our results should be interpreted with caution.25 The differences between the trials raise the issue of how generalisable the study results are to clinical practice. The methods by which venous disease was diagnosed varied, indicating the absence of a widely available diagnostic test and an internationally accepted definition of venous ulceration. Furthermore, different systems of compression were used between and within trials. In the factorial trial6 two different systems of compression were evenly balanced between the two groups. In the remaining trials, various systems were used (Unna boot, short-stretch, two-layer, and four-layer elastic bandages). However, such variation is similar to that in clinical practice, with compression systems adjusted to patients’ needs and preferences. A systematic review of compression for venous ulceration reported no clear differences in the effectiveness of the different types of high-compression systems.2 The direction of effect in four of the five compression trials favoured compression, suggesting that the effect was independent of the type of compression, and therefore might be broadly generalisable. Furthermore, the variation in bandaging systems used probably leads to underestimation rather than overestimation of the treatment effect, and variation in diagnostic criteria will lead to an overestimation of effect only if patients who are misdiagnosed are more likely to benefit from the treatment than those diagnosed correctly. As we have discussed, we do not believe that many patients were misdiagnosed. Response in time-to-healing could have been associated with the dose of pentoxifylline used. Falanga and colleagues19 showed that, although proportions of healed patients were similar with a 1200-mg and 2400-mg daily dose, median time-to-healing favoured the higher dose (71 days vs 83 days). The number of adverse events was also higher in the 2400-mg group. Investigators of one study26 have suggested that adjuvant pentoxifylline is more cost effective than compression alone (US$2190 vs $2570); however, the full details of this study have not been published, and we could not assess how this conclusion was reached. Recommendation of pentoxifylline for every patient does not therefore seem appropriate. Prediction rules3 might serve to screen for patients whose ulcers will probably remain unhealed after 24 weeks of compression treatment. In modelling and validation data sets, ulcers that were bigger than 5 cm2 and those that had persisted for longer than 6 months were highly predictive of failure to heal. Such patients might benefit from a trial of pentoxifylline, as might those unable to tolerate compression treatment.


For personal use. Only reproduce with permission from The Lancet publishing Group.




All authors contributed to the writing of the report. B Arroll and A Jull selected the trials for inclusion in the systematic reveiw. A Jull extracted data from the reports. J Waters checked the extracted data for acccuracy.


Conflict of interest statement


None declared.

Acknowledgments This review was first published in the Cochrane Library and full details of the review are available on CLIB. The Centre for Evidence Based Nursing Aotearoa provided financial assistance for translation of papers. We thank Fujian Song for his assistance with the analysis of publication bias; Sally Bell-Syer (Cochrane Collaboration Wound Review Group), and Elizabeth Royle and Heather Maxwell (Cochrane Collaboration Peripheral Vascular Diseases Review Group) for searching their databases; Ian Murphy (Aventis Pharma), for information from the manufacturer’s databases; Rick and Monika Wiechula, Sarah Hetrick, Jorg Neumann, Martin O’Flaherty, Felice Mussico, and Valery Feigin for assistance with translations; Mary Paula Colgan, Reinhild Eberhardt, and Vincent Falanga for providing clarification regarding their trials; David Margolis, Ann-Marie Bagnall, Charles Ruckley, June Jones, Linda O’Flynn, Helen Handoll, and Rajan Madhok for their advice on the original review; the Clinical Trials Research Unit, University of Auckland for their advice on this paper; Alex Fraser for assistance with artwork; and Nicky Cullum and Andrea Nelson (Cochrane Wounds Group), for their support and advice.








References 1


3 4



7 8

Nelzen O, Bergqvist D, Lindhagen A. Venous and non-venous leg ulcers: clinical history and appearance in a population study. Br J Surg 1994; 81: 182–87. Cullum N, Fletcher AW, Nelson EA, Sheldon TA. Compression bandages and stockings in the treatment of venous ulcers (Cochrane Review). In: The Cochrane Library, Oxford: Update Software, 2000. Margolis DJ, Berlin JA, Strom BL. Which venous leg ulcers will heal with limb compression bandages? Am J Med 2000; 109: 15–19. Ward A, Clissold SP. Pentoxifylline: a review of its pharmacodynamic and pharmacokinetic properties, and its therapeutic efficacy. Drugs 1987; 34: 50–97. Colgan MP, Dormandy JA, Jones PW, Schraibman IG, Shanik DG, Young RAL. Oxpentifylline treatment of venous ulcers of the leg. BMJ 1990; 300: 972–75. Dale J, Ruckley C, Harper D, Gibson B, Nelson E, Prescott R. Randomised, double blind placebo controlled trial of pentoxifylline in the treatment of venous leg ulcers. BMJ 1999; 319: 875–78. Angle N, Bergan JJ. Chronic venous ulcer. BMJ 1997; 314: 1019–23. Arenas R, Atoche C. Postthrombotic leg ulcers: safety and efficacy of treatment with pentoxifylline (double-blind study in 30 patients). Dermatolgia Rev Mexicana Segunda Epoca 1988; 32: 34–38.



20 21




25 26

Barbarino C. Pentoxifylline in the treatment of venous leg ulcers. Curr Med Res Opin 1992; 12: 547–51. Falanga V, Fujitani RM, Diaz C, et al. Systemic treatment of venous leg ulcers with high doses of pentoxifylline; efficacy in a randomized, placebo-controlled trial. Wound Repair Regen 1999; 7: 208–13. Herdy CDC, Thomaz JB, Souza SR, Robadey RA, Rodolfo R. Acao da pentoxifilina na cicatrizacao das ulceras de estase: Um estudo clinico-comparativo. Arq Bras Med 1997; 71: 157–61. Schürmann W, Eberhardt R. Wirksamkeit von Pentoxifyllin als zusatz zu Kompressions- und Lokaltherapie bei Patienten mit Ulcus cruris varicosum/postthromboticum. Therapiewoche 1986; 36: 2343–45. Weitgasser H. The use of pentoxifylline (Trental 400) in the treatment of leg ulcers: results of a double-blind trial. Pharmatherapeutica 1983; 3: 143–51. Coleridge Smith PD. Pathogenesis of chronic venous insufficiency and possible effects of compression and pentoxifylline. Yale J Biol Med 1993; 66: 47–59. Krakauer T. Pentoxifylline inhibits ICAM-1 expression and chemokine production induced proinflammatory cytokines in human pulmonary epithelial cells. Immunopharmacology 2000; 46: 253–61. Neuner P, Klosner G, Pourmojib M, Knobler R, Schwarz T. Pentoxifylline in vivo and in vitro down-regulates the expression of the intercellular adhesion molecule-1 in monocytes. Immunology 1997; 90: 435–39. Peschen M, Lahaye T, Hennig B, Weyl A, Simon JC, Vanscheidt W. Expression of adhesion molecules ICAM-1, VCAM-1, LFA-1 and VLA-4 in the skin is modulated in progressing stages of chronic venous insufficiency. Acta Derm Venereol 1999; 79: 27–32. Bevilacqua MP. Endothelial-leukocyte adhesion molecules. Annu Rev Immunol 1993; 11: 767–804. Franzini E, Sellak H, Marquetty C, Babin-Chevaye C, Hakim J, Pasquier C. Inhibition of human neutrophil binding to hydrogen peroxide-treated endothelial cells by cAMP and hydroxyl radical scavengers. Free Radic Biol Med 1996; 21: 15–23. Crouch SP, Fletcher J. Effect of ingested pentoxifylline on neutrophil superoxide anion production. Infect Immun 1992; 60: 4504–09. Salim AS. The role of oxygen-derived free radicals in the management of venous (varicose) ulceration: a new approach. World J Surg 1991; 15: 264–69. Hood SC, Moher D, Barber GG. Management of intermittent claudication with pentoxifylline: meta-analysis of randomized controlled trials. CMAJ 1996; 155: 1053–59. Callam MJ, Ruckley CV, Dale JJ, Harper DR. Hazards of compression treatment of the leg: an estimate from Scottish surgeons. BMJ 1987; 295: 1382. Schulz KF, Chalmers I, Hays RJ, Altman DG. Empirical evidence of bias: dimensions of methodological quality associated with estimates of treatment effects in controlled trials. JAMA 1995; 273: 408–12. Song F, Eastwood AJ, Gilbody S, Duley L, Sutton AJ. Publication and related biases. Health Technol Assess 2000; 4: 1–115. Bosanquet N, Franks PJ, Brown D, Straub J, Harper DR, Ruckley CV. Cost effectiveness of oxpentifylline in venous ulcer healing. Phlebology 1995; suppl 1: 36.

THE LANCET • Vol 359 • May 4, 2002 •

For personal use. Only reproduce with permission from The Lancet publishing Group.

Suggest Documents