Background
Peripheral arterial disease (PAD) is a highly prevalent and severe atherosclerotic condition characterized by progressive peripheral arterial development of lower extremeity stenosis/occlusions [
1]. Patients affected with PAD suffer from reduced quality of life, and more importantly, increased risk of cardiovascular and cerebrovascular events [
2]. Therefore, a noninvasive and objective method is desirable for diagnostic, prognostic, and therapeutic purposes, such as early detection of physiological function changes, clinical risk stratification for predicting myocardial infarction or stroke, and intervention planning for symptomatic patients.
Noninvasive testing of flow-limiting stenosis typically includes measurement of the ankle–brachial index (ABI), the ratio of ankle systolic blood pressure to arm systolic pressure [
3]. PAD is considered to be present when the ABI is ≤0.90 and severe when the ABI is ≤0.50 [
3]. However, ABI has low sensitivity for PAD diagnosis and may not be necessarily associated with symptom relief after interventions [
4,
5]. Transcutaneous oxygen pressure (TcPO2) measurement is an additional method used to indirectly assess the degree of ischemia in ischemic skeletal muscle by measuring tissue oxygenation [
6]. The use of TcPO2 measurement is limited because it is confined to the skin and thus does not accurately reflect muscle perfusion [
7].
Medical imaging has emerged as an important tool in the diagnosis and management of PAD. Imaging modalities, including computer tomography (CT) angiography, cardiovascular magnetic resonance (CMR) angiography, and digital subtraction angiography, are commonly used to assess abnormal blood vessels and blood flow to the lower extremities. These techniques fail to provide information regarding skeletal muscle microvascular perfusion in the affected extremity [
8]. However, as PAD extends beyond the large-vessels, blood flow impairment leads to microvascular dysfunction. Precise assessment of skeletal muscle perfusion would facilitate the comprehensive evaluation of PAD and could be combined with conventional angiography to reveal both functional and anatomical characteristics.
Several CMR techniques can noninvasively measure microvascular perfusion using endogenous tracers, including arterial spin labeling (ASL), blood oxygenation level-dependent (BOLD), and intravoxel incoherent motion (IVIM). ASL magnetically tags arterial blood using radiofrequency pulses, and the perfusion contrast is given by the signal difference between the tagged image and the nontagged control image obtained without net magnetization perturbation in arterial blood [
9]. BOLD uses the paramagnetic effect of deoxygenated hemoglobin as an intrinsic contrast agent, which decreases the T2* relaxation signal [
10]. IVIM is a variant of conventional diffusion-weighted imaging by separating the effect of blood flow in the randomly oriented capillary network from that of thermally driven water molecular diffusion [
11]. ASL, BOLD, and IVIM have been successfully applied to measure skeletal muscle perfusion in previous studies [
12‐
14]. Since perfusion-related metrics derived from these different CMR techniques are based on completely distinct mechanisms, they may depict different aspects of muscle perfusion properties. ASL is more related to the function of blood delivery to target tissues, BOLD is related to tissue oxygenation, and IVIM is related to pseudodiffusion within capillary beds. We hypothesize that multi-parametric CMR techniques, including ASL, BOLD, and IVIM, could provide complementary information regarding perfusion in skeletal muscles and would represent various alteration patterns in the presence of perfusion deficits.
Hence, this study aimed to 1) test the feasibility of using ASL, BOLD, and IVIM to measure perfusion changes in the lower extremities of healthy subjects under different external compression statuses; 2) validate the associations between ASL-, BOLD-, and IVIM-derived parameters and TcPO2 measurements; 3) evaluate the effects of age on imaging parameter measurements of ASL, BOLD, and IVIM in healthy subjects at rest; 4) use ASL, BOLD, and IVIM to compare perfusion in affected and contralateral (asymptomatic) lower extremities in PAD patients at rest; 5) compare the capabilities of resting-state ASL, BOLD, and IVIM in detecting perfusion differences between PAD patients and age-matched healthy subjects and between mild-to-moderate and severe PAD patients; and 6) investigate the associations between ASL-, BOLD-, and IVIM-derived parameters and ABI in PAD patients.
Discussion
Noninvasive monitoring of skeletal muscle perfusion in the lower extremities is critical for PAD patient management as perfusion can provide insight into microvascular function and endothelial integrity [
5]. Advanced CMR techniques, including ASL, BOLD, and IVIM, have been utilized in the assessment of skeletal muscle perfusion [
12‐
14]; however, to our knowledge, few studies have been performed to directly compare these techniques in the same subject cohort [
5,
26]. Furthermore, there is limited literature on comparisons of these CMR parameters with routinely used parameters in clinical practice such as TcPO2 and ABI [
5,
7,
13]. The results of our study suggested that 1) ASL, BOLD, and IVIM could respond to cuff-induced ischemia in healthy subjects—that is, when the difference reached a significant level, ASL-derived blood flow values, BOLD-derived T2* values, and IVIM-derived
D values tended to decrease with increasing external pressure while IVIM-derived
f values tended to increase under cuff compression; 2) blood flow, T2*, and
f values were all correlated with TcPO2 measurements; 3) ASL of all muscle groups and IVIM of the gastrocnemius group were influenced by age; only BOLD was independent of age in every muscle group; 4) BOLD could detect perfusion differences between the affected and contralateral lower extremities in PAD patients; 5) BOLD could separate PAD patients from healthy old subjects and PAD patients with different severities; and 6) BOLD-derived T2* was correlated with ABI in PAD patients.
Similar to the results in published literature [
19,
27], baseline blood flow in skeletal muscle as measured by ASL in the lower extremity was mostly near or less than 20 ml/100 g/min in our study. For BOLD CMR, we observed a baseline mean T2* value of approximately 25 ms in healthy subjects, which lies within the previously reported range of 22–27 ms [
13,
25,
28,
29]. IVIM imaging of skeletal muscle in the lower extremity has rarely been studied. In other parts of the body, IVIM-derived
f,
D, and
D* values were reported to be 3%, 1.45 × 10
− 3 mm
2/s, and 28.5 × 10
− 3 mm
2/s, respectively, in the forearm muscle at rest by Filli et al. [
14], and 6.6%, 1.45 × 10
− 3 mm
2/s, and 11.7× 10
− 3 mm
2/s, respectively, in the shoulder muscle by Nguyen et al. [
30]. These findings are consistent with our observations (e.g., 5.9%, 1.52 × 10
− 3 mm
2/s, and 18.6 × 10
− 3 mm
2/s, respectively, in the soleus muscle in healthy young subjects).
Under cuff compression conditions, negative ASL and BOLD contrasts in healthy young subjects developed due to ischemic insult, which agreed with previous studies [
7,
12,
13,
25,
31]. ASL is capable of measuring blood flow through muscle tissue microvasculature given that ASL and radiolabeled microsphere measurements in rat leg muscle have shown good correlation for perfusion [
32]. Cuff compression interrupted both arterial inflow and venous outflow simultaneously, thus provoking reduced blood flow obtained by ASL. Although the exact source of the BOLD signal in skeletal muscle is not yet fully understood, it is generally accepted that the signal is primarily associated with capillary and blood oxygenation state [
7]. Lebon et al. also found that the T2* signal in muscle rapidly decreased during ischemia and attributed this change to early hemoglobin desaturation [
33]. This finding is logical given that the BOLD signal changed almost synchronously with hemoglobin desaturation but preceded myoglobin desaturation [
33], as the dissociation constant of hemoglobin is more than 10 times higher than that of myoglobin. IVIM-derived
D values also showed a decreasing trend under cuff-induced ischemic conditions. Local ischemia leads to decreased diffusivity of water molecules within muscles, and this decreased
D was mostly likely attributed to this physical effect. Conversely, the perfusion fraction
f derived from IVIM showed a tendency to increase in the case of arterial occlusion. Given that
f reflects the signal fraction of capillary blood flow in entire water molecule diffusion pool within each voxel [
11], it can be hypothesized that obstruction of venous reflux is probably responsible for the altered
f values. In addition, it has been suggested that decrease in venous oxygen saturation may release relaxing factors than can cause microvascular dilation [
34], which may also increase the
f value. In healthy old subjects, changing trends for the imaging parameters were similar to those observed in the healthy young group except for ASL. ASL is limited by the intrinsic low SNR in skeletal muscle, wherein the ASL signal represents only 0.5%–1% of the raw image intensity [
8]. This notion may account for the lack of statistical significance of ASL measurements, especially given that the healthy old sample number was small (
n = 10).
Interestingly, we also observed aberrant T2* signal changes on the control sides of healthy subjects during the cuff compression experiment on the other lower extremity, in accordance with the findings of a previous study [
35]. Yeung et al. suggested that this might be because of the high sensitivity of BOLD CMR imaging to local magnetic field disturbances caused by magnetic susceptibility effects, which may be induced by oxygen in the air at high pressure during cuff inflation [
35].
Ledermann et al. reported that BOLD CMR imaging correlated with TcPO2 measurements in healthy volunteers during muscle ischemia [
7], which was consistent with our results. However, the correlation observed by Ledermann et al. was stronger (correlation coefficient, 0.96) than that in our study (correlation coefficient, 0.522), primarily because Ledermann et al. averaged signal intensities across all volunteers for statistics, not for individuals. In addition, ASL-derived blood flow and IVIM-derived
f values were also found to correlate with TcPO2 measurements in our study, with correlation coefficients of 0.465 and − 0.351, respectively. The correlation between BOLD-derived T2* and TcPO2 measurements was stronger than those between the other parameters, which may be attributed to the fact that both BOLD CMR imaging and TcPO2 examination are directly associated with the oxygenation state at the microvascular level.
Age effects on parameter measurements varied among the different sequences and muscles. In our study, no significant differences between the healthy young and healthy old groups were observed for the baseline T2* value, indicating that age did not appear to affect BOLD in healthy subjects at rest. This finding was in accordance with other observations [
28]. In contrast, ASL and IVIM were more easily influenced by the age factor especially in the gastrocnemius muscle group. The gastrocnemius is a fast-twitch type muscle, whereas the soleus belongs to the slow-twitch type. Degenerative processes of muscle fibers have been demonstrated to differ with fiber type, and the fast-twitch muscle is more prone to aging and fatigue [
36,
37].
The ABI is a measure providing objective data for diagnosing PAD. When applying these imaging techniques on PAD patients, we found that BOLD was capable of detecting perfusion deficits at rest better than ASL and IVIM. Lower T2* values were related to the presence of PAD and more disease severity stratified by ABI. As discussed earlier, BOLD effect is generally assumed to reflect blood oxygenation state influenced by the ratio of oxygenated to deoxygenated hemoglobin, which is determined by the balance between oxygen supply and consumption [
15]. In PAD, arterial blood flow in the lower extremities is limited, leading to reduced oxygenated hemoglobin. Moreover, the impaired vascular function causes a longer contact time between blood and myocytes, leading to more efficient deoxygenation of hemoglobin [
15]. These two effects both contribute to a reduced T2* value. In a previous study by Englund et al., BOLD-derived metrics under the ischemia-reperfusion paradigm were also found to be correlated with ABI, suggestive of disease severity-dependent impairment of vascular response in PAD patients [
5]. Our results suggest that even at rest, vascular function at the tissue level could be indicative of disease progression. However, unlike CMR imaging, ABI cannot be obtained from all patients especially in patients with critical limb ischemia.
In the cuff-induced arterial occlusion experiment on healthy subjects, varying degrees of ischemia were factitiously induced by different pressures above the systolic pressure, and CMR imaging techniques were able to detect these changes. However, in healthy old subjects and PAD patients with differing degrees of ischemia, only BOLD was effective for this discrimination at rest. One possible explanation for this finding could be that the degree of ischemic insult in PAD patients was less severe than that induced by cuff occlusion. Additionally, collateral arteries developed in skeletal muscles in PAD patients would compensate for the perfusion deficit. BOLD is more sensitive to these less dramatic changes, which may be ASL- or IVIM-insensitive.
Numerous previous studies have used ASL or BOLD to monitor dynamic perfusion changes in skeletal muscles at rest and during ischemia and hyperemia, which allows the measurement of key parameters, such as peak hyperemic value (PHV) and time-to-peak (TTP). In our study, we did not measure continuous temporal changes of ASL, BOLD, and IVIM in our subjects mainly because it was not technically feasible to perform these three sequences sequentially at a high temporal resolution. Nevertheless, our protocol can be regarded as a simplified approach to the dynamic scanning method; a similar approach was also used in a previous study [
38]. Although PHV and TTP proved useful for the assessment of PAD in most published studies, conflicting results also exist [
13]. In addition, the reproducibility of the data is another concern. Versluis et al. investigated the reproducibility of BOLD-derived PHV and TTP values in healthy subjects and PAD patients [
39]. The reproducibility was unsatisfying with a coefficient of variation up to 26.7% and an ICC value as low as 0.59 [
39]. Moreover, due to massive pain or risk of worsening the clinical condition caused by cuff compression, patients with critical limb ischemia, ulceration, necrosis, or gangrene should be considered with caution or even excluded from the study [
40]. Compared with the cuff compression paradigm, the resting-state imaging scheme is simple to perform, less time-consuming, and more acceptable to PAD patients [
24]. In our study, we wanted to investigate and were most concerned with whether the baseline measurements of these techniques had value in assessing PAD.
The present study has several limitations. First, the number of study subjects was relatively small. Second, no gold standard for blood flow, oxygenation, or microvascular perfusion in skeletal muscle could be established in our subjects. For example, mixed venous oxygen saturation measurements would be informative regarding the confirmation of BOLD results. Nevertheless, our results revealed significant relationships with TcPO2 measurements, which are commonly used in clinical routines. However, the use of TcPO2 measurement is limited since it is confined to the skin microvasculature and thus fails to directly analyze the skeletal muscle [
7]. Therefore, once further validated, noninvasive CMR techniques might be used to gain direct information regarding skeletal muscle perfusion in lower extremities. Third, the cuff compression paradigm was used in the study instead of exercise. Exercise is more physiologically and clinically relevant. A previous study showed that muscle perfusion at peak exercise was correlated with 6-min walk distance in lower extremities [
41]. Compared with exercise, cuff compression has improved test-retest reproducibility [
42] and less motion artifacts [
40]. Moreover, cuff compression allow the assessment of muscles which respond less to commonly used ankle flexion exercise [
42]. Lopez et al. suggested the use of cuff compression in a more general study population and the use of exercise in specific PAD therapies in claudicants based on its physiologic and clinical relevance [
42]. Fourth, skeletal muscle energetics were not investigated in the current study.
31P CMR spectroscopy is a useful tool to noninvasively probe skeletal muscle energetics, including adenosine triphosphate and creatine phosphate metabolism [
43], which would help to better understand the perfusion results in our study. Further studies are warranted. Finally, CMR angiography was not performed in the current study.