Treatment of Intermittent Claudication With Physical Training, Smoking Cessation, Pentoxifylline, or Nafronyl:  A Meta-analysis FREE

INTERMITTENT claudication is most frequently caused by atherosclerotic narrowing of the iliac and femoral arteries, often in combination with similar lesions in at least 1 of the more distal arteries of the leg. It is the symptomatic expression of inability of the blood flow to supply the oxygen required by leg musculature during exercise.

Intermittent claudication (stage II of disease according to Fontaine et al¹) and the underlying peripheral arterial obstructive disease is a common disorder that implies relevant treatment costs in Western societies.^2- 3 The prevalence is estimated to be around 5% in men older than 50 years,^4- 5 and the annual incidence increases from approximately 0.12% to 0.19% during middle age to 0.94% after the age of 65 years.⁶ Follow-up studies have shown that symptoms improve spontaneously in 40% of patients and progress in only 10% to 20%,^7- 8 leading to amputation in 7% within 5 years.⁷ Moreover, intermittent claudication is often associated with generalized atherosclerosis, which is responsible for a 2-fold increased death rate in these patients,^5,7 largely from myocardial infarction and stroke, as compared with individuals without intermittent claudication.

Many treatment strategies (nonpharmacological, pharmacological, and revascularization procedures) have been proposed, aimed at eliminating the local atherosclerotic obstruction, slowing progression of the atherosclerotic disease, or simply ameliorating symptoms. The relative value of these treatment modalities is controversial. Conservative treatment of intermittent claudication includes physical training, smoking cessation, and drug therapy.

Physical training and smoking cessation are claimed to be the most effective conservative therapies for intermittent claudication, with the aim of increasing pain-free walking distance.⁹ A recent meta-analysis of mainly nonrandomized trials¹⁰ concluded that physical training should be applied as standard medical care but based this conclusion on pretraining and posttraining results, without a proper comparison group to account for the natural course of the disease. The mechanism that causes improvement with physical training is unclear. Potential explanations include improvement of collateral flow, redistribution of blood flow to ischemic areas, improvement of utilization and extraction of delivered oxygen from erythrocytes, positive influence on the blood viscosity, and raising of pain threshold.^11- 12

Smoking is a well-recognized risk factor for the development^13- 14 and progression^15- 17 of lower-limb atherosclerosis. Discontinuation of smoking is claimed to reduce the risk of amputation.¹⁸ Although the pathophysiological mechanism is still unclear, the harmful consequences of smoking could be explained by its adverse effects on endothelium, platelets, and coagulation. Therefore, encouraging patients to quit smoking appears to be a logical step in the treatment of patients with intermittent claudication, although compliance with the advice to quit smoking permanently is generally low.

The vasoactive drugs pentoxifylline and nafronyl oxalate (naftidrofuryl) are commonly used in the pharmacological treatment of intermittent claudication in the Western world¹⁹; moreover, pentoxifylline is the only drug approved by the Food and Drug Administration for this indication. The rationale for the use of these agents is based on their ability to enhance red blood cell flexibility and to decrease blood viscosity (pentoxifylline), and to enhance aerobic glycolysis and oxygen consumption in ischemic tissues (nafronyl), although their exact mechanism of action is not well defined. However, the role of these vasoactive agents in the treatment of patients with arterial obstructive disease at stage II according to Fontaine et al¹ is controversial.^19- 22

Because of the uncertainty surrounding the effectiveness of these treatment strategies, we performed a meta-analysis to evaluate the evidence from controlled clinical trials.

We performed a MEDLINE computer-assisted search of the English-language medical literature (January 1976 to December 1996). A broad search frame was set with the use of the key words atherosclerosis, arteriosclerosis obliterans, peripheral vascular disease, and intermittent claudication. In addition, reference lists of retrieved articles and reviews on the subject were evaluated to identify additional and earlier articles. Studies were eligible for inclusion if they evaluated primary treatment of patients with intermittent claudication at stage II of disease¹ with physical training, smoking cessation, pentoxifylline, or nafronyl, independent of their design. Studies conducted on selected populations (diabetic patients, hypertensive patients), reviews or meta-analysis of the literature, and case reports were not eligible. The quality of this selection process was evaluated on a random sample of 100 articles analyzed by 3 independent operators (B.G., E.B., and M.H.P.), achieving a κ value ranging from 0.90 to 0.95. Studies were excluded if they did not have a control group or compared 1 of the above-mentioned 4 strategies with another active treatment, were double reports or retrospective, or did not define or assess adequately any of the following outcomes: pain-free and total walking distance or time, ankle-brachial index before or after exercise, rest and peak blood flow, and ankle pressure. For the measurements of pain-free and total walking distance or time, the criterion was the use of a device that forced the patients to walk at a set speed.

Included trials were divided in 4 groups according to treatment type and graded by 2 independent observers (B.G. and E.B.) for the quality of their design in the following categories: level 1, randomized clinical trials, either double-blind or with specified blind assessment; level 2, other randomized trials; and level 3, nonrandomized controlled studies.

Level 3 studies were considered only if no data from level 1 studies were available. Data from included studies were extracted by 2 independent observers (B.G. and E.B.) by means of a standardized form. In case of disagreement, consensus was reached by adding a third observer (M.H.P.). Outcome data were then summarized in tabular format and, whenever possible and justifiable, combined by appropriate meta-analytical statistical procedures.²³ The only criterion used for inclusion in the final summary measure of effectiveness was that the report enabled direct extraction of a difference in effect between the treatment groups and its common SD. Only results obtained at the end of the treatment period were compared, provided that baseline values for the outcomes considered were comparable among studies. Tests for homogeneity were performed, and no heterogeneity was detected. Results were expressed as common differences of the means with 95% confidence interval (CI).

In total, 49 publications evaluating the efficacy of physical training were identified. Of these, 33 were without a control group,^16,24- 55 1 compared physical training with another active intervention,⁵⁶ 3 were double reports,^57- 59 and 1 was a retrospective study.⁶⁰ None of the publications specified blind outcome assessment; hence, level 1 studies were not available. Therefore, 7 level 2 studies^61- 67 and 4 level 3 studies^68- 71 were analyzed. Among level 2 studies, 2 reported data on the same group of patients^61,65; thus, 6 level 2 trials were available for analysis. Details of the study design are presented in Table 1.

Sample size ranged from 14 to 50 patients in the 6 included trials. Patients randomized to active treatment underwent supervised exercise therapy (Table 1), except in 1 study,⁶⁶ in which patients were advised to exercise at home without supervision. In 2 studies the effect of physical treatment was evaluated in patients undergoing a surgical intervention⁶² or receiving aspirin and dipyridamole.⁶³ In 2 studies patients allocated to the control group received placebo tablets,^64,66 while in the other 2 they were advised to continue with their usual lifestyle.^65,67 Treatment duration ranged from 3 to 6 months, while follow-up varied from 3 to 12 months.

Pain-free walking was determined in 4 studies^{62- 63,65- 66} and total walking in 5^{62- 63,65- 67} of the 6 included trials (Table 2 and Figure 1). Although different experimental conditions (treadmill speed and slope), different units of measurement, and different concurrent treatments hampered summarizing calculations, pooled results of all trials yielded a statistically significant increase in pain-free and total walking distance at the end of treatment in favor of physical training compared with the control group (common difference of the means, 139.0 m [95% CI, 31.0 to 246.9 m] and 179.1 m [95% CI, 60.2 to 298.1 m], respectively). One study reported results on quality of life assessed by means of standard questionnaires; however, it was not possible to extract differences in effect between trained and control patients at the end of the treatment period.⁶¹

There were no statistically significant effects on ankle-brachial index (at rest^62- 63,67 or after exercise^63,67) or on calf blood flow (at rest^63- 64 or after exercise^{62- 64,66- 67}).

In the 4 level 3 studies, sample size ranged from 13 to 42 patients.^68- 71 Patients in the active treatment groups were given a supervised physical training program. Control patients received placebo tablets in 2 trials^68- 69 and were advised to continue with their usual lifestyle in 1 trial.⁷⁰ In the other trial the treatment of control group patients was not specified.⁷¹ Treatment duration ranged from 2 to 6 months, while follow-up varied from 2 to 12 months. In general, the results of the level 3 trials were compatible with those of the level 2 trials (data not shown).

In total, 7 publications dealing with the effect of smoking cessation on intermittent claudication were identified. Three were excluded; of these, 1¹⁶ reported data on a subset of patients included in a subsequently published and larger trial,⁷² 1 did not report clinically meaningful outcomes,⁷⁴ and 1 was a retrospective study.⁶⁰ Therefore, 4 level 3 studies were analyzed.^{13,15,72- 73} Details of their design are presented in Table 1. Since none of the included studies reported similar outcomes, summary calculations were not feasible.

Sample size ranged from 52 to 415 patients. All included patients were smokers on referral^{13,15,72- 73} or had stopped smoking within the past 6 months.⁷² Enrolled patients were strongly advised to discontinue smoking at recruitment; in total, 183 (24.2%) of 756 patients stopped smoking during the 10 months to 10 years of follow-up. In the first study, 41 current smokers were compared with 15 ex-smokers.¹⁵ The results were all not statistically significantly in favor of nonsmokers, yielding a difference in total walking distance of 46.7 m (95% CI, −19.3 to 112.7 m), a difference in resting ankle-brachial index of 0.04 (95% CI, −0.08 to 0.16), and differences in resting and postexercise ankle pressure of 11.7 mm Hg (95% CI, −4.5 to 27.9 mm Hg) and 3.7 mm Hg (95% CI, −28.7 to 36.1 mm Hg), respectively. Another study reported that, after 7 years of follow-up, 26 (8.6%) of 304 current smokers shifted from Fontaine stage II to III, vs none of the 39 patients who successfully quit smoking, for a difference of 8.6% (95% CI, 5.4% to 11.7%).⁷² This finding was not confirmed in a larger study⁷³ in which no different incidence of critical limb ischemia and no deterioration in ankle-brachial index were observed between smokers and patients who quit smoking. However, in the same study, a larger number of failed revascularization procedures was observed among heavy smokers (P=.07). The fourth study did not provide raw data.¹³

Of 28 eligible trials, 15 were excluded: 4^75- 78 reported data from a set of patients included in another report published later,⁷⁹ 3 did not have a control group,^80- 82 and 8 compared pentoxifylline with another active treatment.^83- 90 Therefore, 13 studies in which pentoxifylline was compared with placebo were analyzed; 12 were level 1 trials^79,91- 101 and 1¹⁰² was a level 2 trial. In 1 study⁹¹ with a factorial design, 3 groups of patients were randomized to pentoxifylline, acenocoumarol, or placebo; these data were also considered for the placebo-controlled analysis. The details of the design of the included studies are summarized in Table 3.

Sample size ranged from 24 to 154 patients; treatment duration ranged from 1 to 12 months, and a placebo run-in phase varying from 1 to 12 weeks preceded active treatment in 7 studies. Pentoxifylline was mostly used in tablets, at a dosage ranging from 400 to 1200 mg/d. In 2 trials, the active treatment was administered intravenously, in the first¹⁰¹ at a mean dose of 600 mg/d during 2 weeks and in the second¹⁰² at a mean dose of 400 mg/d during 4 weeks. One report included patients belonging to Fontaine stages II to IV.¹⁰⁰ However, stage II patients were allocated to an independent group, and the results were reported separately. Therefore, this part of the study was analyzed.

Pain-free walking distance was evaluated in 9 level 1 trials and in 1 level 2 trial. Three level 1 trials did not provide data in an appropriate format to be included in the final summary table.^91,97,100 The pooled results of the other 6 studies are illustrated in Table 4 and Figure 1, showing a statistically significant effect of pentoxifylline on pain-free walking distance (common difference of the means, 21.0 m [95% CI, 0.7 to 41.3 m]), as compared with placebo.^{79,92- 95,101}

Total walking distance was evaluated in 8 level 1 and in 1 level 2 trials. One level 1 trial⁹⁷ and the level 2 trial¹⁰² did not provide data in an appropriate format to be summarized. Table 4 depicts the pooled results of 7 level 1 studies yielding a statistically significantly beneficial effect of pentoxifylline over placebo (common difference of the means, 43.8 m [95% CI, 14.1 to 73.6 m]).^{79,92- 93,96,98- 99,101}

There were no statistically significant effects on ankle-brachial index at rest^91,94,97,101 or after exercise^91,94,97(Table 4). Resting calf blood flow was evaluated in 1 level 1 trial⁹⁷ and a slight but statistically significant reduction of blood flow was measured in patients treated with pentoxifylline as compared with placebo (−0.4 mL/min per 100 mL [95% CI, −1.2 to 0.37 mL/min per 100 mL). Calf blood pressure was reported by 1 study, and no significant differences were observed among treated and control patients.⁹⁸

Of 11 eligible trials, 5 were excluded: 2 did not report any of the outcomes of interest^103- 104 and 3 compared nafronyl with another active treatment.^26,37,105 Therefore, 6 trials were analyzed. All of these were level 1 studies (Table 3).^106- 111 Sample size ranged from 40 to 180 patients. The study duration ranged from 3 to 6 months, and a placebo run-in phase varying from 2 to 4 weeks preceded the active treatment administration in 5 trials. Nafronyl was only administered orally, with a dosage ranging from 400 to 800 mg/d.

Pain-free walking distance was evaluated in 6 trials. The pooled results of 4 studies are illustrated in Table 5 and Figure 1, showing a statistically significant effect of nafronyl (common difference of the means, 58.6 m [95% CI, 30.4 to 86.8 m]), as compared with placebo.^{107- 109,111} Of the other 2 trials, the data could not be summarized.^106,110

Total walking distance was evaluated in 4 trials. The results of 2 of these trials could be summarized (Table 5), yielding a statistically significant advantage of nafronyl over placebo (common difference of the means, 71.2 m [95% CI, 13.3 to 129.0 m]).^108- 109 The results of the other 2 trials were not in an appropriate format to be summarized.^106,110

There were no statistically significant effects on ankle-brachial index at rest (Table 5).^{107,109- 110}

Intermittent claudication is a disease affecting approximately 5% of elderly men in Western societies, and debate is ongoing about the efficacy of the various generally recommended treatment strategies. Physical exercise and smoking cessation are usually regarded as the first step in the treatment of this condition. The results of our meta-analysis of randomized controlled trials indicate that the initiation of physical exercise in patients at stage II of disease¹ is associated with an increase in pain-free walking distance of approximately 140 m at the end of treatment and a similar increase (180 m) in total walking distance. However, these results were obtained in only 94 and 112 patients, respectively, and the effect is likely to be overestimated. There appears to be a large potential for factors such as group incomparability and publication bias to have influenced these results.

The schedule of prescribed exercise and associated results varied among individual studies. In 1 randomized study with a small patient population, a nonsupervised exercise program determined a significant effect on pain-free walking distance as well as on total walking distance after 6 months,⁶⁶ while in another randomized study that included more patients and with a supervised exercise program, the effect was nonsignificant (Table 2).⁶² Recently, in 2 small trials, patients with intermittent claudication were randomized to supervised vs home-based physical training; preliminary results seem to show an advantage in favor of patients treated in a supervised setting.^112- 113 However, no definite comparison is yet available, and prescribing expensive supervised exercise programs to these patients remains controversial.

Data supporting the usefulness of smoking cessation are derived only from cohort studies. They do suggest a potential benefit, but since it is conceivable that successful smoking cessation is associated with other lifestyle changes, the favorable result could result from these confounding factors. Formal meta-analysis was not possible.

The effectiveness of pentoxifylline and nafronyl with regard to an increase in pain-free and total walking distance is well documented. By means of sensitivity analysis, it can be demonstrated that a study of 40 patients in the case of pentoxifylline and more than 1200 patients in the case of nafronyl with a true zero effect (assuming comparable SDs) would be needed to make the results not statistically significant. Based on our results, the average increase in pain-free walking distance is limited to 20 to 60 m. Moreover, studies that assess the effect of these drugs on degree of disability and quality of life experienced by this group of patients are not available. Hence, although the effectiveness of these 2 agents is convincingly demonstrated, especially for nafronyl, it is questionable whether their prescription to patients with intermittent claudication at stage II of disease would be also clinically relevant. In current clinical practice, the administration of these drugs should therefore be evaluated on a patient-by-patient basis. One earlier and preliminary publication showed similar results and conclusions.¹¹⁴ Other authors obtained similar results with regard to the effect pentoxifylline on walking distance.¹¹⁵

We believe our results are generalizable to patients with intermittent claudication who have stage II disease. Although we limited the data search to English-language literature only, the potential for error because of a language bias is extremely low and, if present, is trivial ("Tower of Babel error").¹¹⁶

We conclude that physical training is a potentially effective treatment modality, as is smoking cessation. However, available evidence derived from intervention studies in favor of these nonpharmacological treatment strategies is weak (small samples in level 2 and level 3 studies, respectively). Thus, it is unclear whether prescribing supervised physical training programs to patients with obstructive arterial disease is cost-effective, since this intervention modality is expensive and time-consuming. Data from larger properly designed clinical trials, assessing also the degree of disability and quality of life,^61,112- 113 are strongly warranted. For now, advising patients to walk and stop smoking is likely to be all that is required. Pentoxifylline and nafronyl are widely used in Western societies, and the results of our analysis confirm their effectiveness in improving walking distance; however, the clinical relevance of such a small effect should be elucidated by properly designed trials that also assess also degree of disability and quality of life.

Accepted for publication May 26, 1998.

Corresponding author: Bruno Girolami, MD, Istituto di Semeiotica Medica, Universitá di Padova, Via Ospedale 105, 35100 Padova, Italy (e-mail: girolami@ux1.unipd.it).