Sage Journals: Discover world-class research

Abstract

Propionyl-l-carnitine (PLC) may improve exercise performance in patients with peripheral artery disease, but results from clinical trials have been inconsistent. The safety and efficacy of PLC for treatment of claudication was evaluated by a systematic review and meta-analysis of clinical trials for which data were available through September 2010. Eighty-five studies were identified, of which 13 were randomized controlled trials. Owing to database availability for the six phase III studies carried out with PLC (1 g orally, twice daily), a patient-level meta-analysis was conducted as the primary analysis. Treadmill performance data from these six studies were harmonized to peak walking distance (PWD) on a 7% grade at a speed of 3 km/hour. PLC (n = 440) was associated with a net 16 meter improvement (95% CI, 8–20 meters) in PWD as compared with placebo (n = 427) in the primary analysis (p = 0.002). The effect of PLC was similar in subpopulations defined using clinical and demographic variables, with possible enhanced benefit in patients engaged in an exercise program or enrolled at study sites in Russia. The systematic review of the effect of PLCs on claudication identified seven additional randomized controlled trials for a total of 13 trials, which included 681 patients on placebo and 672 on PLC. This meta-analysis confirmed a 45 meter net improvement on PLC using a random-effects model. In conclusion, oral PLC is associated with a statistically significant increase in PWD in patients with claudication, which may be clinically relevant.

Keywords

cardiovascular diseases claudication (see intermittent claudication)clinical trials exercise test peripheral arterial disease peripheral vascular diseases walking

Introduction

Therapeutic options for peripheral artery disease-associated claudication remain limited despite increasing recognition of the disability associated with the condition.^1,2 Supervised exercise rehabilitation is efficacious and considered the medical treatment of choice,^2,3 but is available to only a few patients due to cost, a limited number of available facilities, inconvenience and lack of acceptance. Revascularization is beneficial in many patients^4,5 but is expensive and its long-term efficacy is unclear. Pharmacotherapies in the United States are limited to pentoxyfylline, which is associated with only mild improvement in patients⁶ and cilostazol, whose use has been limited due to vasodilatory adverse effects and residual safety concerns based on its mechanism of action.^7,8

Evidence that a secondary myopathy contributes to the exercise limitation in patients with claudication^9,10 has suggested novel targets for new drugs to treat the exercise impairment. Propionyl-l-carnitine (PLC) has many attributes which make it an attractive drug candidate for treating claudication. PLC is taken up by muscle, and generates intracellular carnitine and propionyl-CoA. Carnitine is critical for the normal oxidation of fatty acids.¹¹ Carnitine can also modulate pyruvate dehydrogenase activity, and thus glucose oxidation, under some metabolic conditions.¹² Carnitine metabolism is altered in patients with PAD,^13,14 consistent with an underlying metabolic dysfunction. Propionyl-CoA can generate tricarboxylic acid cycle intermediates which may be important for maintaining oxidative flux under ischemic conditions.¹⁵ Additionally, carnitine administration has been associated with decreased markers of systemic inflammation^16,17 and prevention of muscle fatigue,^18,19 which may be relevant to the treatment of claudication.

PLC has been studied as a treatment for claudication based on these properties. Published studies have supported the hypothesis that PLC is effective but have in general been small and may raise questions of publication bias with respect to the available information. Thus, a comprehensive systematic review of all studies with PLC in patients with claudication performed and for which data were available up to September 2010 was conducted in order to ascertain a better estimate of the effect of PLCs in this population.

Methods

Systematic review of all studies for PLC in claudication

Study identification

A systematic review of all studies of the effect of PLCs on walking performance in patients with claudication was conducted. All studies published in the medical literature in any language, starting from January 1984 and up to September 2010, were retrieved through MEDLINE and EMBASE using the following search strategy: “{[(propionyl AND carnitine) OR propionyl-l-carnitine] WITH [(intermittent AND claudication) OR claudication] AND Leriche-Fontaine stage II}”. The references of all retrieved articles were inspected to identify additional relevant publications; in addition, the Cochrane Library database was also searched.

In order to minimize publication bias, the proceedings of international congresses of vascular medicine were reviewed to identify abstracts of negative studies. Furthermore, the internal database of Sigma-Tau S.p.A (Pomezia, Italy; the manufacturer and principal developer of PLC for peripheral artery disease) was accessed to identify unpublished studies conducted with PLC in patients with claudication.

Patient-level meta-analysis

Study identification – primary analysis

The complete databases of the six phase III randomized, placebo-controlled studies performed by Sigma-Tau S.p.A. in patients with claudication were made available to the authors, including three studies not previously published. A patient-level meta-analysis of these six studies was prospectively designated as the primary analysis. All studies used 1 g PLC twice daily or matching placebo. For convenience, the studies will be referred to as European,²⁰ US Dose Ranging, US/Russia,²¹ European-Diabetes, US-Diabetes, and US-Exercise,²² as detailed in Table 1.

Table 1.

Overview of the phase III trials included in the meta-analysis

		European	US Dose Ranging	US/Russia	European-Diabetes	US- Diabetes	US-Exercise
Primary endpoint – treadmill used		Fixed 7% grade at 3 km/h	Fixed 12% grade at 2 miles/h	Graded^a at 2 miles/h	Fixed 7% grade at 2 miles/h	Fixed 7% grade at 2 miles/h	Graded^a at 2 miles/h
Primary endpoint – variable		MWD at 12 months	MWD at 6 months	PWT at 6 months	MWD at 9 months	MWD at 9 months	PWT at 6 months
Method for defining baseline performance		Mean of highest and lowest of 3 baseline treadmills	Mean of 3 baseline treadmills	Mean of 2 baseline treadmills	Mean of 2 baseline treadmills	Mean of 2 baseline treadmills	Mean of 2 baseline treadmills
Key inclusion criteria	Baseline walking capacity (MWD in meters or PWT in seconds)	50–250 m	50–250 m	60–300 s	50–300 m	50–300 m	90–720 s
	Baseline variability	< 25%	< 25%	< 20%	< 25%	< 25%	< 20%
	ABI	< 0.80	< 0.85	< 0.90	< 0.90	< 0.90	< 0.90
	Claudicants	All	All	All	Type 2 diabetics	Type 2 diabetics	All
Key exclusion criteria		Class 2–4 heart failure	Type 1 diabetes	Critical limb ischemia, recent revascularization	Type 1 diabetes; HbA1c > 10%	Type 1 diabetes; HbA1c > 10%	Critical limb ischemia, exercise program, recent revascularization
Time period of study conduct		1991–1995	1993–1996	1994–1996	1999–2002	2001–2004	2006–2008

All studies compared placebo to 1 g PLC orally twice daily as their primary objective.

MWD, maximum walking distance; PWT, peak walking time.

Graded treadmills were modified Gardner protocols, as detailed in ref. 21.

Analysis populations – primary analysis

In the patient-level meta-analysis, the intention-to-treat (ITT) population pre-specified by each study’s Statistical Analysis Plan for the primary efficacy analysis was used. In the case of the European study, the primary analysis population was composed of patients with baseline walking distances of 50–250 meters. Similarly, in the US Dose Ranging study, patients treated with 1.0 g PLC twice daily formed the basis for that study’s primary efficacy assessment, and have been used here for the primary analysis. When multiple treadmill assessments were made over the course of therapy, the treadmill time point specified in the study’s protocol as the primary endpoint was utilized in the current analysis. Other protocol-specified treadmill assessments were used for the secondary analysis of time-dependence of PLC effect. Furthermore, in order to identify subgroups of patients with a different response to PLC, a broad array of clinical and demographic parameters was used to check for consistency of PLC effect. In this context, concomitant exercise refers to a patient self-reported, regular exercise regimen or a protocol-specified exercise program during the study period. No information on concomitant exercise was available from the US-Diabetes study.

These patient databases were also used for the safety analysis which was based on incidence of adverse events (AEs), serious AEs, deaths and the number of patient discontinuations.

Harmonization of treadmill peak performance assessments

The six phase III studies identified for the primary meta-analysis evaluated maximum walking capacity (either maximum walking distance (MWD) on a constant-load test or peak walking time (PWT) on a graded test) as the primary efficacy parameter. However, these studies used different treadmill methodologies (Table 1), precluding a direct combination of study results. Therefore, harmonization of the results was performed by converting the results of the different treadmill tests into a common measure for analysis (peak walking distance; PWD). This was accomplished by converting each subject’s peak performance on the treadmill test to the equivalent walking distance on a fixed treadmill of 7% slope and a speed of 3 km/hour. This calculation was done by first converting the peak performance to energy expenditure (joules/kg) using a previously developed methodology.^23,24 This energy expenditure was then converted back to the equivalent walking distance at a 7% slope and 3 km/hour speed.

Statistical analysis

Changes in treadmill performance deviated substantially from a normal distribution, and log transformation did not adequately yield an appropriate distribution in most studies. Thus, in accordance with each study’s Statistical Analysis Plan, the meta-analyses of the change in treadmill performance over baseline (PLC vs placebo) were performed using the non-parametric Cochran–Mantel–Haenszel test on the modified ridits, which is equivalent to the van Elteren test.²⁵ Ridit (Relative to an Identified Distribution) analysis is an empirical cumulative probability transformation used on ordinal data which does not rely on an interval scale. Modified ridits are standardized midrank scores. Similarly, non-parametric approaches were used to calculate the 95% confidence intervals (CIs) and p-values.²⁶ Of note, the resulting CIs may be asymmetric due to the non-parametric methods needed for the non-normal distribution. The patient-level data were assessed for heterogeneity based on the chi-squared test. If the data were found to not be heterogeneous (i.e. were in fact homogeneous) then the prespecified decision was to use a fixed-effects model for the meta-analysis. The statistical software was SAS Software v.9.2 for Windows (SAS Institute, Inc., Cary, NC, USA).

Study-level meta-analysis of randomized, controlled trials

Study inclusion and exclusion criteria

Studies identified during the systematic review (see before) were included in the study-level meta-analysis if they: included patients with claudication confirmed by history and PAD confirmed on an ankle–brachial index (ABI) or toe–brachial index (TBI); were randomized, double-blinded and had a comparator arm (active or placebo); had complete data on the change (± standard deviation) from baseline for MWD and/or PWT assessed using a treadmill test; used orally administered PLC; were completed according to their protocol; and randomized patients with Leriche-Fontaine stage II disease (i.e. intermittent claudication). Excluded were studies with: no confirmation of claudication; no comparator arm; incomplete or no treadmill data on MWD or PWT; intravenous administration of PLC; incomplete patient data sets due to study discontinuation; and patients with ischemic rest pain or trophic lesions (Leriche-Fontaine stages III–IV). When duplicative studies were identified, only the most recent publication was considered.

Analysis population

In the retrieved articles, the mean change over baseline observed in the ITT population was used. If this endpoint was not available, the result from the randomized patient population was used.

Statistical analyses

Data from each retrieved study were abstracted using a data collection form, and results compared. Quality of the included studies was not graded. For each identified study, the treatment effect size was assessed by computing the mean difference from end of treatment relative to baseline for each group (PLC and placebo) and the difference between these group means was then normalized to their pooled standard deviation. Heterogeneity across studies was assessed by formal statistical testing, using a Q-test for heterogeneity.²⁷ Trial results were pooled in a statistical meta-analysis using weighted standardized mean differences and the estimator of Hedges and Olkin (Hedges’ g ), as described by Boissel et al.²⁷ The decision to use a fixed-effects and/or a random-effects model was used based on heterogeneity testing that defined the best model to use. Commercial software was used for all the analyses (MetaWin, statistical software for meta-analysis (v.2.1) by Rosenberg, Adams, and Gurevitch; Sinauer Associates Inc., 2007).

Results

Systematic review of all studies for PLC in claudication

Identified studies

A total of 85 studies including 3394 patients with claudication were identified using the above search strategy. Of these, 72 were excluded from subsequent analysis (n = 1540 patients, 45% of the total) because the reports were incomplete or had no treadmill data (n = 23), were reviews (n = 21), used intravenous PLC (n = 27), or oral PLC was administered only as a single dose (n = 1) (Figure 1). Appendix 1 (online only) provides a complete list of the excluded studies.

Figure 1.

Systematic review: disposition of identified studies using PLC in intermittent claudication (January 1984 – September 2010). N refers to the number of studies and patient numbers are the total number of randomized subjects.

The remaining 13 studies (n = 1854 patients) were included in the subsequent analyses (Figure 1). All 13 studies were randomized controlled trials, exposing 1002 randomized patients with Leriche-Fontaine stage II claudication to a median dose of 2000 mg/day oral PLC (range: 1000–4000 mg/day), mostly in a twice daily schedule, for a median treatment duration of 6 months (range: 6–12). Six out of the 13 studies were phase III trials and contributed substantially to this cohort (n = 1402 randomized patients; Figure 1). An overview of the designs of the six phase III trials is provided in Table 1. The designs of the additional seven studies are summarized in Table 2.

Table 2.

Clinical studies meeting criteria for inclusion in the systematic review

Study (first author or principal investigator)	Total no. of patients	PLC daily dose (g)	Change in MWD(PLC – Placebo)^a	p-value
Greco et al.⁴⁰	20	1.5	146	0.001
Brevetti et al.⁴¹	139	1.0–3.0	50	0.09
Di Somma et al.⁴¹	106	1.0–3.0	48	0.001
Strano⁴²	118	1.5	52	0.087
Bagliani (ST261-124-87)^b	22	1.5	152	0.054
Galante (ST261-123-87)^b	18	1.0	−22	NS
Poggi (ST261-116-90)^b	29	2.0	27	NS

The studies by Brevetti and Di Somma were published in a single manuscript. The individual trial results were made available by the trials’ sponsor. Unpublished studies are identified by the protocol number made available by Sigma-Tau S.p.A.

The change in maximal walking distance (MWD) was calculated as the difference in the group means PLC vs placebo.

Unpublished studies, made available by Sigma-Tau S.p.A. (Pomezia, Italy).

Patient-level meta-analysis

Analysis population

The patient demographics comprising the primary analysis cohorts in the six phase III trials are described in Table 3. Patients were predominately male, were on average in their seventh decade of life, and had a high prevalence of current smoking. Studies differed in their inclusion of diabetics, with two studies (US-Diabetes and European-Diabetes) requiring a diagnosis of diabetes for subjects to be included, and one study (European) tending to under-represent diabetics. Baseline treadmill performance was similar in all studies. The concomitant use of statin and antiplatelet therapies appeared to increase in the more recent versus the earlier studies (Table 3).

Table 3.

Demographics of the ITT populations in the phase III trials included in the primary meta-analysis

	European²⁰		US Dose Ranging		US/Russia		European-Diabetes		US-Diabetes		US-Exercise
	Plac	PLC	Plac	PLC	Plac	PLC	Plac	PLC	Plac	PLC	Plac	PLC
n (ITT population)	83	80	75	80	73	82	85	86	81	80	30	32
Sex, % male	78.3%	83.8%	92.0%	82.5%	87.7%	81.7%	80.0%	69.8%	87.7%	81.3%	83.3%	62.5%
Age, mean years	61.7	61.0	64.7	64.5	60.5	62.0	64.8	65.1	67.6	65.2	66.6	67.4
Smoking status, % current smokers	34.9%	41.3%	42.7%	48.8%	53.4%	51.2%	30.6%	29.1%	29.6%	30.0%	16.7%	37.5%
Baseline ABI, mean	0.56	0.57	0.57	0.59	0.57	0.59	0.59	0.62	0.61	0.64	0.65	0.65
Baseline peak walking distance, mean (m)	173	163	133	138	218	219	154	146	149	159	221	231
Baseline use of statins, % taking at baseline	1.2%	0%	18.7%	11.3%	6.8%	15.9%	15.3%	23.3%	65.4%	63.8%	63.3%	53.1%
Baseline use of antiplatelet drugs	0%	0%	26.7%	21.3%	13.7%	14.6%	17.6%	24.4%	11.1%	10.0%	13.3%	15.6%
Diabetes, % with diabetes	4.8%	7.5%	21.3%	26.3%	16.4%	14.6%	100%	100%	100%	100%	33.3%	9.4%
Concomitant exercise, % self-reported participation	18.1%	17.5%	6.7%	8.8%	12.3%	19.5%	2.4%	2.3%	Not assessed		100%	100%

Baseline peak walking distances are the harmonized treadmill performances at baseline (see Methods).

Plac, placebo; PLC, propionyl-l-carnitine.

Effect of oral PLC on peak walking distance

The ITT populations included a total of 427 patients randomized to placebo and 440 patients randomized to 1 g PLC orally twice daily. Analysis of the ITT populations demonstrated that PLC was associated with a net improvement of 16 meters (95% CI, 8–20 meters) in PWD as compared with placebo (p = 0.002; Figure 2, standardized effect size 0.11). Examination of the six individual studies demonstrated no heterogeneity in effect size (χ²: 9.36, df: 5; p > 0.05), consistent with the overlapping CIs across studies (Figure 2).

Figure 2.

Effect of PLC on harmonized peak walking distance for patient-level meta-analysis. The results from ITT populations in the six phase III studies and the meta-analysis are shown as change over baseline vs placebo (± 95% CIs). The CIs are asymmetric due to the use of non-parametric analyses (see Methods).

The effect of PLC on PWD appeared consistent across a broad array of clinical and demographic parameters, except for suggestions of increased effect in patients who were participating in a concomitant exercise program during the study period, or who were randomized at sites in Russia (Figure 3).

Figure 3.

Effect of PLC on harmonized peak walking distance in patient subgroups (ITT populations) for patient-level meta-analysis. The results are from the six phase III studies and are based on patient characteristics at study entry. The data are expressed as change in peak walking distance over baseline vs placebo ± 95% CIs. The CIs are asymmetric, reflecting the use of non-parametric analyses (see Methods). N refers to the number of subjects in each subgroup and the point estimate is presented numerically. Conc, concomitant; PWD, peak walking distance (harmonized).

The six studies varied in duration from 6 to 12 months (Table 1) and included protocol-defined treadmill assessments at earlier time points on a protocol-specific basis. Estimates of improvement in PWD as a function of time suggested that the benefits of PLC were maintained over a treatment interval of 3–12 months (Figure 4).

Figure 4.

Time dependence of PLC effect on harmonized peak walking distance for patient-level analysis. Protocol-specified treadmill assessments were used. As the time points studied varied by protocol, the number of patients contributing to each time point in the figure also varies and a weighted regression was used. Results are the change over baseline vs placebo ± 95% CIs based on the ITT populations. Numerical values for the point estimate at each time point are provided. The CIs are asymmetric, reflecting the use of non-parametric analyses (see Methods). A trend line is shown for descriptive purposes.

Safety of oral PLC

PLC was generally well tolerated by patients with claudication, with discontinuation rates similar between placebo (n = 626) and PLC groups (n = 797) (Table 4). The incidence of serious AEs, cardiovascular serious AEs and deaths were balanced across the placebo and PLC arms. Across the six studies there were a total of 95 (15.2%) patients with serious AEs in patients randomized to placebo and 133 (16.7%) patients with serious events in the PLC arms. Gastrointestinal adverse events, the most common adverse events associated with PLC,²⁸ were similar between treatment groups (11.8% placebo, 14.3% PLC).

Table 4.

Safety profile of PLC (1 g orally twice daily) in patients with claudication (Safety population)

	European		US Dose Ranging		US/Russia		European-Diabetes		US-Diabetes		US-Exercise
	Plac	PLC	Plac	PLC	Plac	PLC	Plac	PLC	Plac	PLC	Plac	PLC
Safety population	253	248	75	244	76	85	97	94	90	92	35	34
Any adverse events^a	131	123	49	173	32	42	48	52	277	274	24	24
Gastrointestinal adverse events^a	25	25	12	44	8	7	7	6	15	25	7	7
Serious adverse events	35	34	11	51	9	6	16	20	19	19	5	3
Cardiovascular serious adverse events^a	20	21	5	32	5	3	9	12	8	11	3	0
Deaths	5	5	0	4	2	0	2	1	0	3	0	0
Discontinuations^b	53	60	9	34	5	4	18	14	27	24	6	4

Values are numbers of patients with at least one treatment-emergent adverse event.

The Safety population may be larger than the ITT population as it includes other doses or cohorts not designated as the study’s primary efficacy analysis population.

Plac, placebo; PLC, propionyl-l-carnitine.

Based on COSTART or MeDRA coding.

Discontinuations for any cause.

Meta-analyis of all randomized controlled trials

Effect of oral PLC on peak walking performance

An inspection of the funnel plot of the standardized difference in PWD observed in each of the 13 included studies versus the study’s sample size was consistent with an absence of publication bias, as confirmed by a Kendall’s test (tau: –0.29, p = 0.17; Figure 5). Similar results were obtained when the variance in each study effect size was used instead of sample size (data not shown).

Figure 5.

Funnel plot of PLC effect size (standardized difference) vs sample size (total number of randomized patients) in each individual study. Diamonds identify published studies.

The study-level meta-analysis of the 13 randomized controlled trials identified a Q-test for heterogeneity that was not significant at p = 0.08 when using a random-effects model. When the random-effects model was used, the improvement in PWD induced by PLC was 45 meters (95% CI, 7–84 meters, standardized effect size 0.28; Figure 6). In contrast, a fixed-effects model demonstrated the studies were heterogeneous (Q-test for heterogeneity, p < 0.01). With the fixed-effects model there was a statistically significant effect of PLC in patients with claudication, with an improvement in PWD of 35 meters (95% CI, 13–57 meters, standardized effect size 0.22).

Figure 6.

Effect of PLC on maximum walking distance. The results from all included studies and their meta-analysis are shown as change over baseline vs placebo (± 95% CIs, random-effects model).

Discussion

Efficacy of PLC

In PAD patients with claudication, a patient-level meta-analysis of six phase III trials demonstrated that oral PLC (1 g twice daily) increased harmonized treadmill walking distance by 16 meters (95% CI, 8–20 meters), as compared with placebo. This conclusion was confirmed by a systematic review and study-level meta-analysis that included 13 studies overall (including the above six phase III studies), which demonstrated a 45 meter improvement (95% CI, 7–84 meters, random-effects model). These improvements corresponded to standardized effect sizes of 0.11 and 0.28, respectively. The benefit of oral PLC did not appear to be restricted to any subgroup based on demographic or clinical features (Figure 3). Patients were randomized at sites in Russia in two studies (European and US-Russia), and this patient cohort seemed to benefit more from PLC treatment for reasons that may be associated with either the demographics or the standard of care of these patients (Figure 6). However, the specific reasons for this difference are not known, but both studies were conducted to Good Clinical Practice standards and were monitored and audited. Of note, in the US-Russia study, the drug effect was statistically significant even after adjusting for the country effect.

The patient-level meta-analysis was considered primary and the most robust as all individual phase III patient data were included, thus allowing for more accurate estimates. As the total cohort was not truly randomized, but only randomized within study, this approach may have some limitations. Nonetheless, each study was sufficiently large to minimize randomization bias and the effects of PLC were not significantly heterogeneous across trials. The effect estimates from the patient-level meta-analysis differ from those of the study-level meta-analysis because of the different patient groups selected, lack of treadmill data harmonization in the latter, and the specific analysis methods used. Of note, although the study-level meta-analysis included several smaller studies, varied treadmill methods, and PLC administered at different doses, its results overlap with those of the patient-level meta-analysis and thus can be viewed as supportive of its conclusions.

The importance of accessing unpublished data and the potential bias of the published literature are illustrated by the changes in PWD associated with PLC in published versus unpublished studies (Tables 1 and 2). On a sample size-weighted basis, the improvement with PLC in published studies was approximately 44.5 meters versus 8.5 meters in the unpublished trials.

The estimates of the effect of PLCs on maximal walking performance observed in the current analyses are similar to that of other drugs approved for claudication in the United States, including pentoxyphylline (44 meters; 95% CI, 14–74 meters)²⁹ and cilostazol (42 meters; 95% CI, 21–64 meters).³⁰ The effect of all drugs, however, appears smaller than that of non-pharmacological interventions, such as supervised exercise training (113 meters; 95% CI, 95–131 meters).³¹ Importantly, these apparently modest changes in treadmill walking performance on drug therapy have been associated with patient-perceived increases in ambulatory walking ability.³²

The clinical pharmacology of PLC may limit its efficacy as an oral therapy for claudication. The oral systemic bioavailability of PLC is likely not higher than the approximately 10–15% observed with carnitine,³³ and peak PLC plasma concentrations of only 5 µM are achieved.³⁴ Further, PLC may have limited access to intracellular sites of action. The skeletal muscle carnitine pool is kinetically isolated from the plasma pool.³⁵ Uptake of carnitine or acylcarnitines into muscle under normal conditions is slow, and even large doses of intravenous carnitine have minimal acute effects on the muscle carnitine content.³⁵ Thus, it can be hypothesized that better delivery of PLC to skeletal muscle in patients with PAD might improve its efficacy. Supporting this concept are data showing short-term benefit of intravenous carnitine or PLC in patients with PAD.^36,37 In this context it is interesting to note that the full benefit on walking performance of oral PLC appears to require a long duration of treatment (Figure 4), thus suggesting slow accumulation in muscle. Insulin has been shown to enhance muscle uptake of carnitine,³⁸ but it is unclear whether a similar effect on PLC uptake exists. In this context, a trend towards greater efficacy in non-diabetic patients (i.e. patients without insulin resistance) was observed (Figure 3).

Safety of oral PLC

Published reviews of PLC and other carnitine therapeutics conclude that the drugs have a low incidence of AEs and no serious toxicities.^28,39 The data from the six phase III trials support this conclusion in the PAD population. The drug appeared to be well tolerated with discontinuation rates similar between the treatment arms. Gastrointestinal AEs occurred at a low rate, without relevant treatment-related differences observed. The incidence of serious AEs, cardiovascular serious AEs and deaths were balanced between treatment groups. Across the studies, a total of nine (1.4%) deaths were reported in patients randomized to placebo and 13 (1.6%) in patients on PLC.

Limitations of the analysis.

No meta-analysis can likely support conclusions as robust as those obtained from a single, adequately powered, well-conducted clinical trial. While the primary meta-analysis was able to utilize patient-level data, it is important to note that the six phase III studies randomized different patient populations, treatment duration varied by study, and background therapy differed either by study design (e.g. background exercise rehabilitation) or reflecting evolution of standards of care during the 15-year period covered by the six studies (e.g. use of statins). Differences in discontinuation rates across studies (Table 4), while balanced between treatment arms, also suggest potential unrecognized heterogeneity between such studies. Nonetheless, both the overlap of the 95% CIs for the estimated PLC effects (Figure 3) and the result of the chi-squared test for homogeneity suggest that the PLC effect did not vary appreciably across studies. Further, analysis of subpopulations failed to identify dramatic differences in PLC effect, with very few notable exceptions (Figure 3), although female patients were under-represented. It is possible, but unlikely, that additional clinical trial data with more unfavorable results exist but were not included in this systematic review. Finally, while treadmill performance is an accepted surrogate for clinical function based on quality of life and other ambulatory assessments, the current analyses focused exclusively on treadmill performance and did not directly assess clinical impact.

Conclusions

Analyses of patient-level data from six phase III studies, as well as a study-level meta-analysis of clinical trial data from 13 randomized controlled trials, demonstrate that PLC improves walking performance in patents with claudication. The overall standardized effect size was 0.11 in the patient-level analysis of the phase III studies and 0.28 in the publication-level meta-analysis, each of which was statistically significant (p < 0.002) and of a magnitude representing a net benefit of 16–45 meters. PLC was in general well tolerated, with no relevant safety signals identified.

Footnotes

Acknowledgements

The authors thank Sigma-Tau S.p.A. and Joe Testa at Sigma-Tau Research, Inc. for facilitating access to the study databases for analysis.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Conflict of interest

EPB and WRH were consultants to Sigma-Tau S.p.A. DK is an active consultant for Sigma-Tau S.p.A. AA is an employee of Sigma-Tau S.p.A.

References

Hiatt

. Medical treatment of peripheral arterial disease and claudication. N Engl J Med 2001; 344: 1608–1621.

Gardner

Afaq

. Management of lower extremity peripheral arterial disease. J Cardiopulm Rehabil Prev 2008; 28: 349–357.

Falcone

Hirsch

Regensteiner

. Peripheral arterial disease rehabilitation: A review. J Cardiopulm Rehabil 2003; 23: 170–175.

Nylaende

Abdelnoor

Stranden

. The Oslo balloon angioplasty versus conservative treatment study (OBACT)—The 2-years results of a single centre, prospective, randomised study in patients with intermittent claudication. Eur J Vasc Endovasc Surg 2007; 33: 3–12.

Almahameed

Bhatt

. Contemporary management of peripheral arterial disease: III. Endovascular and surgical management. Cleve Clin J Med 2006; 73(Suppl 4): S45–51.

Radack

Wyderski

. Conservative management of intermittent claudication. Ann Intern Med 1990; 113: 135–146.

Sorkin

Markham

. Cilostazol. Drugs Aging 1999; 14: 63–71; discussion 2–3.

Hiatt

Money

Brass

. Long-term safety of cilostazol in patients with peripheral artery disease: The CASTLE study (Cilostazol: A Study in Long-term Effects). J Vasc Surg 2008; 47: 330–336.

Brass

Hiatt

Green

. Skeletal muscle metabolic changes in peripheral arterial disease contribute to exercise intolerance: A point-counterpoint discussion. Vasc Med 2004; 9: 293–301.

10.

Pipinos

Judge

Selsby

. The myopathy of peripheral arterial occlusive disease: Part 2. Oxidative stress, neuropathy, and shift in muscle fiber type. Vasc Endovascular Surg 2008; 42: 101–112.

11.

Bremer

. Carnitine—metabolism and functions. Physiol Rev 1983; 63: 1420–1480.

12.

Stephens

Constantin-Teodosiu

Greenhaff

. New insights concerning the role of carnitine in the regulation of fuel metabolism in skeletal muscle. J Physiol 2007; 581: 431–444.

13.

Hiatt

Nawaz

Brass

. Carnitine metabolism during exercise in patients with peripheral vascular disease. J Appl Physiol 1987; 62: 2383–2387.

14.

Hiatt

Wolfel

Regensteiner

Brass

. Skeletal muscle carnitine metabolism in patients with unilateral peripheral arterial disease. J Appl Physiol 1992; 73: 346–353.

15.

Di Lisa

Menabo

Siliprandi

. L-propionyl-carnitine protection of mitochondria in ischemic rat hearts. Mol Cell Biochem 1989; 88: 169–173.

16.

Malaguarnera

Gargante

Russo

. L-carnitine supplementation to diet: A new tool in treatment of nonalcoholic steatohepatitis—A randomized and controlled clinical trial. Am J Gastroenterol 2010; 105: 1338–1345.

17.

Shakeri

Tabibi

Hedayati

. Effects of l-carnitine supplement on serum inflammatory cytokines, C-reactive protein, lipoprotein (a), and oxidative stress in hemodialysis patients with Lp (a) hyperlipoproteinemia. Hemodial Int 2010; 14: 498–504.

18.

Brass

Scarrow

Ruff

Masterson

Van Lunteren

. Carnitine delays rat skeletal muscle fatigue in vitro. J Appl Physiol 1993; 75: 1595–1600.

19.

Dutta

Ray

Singh

Vats

Singh

. L-carnitine supplementation attenuates intermittent hypoxia-induced oxidative stress and delays muscle fatigue in rats. Exp Physiol 2008; 93: 1139–1146.

20.

Brevetti

Diehm

Lambert

. European multicenter study on propionyl-L-carnitine in intermittent claudication. J Am Coll Cardiol 1999; 34: 1618–1624.

21.

Hiatt

Regensteiner

Creager

. Propionyl-L-carnitine improves exercise performance and functional status in patients with claudication. Am J Med 2001; 110: 616–622.

22.

Hiatt

Creager

Amato

Brass

. Effect of propionyl-L-carnitine on a background of monitored exercise in patients with claudication secondary to peripheral artery disease. J Cardiopulm Rehabil Prev 2011; 31: 125–132.

23.

Margaria

. Sulla fisiologia e specialmente sul consumo energetico della Marcia e della corsa verie velocità ed inclinazioni del terreno. Atti Accad Naz Lincei (Roma) 1938; 7: 299–368.

24.

Marconi

Ferretti

Anchisi

. Energetics of walking in patients with peripheral arterial disease: A proposed functional evaluation protocol. Clin Sci (Lond) 2003; 105: 105–111.

25.

Mantel

. Chi-square test with one degree of freedom: Extension of the Mantel–Haenszel procedure. J Am Stat Assoc 1963; 58: 690–700.

26.

Lehmann

. Non Parametrics: Statistical Methods Based on Ranks. San Francisco: Holden-Day, 1975.

27.

Boissel

Blanchard

Panak

Peyrieux

Sacks

. Considerations for the meta-analysis of randomized clinical trials. Summary of a panel discussion. Control Clin Trials 1989; 10: 254–281.

28.

Goa

Brogden

. l-Carnitine. A preliminary review of its pharmacokinetics, and its therapeutic use in ischaemic cardiac disease and primary and secondary carnitine deficiencies in relationship to its role in fatty acid metabolism. Drugs 1987; 34: 1–24.

29.

Girolami

Bernardi

Prins

. Treatment of intermittent claudication with physical training, smoking cessation, pentoxifylline, or nafronyl: A meta-analysis. Arch Intern Med 1999; 159: 337–345.

30.

Pande

Hiatt

Zhang

Hittel

Creager

McDermott

. A pooled analysis of the durability and predictors of treatment response of cilostazol in patients with intermittent claudication. Vasc Med 2010; 15: 181–188.

31.

Watson

Ellis

Leng

. Exercise for intermittent claudication. Cochrane Database Syst Rev 2008; CD000990.

32.

Regensteiner

Ware

Jr McCarthy

. Effect of cilostazol on treadmill walking, community-based walking ability, and health-related quality of life in patients with intermittent claudication due to peripheral arterial disease: Meta-analysis of six randomized controlled trials. J Am Geriatr Soc 2002; 50: 1939–1946.

33.

Evans

Fornasini

. Pharmacokinetics of L-carnitine. Clin Pharmacokinet 2003; 42: 941–967.

34.

Cao

Wang

Liu

Wang

Han

Wang

. Comparison of pharmacokinetics of L-carnitine, acetyl-L-carnitine and propionyl-L-carnitine after single oral administration of L-carnitine in healthy volunteers. Clin Invest Med 2009; 32: E13–19.

35.

Brass

Hoppel

Hiatt

. Effect of intravenous L-carnitine on carnitine homeostasis and fuel metabolism during exercise in humans. Clin Pharmacol Ther 1994; 55: 681–692.

36.

Brevetti

Perna

Sabba

. Superiority of L-propionylcarnitine vs L-carnitine in improving walking capacity in patients with peripheral vascular disease: An acute, intravenous, double-blind, cross-over study. Eur Heart J 1992; 13: 251–255.

37.

Brevetti

Chiariello

Ferulano

. Increases in walking distance in patients with peripheral vascular disease treated with L-carnitine: A double-blind, cross-over study. Circulation 1988; 77: 767–773.

38.

Stephens

Constantin-Teodosiu

Laithwaite

Simpson

Greenhaff

. A threshold exists for the stimulatory effect of insulin on plasma L-carnitine clearance in humans. Am J Physiol Endocrinol Metab 2007; 292: E637–641.

39.

Wiseman

Brogden

. Propionyl-L-carnitine. Drugs Aging 1998; 12: 243–248; discussion 249–250.

40.

Greco

Mingrone

Bianchi

Ghirlanda

. Effect of propionyl-L-carnitine in the treatment of diabetic angiopathy: Controlled double blind trial versus placebo. Drugs Exp Clin Res 1992; 18: 69–80.

41.

Brevetti

Perna

Sabba

Martone

Condorelli

. Propionyl-L-carnitine in intermittent claudication: Double-blind, placebo-controlled, dose titration, multicenter study. J Am Coll Cardiol 1995; 26: 1411–1416.

42.

Strano

and the Italian propionyl-L-carnitine working group. Propionyl-L-carnitine versus pentoxifylline: Improvement in walking capacity in patients with intermittent claudication. Clin Drug Invest 2002; 22(Suppl 1): 1–6.

A systematic review and meta-analysis of propionyl- l -carnitine effects on exercise performance in patients with claudication

Abstract

Keywords

Introduction

Methods

Systematic review of all studies for PLC in claudication

Study identification

Patient-level meta-analysis

Study identification – primary analysis

Analysis populations – primary analysis

Harmonization of treadmill peak performance assessments

Statistical analysis

Study-level meta-analysis of randomized, controlled trials

Study inclusion and exclusion criteria

Analysis population

Statistical analyses

Results

Systematic review of all studies for PLC in claudication

Identified studies

Patient-level meta-analysis

Analysis population

Effect of oral PLC on peak walking distance

Safety of oral PLC

Meta-analyis of all randomized controlled trials

Effect of oral PLC on peak walking performance

Discussion

Efficacy of PLC

Safety of oral PLC

Limitations of the analysis.

Conclusions

Footnotes

Acknowledgements

Funding

Conflict of interest

References