Blood Oxygenation Level-Dependent CMR-Derived Measures in Critical Limb Ischemia and Changes With Revascularization

Background Use of blood oxygenation level-dependent cardiovascular magnetic resonance (BOLD-CMR) to assess perfusion in the lower limb has been hampered by poor reproducibility and a failure to reliably detect post-revascularization improvements in patients with critical limb ischemia (CLI). Objectives This study sought to develop BOLD-CMR as an objective, reliable clinical tool for measuring calf muscle perfusion in patients with CLI. Methods The calf was imaged at 3-T in young healthy control subjects (n = 12), age-matched control subjects (n = 10), and patients with CLI (n = 34). Signal intensity time curves were generated for each muscle group and curve parameters, including signal reduction during ischemia (SRi) and gradient during reactive hyperemia (Grad). BOLD-CMR was used to assess changes in perfusion following revascularization in 12 CLI patients. Muscle biopsies (n = 28), obtained at the level of BOLD-CMR measurement and from healthy proximal muscle of patients undergoing lower limb amputation (n = 3), were analyzed for capillary-fiber ratio. Results There was good interuser and interscan reproducibility for Grad and SRi (all p < 0.0001). The ischemic limb had lower Grad and SRi compared with the contralateral asymptomatic limb, age-matched control subjects, and young control subjects (p < 0.001 for all comparisons). Successful revascularization resulted in improvement in Grad (p < 0.0001) and SRi (p < 0.0005). There was a significant correlation between capillary-fiber ratio (p < 0.01) in muscle biopsies from amputated limbs and Grad measured pre-operatively at the corresponding level. Conclusions BOLD-CMR showed promise as a reliable tool for assessing perfusion in the lower limb musculature and merits further investigation in a clinical trial.

P eripheral arterial disease (PAD) affects 27 million people in North America and Europe (1) and is characterized by progressive development of arterial stenoses and occlusions. Most individuals with PAD are asymptomatic (2), with the remainder developing intermittent claudication or critical limb ischemia (CLI). The latter is characterized by pain at rest, ulceration or gangrene, and high morbidity and mortality. Up to 25% of patients even- times. This was first demonstrated when mapping areas of brain activation, where blood oxygenation is inversely proportional to T2* relaxation time (15,16).
BOLD-CMR has subsequently been used to delineate areas of poorly perfused myocardium (17,18), but the lack of reproducibility and objectivity when measuring perfusion in the lower limb has limited its utility (12).
The present study aimed to: 1) test the feasibility and reproducibility of using BOLD-CMR to measure perfusion in the lower limb; 2) use BOLD-CMR to compare perfusion in the critically ischemic and contralateral (asymptomatic) limb in the same patient, as well as To standardize scans, all subjects refrained from caffeine and exercise 4 h before imaging and rested supine for 5 min. All subjects were imaged with a 3-T Philips Achieva scanner (Philips Healthcare, Best, the Netherlands) using a multi-echo single-shot gradient recalled echo (GRE) sequence.
Reactive hyperemia, instigated by cuff-induced arterial occlusion followed by rapid cuff deflation, was used to elicit T2* changes ( Figure 1A, Online Appendix).   Figure 1B

RESULTS
A total of 37 patients with CLI, 10 age-matched control subjects, and 12 young healthy control subjects were recruited into the study ( Table 1, Online Figure 1).
Three CLI patients were excluded, as they were unable to lie supine long enough to undergo CMR because of rest pain severity.
A T2-weighted TSE image allowed accurate visual delineation of muscle groups ( Figure 2A). A high spatial resolution was achieved with the GRE-BOLD-CMR sequence, which allowed for accurate exclusion of major blood vessels, bone, and soft tissue from muscle ROIs ( Figure 2B). T2* perfusion maps highlight changes in muscle perfusion during cuffing and reactive hyperemia ( Figures 2C and 2D).
Automated analysis identified Grad and SRi as the most informative in discriminating between ischemic and patient contralateral limbs (p < 0.0001 for both) (  3F). Bland-Altman analysis confirmed good reproducibility between scans for Grad (mean bias: 0.0004; 95% limits of agreement: À0.14 to 0.14) and SRi (mean bias: À0.92; 95% limits of agreement: À4.88 to 3.04) ( Figures 3G and 3H). No significant difference was found for Grad and SRi between interval scans for all muscle groups analyzed (Online Figure 2B).  Values are median (range) or n.
CLI ¼ critical limb ischemia.   Figure 4).    Table 1). One patient was excluded because of the presence of significant edema on both pre-and post-intervention imaging. T2* curves in the ischemic limb generated prior to revascularization were consistent with curves typical of limbs with impaired perfusion ( Figure 5A). This was confirmed by significantly lower Grad and SRi measurements compared with both the contralateral asymptomatic legs (Grad: 0.14 AE 0.10 ms/s vs. 0.28 AE 0.14 ms/s; p < 0.0001; and SRi: 8.08 AE 5.04% vs. 11.03 AE 5.62%; p < 0.005) and age-matched control subjects (Grad: 0.38 AE 0.17 ms/s; p < 0.0001; and SRi: 13.77 AE 6.33%; p < 0.0001).

patients (Online
Revascularization resolved rest pain in all patients and successfully healed tissue loss in all but 1 patient, who required bypass surgery because of stent occlusion 3 months post-intervention. Improved ABPIs were measured in 8 patients. In 5 patients, ABPI was not measurable either pre-or post-operatively due to crural vessel calcification ( Figure 5B).
Repeat BOLD-CMR after revascularization resulted in T2* curves consistent with improved perfusion, resembling those generated from the patient contralateral limbs ( Figure 5A). There were significant improvements in Grad and SRi after limb revascularization (Grad: 0.14 AE 0.10 ms/s vs. 0.32 AE 0.18 ms/s; p < 0.0001; and SRi 8.08 AE 5.04% vs. 11.57 AE 6.31%; p < 0.005) ( Figures 5C and 5D), suggesting that intervention had increased perfusion in the calf. p > 0.05) ( Figure 5G). The change in ABPI that   (A) Mean signal intensity time course curves for the 5 muscle groups (anterior, lateral, soleus, gastrocnemius, and deep) highlight differences in T2* signal intensity provoked by reactive hyperemia. Consistent differences, particularly during ischemia and reactive hyperemia, are seen among the young control subjects (blue), age-matched control subjects (orange), the patient's contralateral limb (gray), and the patient's ischemic limb (red). Values (mean AE SD) for Grad (B) and SRi (C) are significantly lower in the ischemic limb compared with the patients' contralateral limb, age-matched control subjects, and young healthy control subjects. ANOVA ¼ analysis of variance; other abbreviations as in Figure 1.

DISCUSSION
We believe that the present study is the first to consistently demonstrate that BOLD-CMR is an fibrosis in patients with renal impairment. One advantage of ASL and DCE sequences is that, unlike BOLD, they allow absolute quantification of perfusion, but this is dependent on tracer kinetic modeling, which can introduce inaccuracies (19).
To determine whether BOLD-CMR-derived parameters (Grad and SRi) reflect the extent of tissue vascularity, these were correlated with the C:F ratio determined using histological analysis of corresponding muscle biopsies from amputated limbs.
Only Grad showed a positive correlation with C:F ratio, suggesting that this parameter may be sensitive to the extent of tissue vascularity in the ischemic calf. Consistent with previous reports using BOLD-CMR, our study demonstrates a greater fall in T2* signal during cuffing of the control versus ischemic limbs (11). The earlier leveling off in T2* signal after cuffing in CLI patients may represent a reduced physiological

A B
The capillary-fiber (C:F) ratio in the poorly-perfused calf musculature of amputated limbs correlates with corresponding Grad (A) measurements but not with SRi (B). Abbreviations as in Figure 1. reserve of muscle to ischemic stress in these patients (21). Similarly, the slower rise in T2* signal seen during reactive hyperemia in CLI patients may be due to less efficient delivery of oxygenated hemoglobin to muscle.
In patients with CLI, we measured a higher Grad  Kingdom. E-mail: bijan.modarai@kcl.ac.uk.