Exercise-induced arterial endofibrosis (EIAE) affects highly trained endurance athletes, mainly cyclists, with no risk factors for cardiovascular diseases. It usually results in an isolated narrowing (10-20% diameter reduction) in the proximal segment of the external iliac artery, although other localization may be found (1,2,8). Although the exact prevalence remains unknown, it could be as high as 30% of athletes in some cycling teams. The ankle-to-brachial systolic blood pressure index (ABI) is widely used as a significant index for the presence of EIAE (1-4,6,8,9,16,17,20). Because ABI at rest remains within normal limits in athletes with EIAE, exercise testing facilitates the diagnosis by increasing the flow and, thereby, increasing the pressure drop, through arterial lesions (1,2). In lower-extremity arterial disease (LEAD), the amplitude and duration of postexercise ABI decrease depends on both the severity of the arterial lesions and the intensity of exercise (18,22). Thus, as previously suggested, it is expected that early recordings after heavy-load exercise may facilitate the detection of EIAE with ABI (2,20).
Multiple pressure values are required to calculate ABI. During the early recovery period after maximal exercise, systemic systolic arterial pressure decreases rapidly (18). Thereby, minute differences in the time taken for recording, as would result from the recording of the four limbs' arterial pressure values one by one, may result in wide differences in hemodynamic parameters at the time of pressure-value recordings. As a result, it is likely that the absence of simultaneity of pressure recordings for all four limbs may affect the performance of ABI in differentiating normal arteries from arteries with EIAE. This may also be an explanation of the differences observed among the studies reporting ABI diagnostic performance in EIAE (2). Indeed, our previously published recommendation, as well as that of Taylor and George (20), that simultaneous pressure recordings should be preferred, relied on the intuitive conviction that this was preferable to consecutive recordings. In fact, there is no proof of this assumption in the literature, and comparison of the diagnostic performance of pressure measurements issued from simultaneous versus consecutive recording has never been reported.
We intended to check whether after exercise, ABI calculated from simultaneous measurements was better than ABI calculated from consecutive measurements for differentiating athletes with suspected unilateral EIAE from normal athletes. For this purpose, we compared normal athletes with athletes complaining about unilateral pain related to EIAE. Although a specific disease (as compared with atherosclerosis), EIAE is an interesting and unique clinical situation to study the hemodynamic consequences of (surgically proved) isolated moderate arterial lesions on arterial pressures after exercise.
In our laboratory in the past years, all referred cyclists had a systematic simultaneous measurement of arterial pressure on arms and legs. This standard procedure has been approved by our ethics committee. A retrospective study was performed among competition cyclists referred to our laboratory for suspected unilateral EIAE. For the present study, only patients for whom the diagnosis of EIAE could finally be ascertained by histological findings of the operated artery and who were able to return to competition without symptoms were included. This enabled us to confirm the absence of eventual other causes for exercise-induced pain. Between 1993 and 2005, among the patients referred to the laboratory for suspected EIAE, 42 patients (41 males) were operated on and were able to return to competition without symptoms, thus confirming the vascular origin of their unilateral exercise-related pain. In one patient, EIAE was found and treated on both sides, although the patient reported unilateral symptoms. None of the patients with EIAE had any personal or familial risk factor for atherosclerosis. For those 42 patients with EIAE, age, height, and weight were 30 ± 11 yr, 175 ± 7 cm, and 68 ± 7 kg, respectively. Normal controls were healthy asymptomatic competition cyclists who were submitted to the program for obligatory systematic survey of competition athletes and who were selected to be matched on gender, age (± 2 yr), height (± 5 cm), and weight (± 5 kg) to the 42 patients with EIAE. None of the normal subjects had any personal or familial risk factor for atherosclerosis. For the 42 normal controls, age, height, and weight were 29 ± 10 yr, 176 ± 5 cm, and 71 ± 7 kg, respectively. All experiments were performed in a 22-23°C air-conditioned room. Satisfactory pressure recordings were obtained for all tests.
Arterial pressure measurements.
Brachial (BSBP) and ankle (ASBP) systolic blood pressure levels on both sides were measured simultaneously in the supine position after a 5-min rest period before exercise. Simultaneous measurements for all four limbs were repeated every minute from minute 1 to minute 4 during the recovery period after incremental maximal exercise. For pressure measurements, 15-cm arm cuffs were positioned at the level of the brachial and posterotibial arteries, both at rest and during recovery, but these were removed for exercise testing. BSBP and ASBP were measured using four calibrated automatic synchronized sphygmomanometers (Dynamap, 1846SXP Kritikon, Johnson & Johnson, Tampa, FL). The accuracy of pressure measurements obtained with this technique has previously been demonstrated (13). All values were automatically recorded on paper and then transferred to a personal computer for ABI calculation.
In both groups, the subjects were asked to perform one incremental bicycle exercise test in the sitting position (VP100, Techmachine, France). After a warm-up period of 5-10 min at 70 W, work load for the test was started for all subjects at 100 W and was increased by 50 W up to 300 W and then by 30 W every 3 min. Exercise was performed up to exhaustion in asymptomatic athletes or ended on pain limitation in EIAE.
Analysis of results.
We studied both absolute single-leg values for ABI and ASBP and between-leg differences (Δ) in ASBP and ABI from the single incremental bicycle exercise test performed by the subjects (control and patients). Because of the wide day-to-day variability (~15%) reported for arterial pressure measurements in test-retest recordings, the choice was made to mimic the effects of consecutive measurements in a single exercise test rather than to perform two consecutive exercise tests for each subject on different days (one with simultaneous and one with consecutive recordings). Measurements at rest were calculated as for simultaneous postexercise values (Table 1). After exercise ASBP, ABI, ΔASBP, and ΔABI calculations were performed from simultaneous (simu) and consecutive measurements according to Table 1. When calculating from consecutive (cons) measurements, one approach was to start with the suspected limb in EIAE, which appears logical from a clinical point of view (left limb in controls), before using the contralateral limb and, finally, the arms. Note that this results in ASBPsimu being equal to ASBPcons. As a consequence, it is not reported in the results. The other way was to use the values one at a time but in a random order (rand). Randomization was obtained through the generation of 84 random series of four numbers from 1 to 4, each number corresponding to the minute to be used for the calculation and being attributed to one limb.
For Δ values, we kept the absolute values of the calculation to exhibit only positive results.
Sample size determination.
At the first minute of the recovery period from maximal exercise, ABI calculated from simultaneous measurements showed a 90% sensitivity and 87% specificity rate in the diagnosis of moderate arterial lesions in athletes (1). We estimated that the accuracy for the detection of arterial lesion would be 20% lower with parameters calculated from consecutive versus simultaneous measurements with a standard deviation of 25%. Using a binomial distribution (http://calculators.stat.ucla.edu) the minimal number of subjects to be included in each group to reach a two-sided statistical significance level of 0.05 and 80% power was 32.
We used the receiver operating curve (ROC) analysis to study the diagnosis performance of each parameter defined in Table 1. The ROC analysis is a means of calculating the sensibility and specificity of a test for each value of the studied variable in the diagnosis of a disease (10). This approach has the triple advantage of allowing for a) The objective determination of the performance of a test through the calculation of the area under the sensitivity/specificity relationship curve. A perfect test would provide an area under the curve of 1, whereas an area of 0.5 would be the result of a random choice. b) The objective determination of the cutoff point to be proposed for clinical use. The value of the variable resulting in the lowest distance to the 100/100% sensitivity/specificity angle is considered to be the best compromise of sensitivity and specificity for clinical use, assuming an equal "cost" of false-positive and false-negative results. c) The statistical comparison of the area under the ROC curve (area ± SE of area) obtained with different parameters in the same population using the method described by Hanley and McNeil (11).
Using the cutoff point determined through ROC curve analysis, we calculated the sensitivity, specificity, positive (PPV) and negative (NPV) predictive values, and accuracy of the studied parameter. These values are presented with their 95% confidence interval (19). An unpaired t-test was used as a mean of testing whether differences existed between EIAE and normal controls subjects for ASBPsimu, ABIsimu, ΔASBPsimu, and ΔABIsimu, both at rest and during recovery.
For all statistical tests, a two-tailed probability level of P < 0.05 was used to indicate statistical significance.
At rest, ASBPsimu and ABIsimu in EIAE (145 ± 17 mm Hg and 1.13 ± 0.08, respectively) were not significantly different from normal controls (144 ± 24 mm Hg and 1.15 ± 0.10 for ASBPsimu and ABIsimu, respectively). Areas under the ROC curve for single-leg ASBPsimu and ABIsimu were 0.54 ± 0.06 for both parameters (not significant from a random choice). For ΔASBPsimu and ΔABIsimu, no significant difference was found between EIAE (8 ± 13 mm Hg and 0.06 ± 0.11, respectively) and normal controls (8 ± 6 mm Hg and 0.06 ± 0.05, respectively). Areas under the ROC curve using ΔASBPsimu and ΔABIsimu at rest were 0.57 ± 0.06 for both parameters (not significant from a random choice).
After exercise, ASBPsimu and ABIsimu results were significantly (P < 0.0001) lower in the EIAE group compared with normal controls (97 ± 35 mm Hg and 0.49 ± 0.17 compared with 153 ± 35 mm Hg and 0.80 ± 0.15, respectively). Areas under the ROC curves for postexercise values are reported in Table 2. Areas for the ROC curve for single-leg ASBP and ABI values with simultaneous, consecutive, and random approaches ranged from 0.83 to 0.90. As shown in Table 3, the optimal cutoff point to differentiate EIAE from normal athletes was 111 mm Hg for ASBPsimu and 0.64 for ABIsimu. Those cutoff points provided the highest accuracy among the different approaches to single-leg ASBP and ABI, but calculation from simultaneous recordings showed little influence over the results provided by single-leg values. Indeed, ROC curve areas were significantly greater only for ABIsimu compared with ASBPsimu (P < 0.05, r = 0.91) and ASBPrand (P < 0.05, r = 0.68). No significant difference was found between ROC curve areas of ABIsimu and other ways of calculating single-leg ABI.
After exercise, ΔASBPsimu and ΔABIsimu results were significantly (P < 0.0001) higher in the EIAE group compared with normal controls (64 ± 33 mm Hg and 0.32 ± 0.16 compared with 10 ± 7 mm Hg and 0.05 ± 0.04, respectively). For between-leg difference parameters, the highest area under the ROC curve was obtained using ΔASBPsimu and ΔABIsimu, with no difference between the two parameters (Table 2 and Fig. 1). Areas under the ROC curve for ΔASBPsimu and ΔABIsimu were higher than areas under the ROC curve of ΔASBP and ΔABI calculated from consecutive and random measurements (P < 0.01). Further, ΔASBPsimu and ΔABIsimu showed significantly higher areas under the ROC curve than all single-leg parameters and, specifically, single-leg ASBPsimu and ABIsimu values (P < 0.05).
FIGURE 1-Receiver op...Image Tools
Accuracy for ΔASBPsimu and ΔABIsimu in discriminating EIAE from normal subjects with cutoff points of 22 mm Hg and 0.10, respectively, were 93% for both parameters, as shown in Table 3.
ABI is a widely used index for the presence of LEAD. The decrease of ABI at rest is proportional to the presence and severity of arterial lesions. In EIAE of the external artery, a unique example of moderate arterial lesions, absolute ASBP, and ABI remain within the normal range at rest. Increasing the blood flow through a stenosis increases the pressure gradient through the lesion. Therefore, submitting patients to high-work load exercise should theoretically facilitate the diagnosis of moderate arterial lesions. Unfortunately, after heavy-load exercise, ABI also decreases as a function of exercise intensity in normal subjects (7). This is probably why the normal limits to be used after maximal exercise to detect EIAE are still subject to debate (2).
Previous reports have emphasized that pressure should be measured early in the recovery period to improve the diagnostic performance of pressure measurements of moderate arterial lesions (2,20). Unfortunately, in the early recovery period, very small differences in time for blood pressure measurements (even less than a minute) are expected to result in important differences in measured values.
The standard method for indirect pressure measurement using a mercury sphygmomanometer is the auscultatory method in the upper limb (14) and the Doppler ultrasound technique in the lower limb (21). Observer's experience and differences between observers' clinical statutes are common sources of error in the recording of arterial pressure (5,12). Even in trained hands, we assume that the time required to manually measure ASBP approximates the minute. In this case, simultaneous measurements on both limbs and arms after exercise can hardly be attained through manual recordings. The present study confirmed that simultaneous measurements significantly increased the ability to differentiate athletes with EIAE from normal athletes as opposed to calculation from consecutive recordings, specifically when studying the between-leg ASBP and ABI differences. Clinically, the effect of simultaneous versus consecutive systolic blood pressure recordings results in six patients being misdiagnosed with ΔABIsimu versus 17 and 22 patients with ΔABIcons and ΔABIrand, respectively, in our group of 84 patients. When searching for moderate arterial lesions after maximal heavy-load exercise, automated pressure measurements should be used to enable simultaneous recordings. Nevertheless, future studies are required comparing manual with automated recordings for the detection of EIAE.
Previous reports have emphasized the interest of postexercise ABI in the diagnosis of EIAE (1-4,6,8,9,16,17,20). Cutoff points reported in the literature using postexercise single-leg absolute ABI for the diagnosis of EIAE range from 0.5 to 0.66 (2), which is consistent with the values found in the present population. In the present study, a between-leg ASBP or ABI difference in excess of 22 mm Hg or 0.10 at the first minute of recovery after maximal exercise were found to provide the best compromises of sensitivity and specificity to differentiate EIAE patients from normal athletes. The between-leg ABI difference is lower than in previous reports, but it results in a higher positive predictive value (20).
In recent years, Dutch groups have suggested magnetic resonance angiography as the best approach to diagnose "flow limitation in athletes" (15-17). We have recently commented on the difference of the concepts between the Dutch approach, including lengthened arteries, and the French concept, including only fixed stenoses (2). One should remember that the present study is dealing with endofibrosis and not with lengthened arteries. Endofibrosis is defined as an intimal thickening of the artery responsible for a stenosis of the lumen independently of the presence of lengthened artery. Indeed, many patients with endofibrosis present no lengthening of the artery. The concept of flow limitation from flexural arteries is an interesting concept. It sometimes overlaps with endofibrosis, but it may also occur without any significant endoluminal intimal lesion. One should not mix endofibrosis and arterial lengthening. It is clear that in this latter concept, blood pressure measurement and ABI calculation are expected to be of low accuracy.
It is possible that the repeated measurements did not provide pressure values similar to the values that would have been obtained while measuring one limb at a time. It cannot be excluded that taking measurements on all four limbs may have interfered with the results for pressure obtained in the subsequent readings and that measurement of one limb at a time should have been performed. Nevertheless, the approach used in the present study is of advantage to test the sole effect of simultaneity by completely abolishing the risk of differences attributable to technical or physiological changes in test-retest experiments.
Because the present study reports the results of a retrospective analysis of patients diagnosed in our laboratory, it could be suggested that a methodological bias has been introduced. Indeed, single-leg ABI calculations from simultaneous measurements were part of the diagnostic algorithm used to detect the 42 EIAE patients who were included (according to the results observed in the population of cyclists suspected of EIAE and studied between 1993 and 2000). Had the single-leg ABIsimu value shown better diagnostic accuracy than the other approaches, we would not be able to rule out that our results might be biased. This probably cannot be opposed to the diagnostic superiority of ΔABI over single-leg ABI values.
Because of the relatively high prevalence of EIAE in highly trained cyclists, one could argue that there was a risk that subjects included in the normal group could have suffered early asymptomatic lesions. We argue that it does not interfere with the comparison of ABI calculated from simultaneous versus consecutive recordings and with the conclusion about the diagnostic superiority of ΔABI over other calculated parameters. Could the fact that one patient was finally proven to have bilateral lesions in the EIAE group have interfered with the results? It is likely, but at worse this may have lowered the (significant) differences in diagnostic performance between the ΔASBP and ΔABI versus single-leg ASBP and ABI values to discriminate EIAE from normal results.
In conclusion, we suggest that 1) between-leg differences should be preferred to single-leg ASBP and ABI values for the detection of EIAE; 2) to discriminate athletes suspected of unilateral EIAE from normal athletes, simultaneous recordings of all four limbs should be performed; and 3) because simultaneous recordings of all four limbs can hardly be obtained manually, it is likely that automated devices should be used after exercise to improve the detection of EIAE, although comparison with manual detection is required. Future studies should also be performed to prospectively test the diagnostic performance of a ΔASBPsimu of more than 22 mm Hg and a ΔABIsimu of more than 0.10 after maximal exercise for the diagnosis of EIAE in symptomatic athletes. EIAE is a unique situation of surgically proved moderate LEAD, but it is a specific disease. Whether the results of this study are applicable to detect arterial lesions resulting from early atherosclerosis in sedentary subjects remains to be determined.
Alexis Le Faucheur is supportted by a grant from the Conseil Général de Maine et Loire.
The authors thank Dr B. Vielle (Department of biostatistics) for the ROC curve calculation and their comparisons, I. Laporte for technical help, and N. Boileau, who corrected the manuscript.
1. Abraham, P., S. Bickert, B. Vielle, and M. Chevalier. Pressure measurements at rest and after heavy exercise to detect moderate arterial lesions in athletes. J. Vasc. Surg.
2. Abraham, P., P. Bouye, I. Quere, J. M. Chevalier, and J. L. Saumet. Past, present and future of arterial endofibrosis in athletes: a point of view. Sports Med.
3. Abraham, P., J. M. Chevalier, G. Leftheriotis, and J. L. Saumet. Lower extremity arterial disease in sports. Am. J. Sports Med.
4. Abraham, P., J. L. Saumet, and J. M. Chevalier. External iliac artery endofibrosis in athletes. Sports Med.
5. Bailey, R. H., and J. H. Bauer. A review of common errors in the indirect measurement of blood pressure. Sphygmomanometry. Arch. Intern. Med.
6. Chevalier, J. M., P. L'Hoste, J. P. Saint André, A. Mégret, and G. Leftheriotis. Iliac arterial endofibrosis: a top athlete's disease. In: Les lésions occlusives aorto-iliaques chroniques. Actualités de chirurgie vasculaire,
E. Kieffer (Ed.). Paris: AERCV, pp. 443-452, 1991.
7. Desvaux, B., P. Abraham, D. Colin, G. Leftheriotis, and J. L. Saumet. Ankle systolic blood pressure following sub-maximal and maximal exercises in healthy young men. J. Sports Med. Phys. Fitness
8. Ehsan, O., A. Darwish, C. Edmundson, V. Mills, and H. Al-Khaffaf. Non-traumatic lower limb vascular complications in endurance athletes. Review of literature. Eur. J. Vasc. Endovasc. Surg.
9. Fernandez-Garcia, B., J. Alvarez Fernandez, F. Vega Garcia, et al. Diagnosing external iliac endofibrosis by postexercise ankle to arm index in cyclists. Med. Sci. Sports Exerc.
10. Hanley, J. A., and B. J. McNeil. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology
11. Hanley, J. A., and B. J. McNeil. A method of comparing the areas under receiver operating characteristic curves derived from the same cases. Radiology
12. Jeelani, N. U., B. D. Braithwaite, C. Tomlin, and S. T. MacSweeney. Variation of method for measurement of brachial artery pressure significantly affects ankle-brachial pressure index values. Eur. J. Vasc. Endovasc. Surg.
13. Mundt, K. A., L. E. Chambless, and C. B. Burnham. Measuring ankle systolic blood pressure: validation of the Dinamap 1846 SX. Angiology
14. Pickering, T. G., J. E. Hall, L. J. Appel, et al. Recommendations for blood pressure measurement in humans and experimental animals: part 1: blood pressure measurement in humans: a statement for professionals from the subcommittee of professional and public education of the american heart association council on high blood pressure research. Circulation
15. Schep, G., M. H. Bender, D. Kaandorp, E. Hammacher, and W. R. de Vries. Flow limitations in the iliac arteries in endurance athletes. Current knowledge and directions for the future. Int. J. Sports Med.
16. Schep, G., M. H. Bender, G. van de Tempel, P. F. Wijn, W. R. de Vries, and B. C. Eikelboom. Detection and treatment of claudication due to functional iliac obstruction in top endurance athletes: a prospective study. Lancet
17. Schep, G., S. L. Schmikli, M. H. Bender, W. L. Mosterd, E. R. Hammacher, and P. F. Wijn. Recognising vascular causes of leg complaints in endurance athletes. Part 1: validation of a decision algorithm. Int. J. Sports Med.
18. Strandness, D. E. Jr, and R. E. Zierler. Exercise ankle pressure measurements in arterial disease. In: Vascular Diagnosis,
E. F. Bernstein (Ed.). St. Louis, MO: Mosby pp. 547-553, 1993.
19. Sumner, D. S. Evaluation of noninvasive testing procedures: data analysis and interpretation. In: Vascular Diagnosis
. E. F. Bernstein (Ed.). St. Louis, MO: Mosby pp. 39-63, 1993.
20. Taylor, A. J., and K. P. George. Ankle to brachial pressure index in normal subjects and trained cyclists with exercise-induced leg pain. Med. Sci. Sports Exerc.
21. Yao, J. S. T. Pressure measurement in the extremity. In: Vascular Diagnosis,
E. F. Bernstein (Ed.). St. Louis, MO: Mosby, pp. 169-175, 1993.
22. Yao, S. T. Haemodynamic studies in peripheral arterial disease. Br. J. Surg.