Racial Heterogeneity in Treatment Effects in Peripheral Artery Disease: Insights from the CLEVER Trial

Improving symptoms, functions and quality-of-life (i.e., health status) is one of the important goals in treatment of patients with peripheral artery disease (PAD).¹ However, it is unknown whether health status responses differ by race (black vs. white) with alternative PAD treatment modalities. Such differences may exist given the disproportionate burden of PAD in minority populations, and rapid disease progression as compared with white,² as well as differences in psychosocial and economic factors and possibly differences in exercise level, which is known to improve outcomes in patients with PAD.^3,4 Understanding whether black and white respond differently to treatments can help us better support targeted therapy to improve quality of care. This is relevant because until now supervised exercise (SE) programs for PAD was not available in the United States, and recently the Centers for Medicare and Medicaid Services agreed to reimburse for SE therapy.⁵ If there is heterogeneity in response to SE, knowing this difference is important to provide patient-centered care and to get maximum treatment benefit for each unique patient population.

The Claudication: Exercise Versus Endoluminal Revascularization (CLEVER) trial randomized patients with claudication from aortoiliac disease to SE, stent therapy (ST), or optimal medical care (OMC). Short-term results indicated superior treadmill walking performance (i.e., peak walking time and claudication onset time) with SE than either ST or OMC. Conversely, benefit in PAD-specific health status (Peripheral Artery Questionnaire [PAQ] summary score)⁶ was more favorable for ST than either SE or OMC.³ Similarly, general quality-of-life benefit as assessed with Short Form-12 Physical Component Summary [SF-12 PCS]⁷ was similar for ST as well as SE when compared with OMC.³ Long-term results, however, showed similar and sustained benefits for both ST and SE over OMC in treadmill walking performance, but PAQ summary score was more favorable for ST when compared with SE or OMC.⁴ Improvement in SF-12 PCS was seen only with SE.⁴

The purpose of this study is to understand whether there is heterogeneity in response to alternative treatment modalities such as SE, ST or OMC by race (black vs. white) and whether any difference in treatment varies over time, using data from the CLEVER trial.

Methods and Results

Details about the CLEVER trial have been reported before.^3,4 Briefly, it examined the benefits of ST, SE or OMC on both walking outcomes and quality-of-life measures in 119 patients with moderate to severe intermittent claudication and hemodynamically significant aortoiliac arterial stenosis from 22 sites in the United States and Canada. For the current analysis, data were accessed through National Heart, Lung and Blood Institute data repository (https://biolincc.nhlbi.nih.gov/studies/clever/?q=clever) with Institutional Review Board approval from Saint Luke’s Hospital of Kansas City, Missouri. Our primary outcomes of interest were changes in PAQ and SF-12 PCS scores from baseline at 6 and 18 months following randomization. Higher change scores represent greater health status improvements. We also examined changes in treadmill walking performance (peak walking time and claudication onset time) and walking impairment questionnaire (WIQ). We did not assess other outcomes that did not vary by treatment in the CLEVER study.³ We excluded 7 patients of races other than black or white. Change from baseline was analyzed using linear mixed effect models, including treatment groups, race and follow-up time in months as fixed effects. We examined all 2- and 3-way interaction terms to test for differences in treatment response over time between race groups, and used an unstructured covariance matrix to account for repeated measurements. The models were adjusted for baseline health status and baseline characteristics that differed within races or by treatment within races (age, sex, hypertension, smoking status, diabetes, arthritis/musculoskeletal disorders, stroke, myocardial infarction and percutaneous coronary intervention).

Among 104 eligible patients, 41, 43 and 20 patients were randomized to SE, ST and OMC, respectively. The mean age was 64.2 years and 37.5% were women. Seventy-five patients were white (OMC 14, SE 25, ST 36) and 29 were black (OMC 6, SE 16, ST 7). Follow up at 18 months were similar for white vs. black (86.7% vs. 89.7%, respectively) with no intervening death. At baseline there were no significant racial differences in resting ankle brachial index, use of antiplatelet, statin or cilostazol therapy, PAQ summary scores, SF-12 PCS, peak walking time, claudication onset time or walking impairment questionnaire. However, compared with white participants, black participants were more likely to be women (55.2% vs. 30.7%), current smoker (58.6% vs. 52.0%), have diabetes (40.7% vs. 18.7%) and arthritis/other musculoskeletal disorder (44.8% vs. 22.5%). Within each racial group, there were no significant baseline differences by treatment groups in resting ankle brachial index, PAQ summary scores, SF-12 PCS, peak walking time, claudication onset time or walking impairment questionnaire.

There was a significant race-treatment interaction for PAQ summary scores (p=0.035); in white, PAQ scores increased only with ST, whereas in black they increased with both ST and SE, compared with OMC (Figure, upper panel). Interestingly, in the OMC group, PAQ score decreased over time in black but not in white. Model-estimated mean changes (95% CI) in PAQ summary scores in SE and ST compared with OMC were 2.9 (−10.0, 15.8) and 26.6 (14.6, 38.6) in white, and 28.2 (8.7, 47.7) and 31.8 (10.4, 53.2) in black, respectively, which was unchanged at 6 and 18 months (p=0.22 for race-month interaction and p=0.38 for race-treatment-month interaction). A significant race-treatment interaction was also found for SF-12 PCS (p=0.005), which increased only with ST in white and only with SE in black, compared with OMC (Figure, lower panel). Model estimated mean changes in SF-12 PCS scores in SE and ST compared with OMC were 3.9 (−0.6, 8.4) and 6.7 (2.5, 10.9) in white, and 15.9 (7.0, 24.8) and 5.8 (−2.4, 14.1) in black, respectively which was unchanged at 6 and 18 months (p=0.28 for race-month interaction and p=0.64 for race-treatment-month interaction). Similarly, model-estimated changes in PAQ scores were −23.7 (−34.4, −13.0) and −3.6 (−21.2, 14.0) in SE compared with ST in white and black, respectively, and for SF-12 the scores were −2.8 (−6.8, 1.2) and 10.0 (3.1, 17.0), respectively. The race-treatment, race-month and race-treatment-month interactions were not significant for other outcomes (all p>0.05).

Figure.

Unadjusted PAQ summary (upper panel) and SF-12 PCS (lower panel) scores means by race and treatment. PAQ = peripheral artery disease questionnaire. SF-12 PCS = short form-12 physical component score

Comment

We found that although PAD-specific health status scores were greater with both SE and ST compared with OMC in black, such difference was seen only with ST in white. Furthermore, compared with ST, PAQ summary scores were lower with SE in white, but were not different in black. Compared with OMC, general-health status scores were greater only with SE in black and only with ST in white. However, when compared with ST, SF-12 PCS scores were greater with SE in black, but were not different in white. No significant differences were noted for treadmill walking performance or WIQ.

While the smaller sample size reduces statistical power, the observed significant racial differences in health status may suggest racial heterogeneity in treatment responses. Should these findings be replicated in larger studies, we need to understand why SE may be more beneficial in black, as compared with white. Identifying whether these differences in treatment response are mediated by psychosocial stress, economic factors, adherence, baseline exercise level or other unknown factors could help identify opportunities to tailor PAD treatment strategies to specific racial groups or other patient-centered factors who would benefit the most. Black may have a more compromised starting situation, such as socio-economic, mental health or risk factor control, as seen in this study. Engaging in exercise could provide greater overall benefit in black.

A difference of 8 points in PAQ and >5 points in SF-12 are considered clinically important.³ The observed changes in health status are clinically significant. If substantiated later, additional studies should corroborate these improvements with other clinical measures to improve interpretability. Furthermore, the Centers for Medicare and Medicaid Services recently agreed to reimburse SE programs,⁵ and targeting such therapy to right patient population will be most impactful.

The CLEVER trial enrolled selected patients with aortoiliac disease irrespective of femoropopliteal PAD. Therefore, our findings may not extend to patients with isolated femoropopliteal lesions or patients not meeting trial eligibility criteria. Despite three randomization arms (SE, ST and OMC), OMC was provided in all patients and the primary interest of the trial was to compare the effect of SE with ST.³ So our results require careful interpretation when considering OMC as a control.

While some overlap between PAQ and SF-12 PCS is expected, PAQ provides PAD-specific information that SF-12 does not. We did not find significant interactions for mobility-based outcomes, like WIQ and treadmill-based measures. While WIQ provides information on PAD-specific mobility, it does not provide other quality-of-life information.⁸ Whether this explains the disparate findings requires further study.

This hypothesis-generating post hoc analysis of the CLEVER trial demonstrates differential quality-of-life benefits by race with alternative PAD treatment modalities. These findings warrant further examination to confirm the veracity of these observations and to understand the mechanisms responsible for observed treatment responses so that treatment approaches can be optimized to fit patients’ needs, preferences and potential benefits.