Background
The incidence of critical limb ischemia (CLI) for the general population is approximately 500–1000 per million per year in European and North American populations. The estimated CLI prevalence for people aged 60–90 years is approximately 1.0% (range, 0.5–1.2%) [
1]. Conversely, 6.0% of dialysis patients with peripheral arterial disease (PAD) have undergone amputation (range, less than 2.0% in Japan to 10.0% in the United States) in the DOPPS study [
2]. Many mechanisms have been used to detect PAD and lower limb ischemia [
3,
4]. Kovacs et al. [
5] reported that toe pressure and transcutaneous oximetry (TcPO
2) with exercise were used to detect PAD in 120 patients. The results indicated that the toe brachial index (TBI) with exercise provides a reliable receiver-operating characteristic (ROC) curve for PAD. Conversely, Jalkanen et al. [
6] used the crural index to divide patients into five groups, and those in the crural index group IV had unfavorable survival outcomes. Hardman et al. [
7] also classified PAD severity based on atherosclerotic lesions and suggested a potential link between PAD severity and therapeutic treatment.
Skin perfusion pressure (SPP) has been considered a useful tool for detecting PAD severity with TcPO
2 [
8]. The disadvantages of SPP are that it is time-consuming, uncomfortable because of the blood pressure cuff used for patients and the pressure created in the lower extremities, and relatively expensive. Recently, Hijden et al. [
9] divided patients into low, middle, and high body mass index groups and performed peripheral arterial tonometry, laser Doppler flowmetry (LDF), and digital thermal monitoring to determine endothelial functions; their data suggested that LDF could be used to predict the prevalence of cardiovascular events and endothelial dysfunction.
For hemodialysis patients with a higher prevalence of atherosclerotic disease and amputation, a large volume of blood is removed during each dialysis session. In the Japanese Society of Dialysis (JSDT) guideline [10], Critical Limb Ischemia (CLI) was defined as Fontaine’s classification stage III or IV. However, it is too late to diagnose PAD with rest pain or ulcer with Fontaine stage III and IV to prevent progression of PAD to severe stage. Therefore, early detection of PAD is required in Fontaine II; however, dialysis patients often have complications with bone and joint disorders, and it is difficult to evaluate intermittent claudication of Fontaine stage II. Now, we tried to detect PAD in early stages, and this was the first time we used mini-laser Doppler blood flowmetry to measure blood flow in the lower limbs of hemodialysis patients. The system comprises a laser Doppler probe and a small handheld monitor with a laptop computer (Additional file
1: Figure S1a). This study aimed to determine the correlation between reduction of blood flow in early stages of PAD. We defined PAD as JSDT guideline plus Fontaine stage II. The information gained could be helpful for rapidly diagnosing lower limb ischemia in dialysis patients.
Discussion
To our knowledge, this is the first study to evaluate skin microcirculation using LDF for the lower limbs of dialysis patients with or without PAD and to compare the results with those of healthy controls, which allowed for further investigation of skin blood flow changes during dialysis. Early-stage diagnosis of PAD in dialysis patients to prevent deterioration in blood flow, ulcers, necrosis, and systemic infections is critical. Furthermore, surveillance of PAD is required, but the current technology used for SPP measurement only comprises large machines that are not portable, potentially leading to longer time required for the diagnosis. The pocket LDF is handheld and has a small sensor compatible with a laptop computer. Its small size makes the system easily transportable to various clinics for diagnosing PAD.
In conjunction with LDF, the mechanism used to measure the capillary blood flow directly under the skin during dialysis has been previously reported [
28‐
30]. This method was developed to measure the blood flow at the skin surface and to evaluate the function of autonomic neurons. Autonomic dysfunction could be sensitively detected by comparing ear blood flow using LDF for patients with and without DM. In this study, a small and sensitive system was applied to detect PAD among dialysis patients and measure the lower limb blood flow on the skin surface of the dorsal and plantar areas of the foot. According to the logistic regression, SPP-Dorsal Area was significantly correlated with outcomes, but not with SPP-Plantar Area, and LDF-Plantar-Qb was significantly correlated with outcomes, but not significantly correlated with LDF-Dorsal-Qb. We inferred that SPP-Dorsal Area detected blood pressure of the dorsal arteries, whereas the LDF measurement was responsive to microperfusion of the skin. The dorsalis pedis artery is located nearer to the dorsal surface than to the plantar area of the foot. Therefore, SPP more accurately reflects micro-blood pressure in the dorsal area, but the dorsal area is rich with tendons and the bulb of the hair root, which has less micro-blood flow than the plantar area, resulting in a more accurate reflection of skin micro-perfusion than SPP in the plantar area.
Based on the univariate logistic analysis of PAD, covariates with
p < 0.05 were included in the multivariate analysis (serum creatinine, C-reactive protein, total choline, SPP-Dorsal Area, and LDF-Plantar-Qb) and were detected as the final covariates. We attempted to evaluate the cut-off value for both SPP-Dorsal Area and LDF-Plantar-Qb. With real-world data, we would evaluate the outcomes with the multivariate analysis, but not with the univariate logistic analysis. Nevertheless, using the ROC analysis, outcomes were evaluated based on the univariate logistic analysis. Based on this method, the ROC analysis should be performed with the multivariate logistic analysis [
31‐
34] as covariate-adjusted ROC curves. We adopted the same covariates as those used in the stepwise multivariate logistic analysis (creatinine, C-reactive protein, total choline). For the univariate ROC curves, the cut-off value was 28.5 mL/min (Additional file
1: Figure S1d). However, with adjusted covariates, the cut-off value of LDF-Plantar-Qb was 20.0 mL/min (Fig.
4). The other side adjusted ROC curve of SPP-Dorsal Area was unclear with many similar peaks for outcome (Additional file
1: Figure S1e, f). SPP-Dorsal Area value in Fontaine stage III was significantly decreased to 46.7 +/− 23.2 mmHg (Fig.
3b), but difference between stage I and II was 11.1 mmHg with no significance (Additional file
2: Table S3). The ability to detect PAD among Fontaine stage II might be low in SPP compared with LDF.
SPP measurements were divided according to the Fontaine classification. According to a previous study, the SPP threshold for amputation was 40.0 mmHg [
21]. Based on our results, Fontaine group IV had an SPP-Dorsal Area value of 55.1 mmHg. LDF in the plantar area also indicated a similar tendency. Therefore, we suggest that blood flow to the skin in the plantar area of the foot would precisely reflect the blood flow in the lower limb, especially in the peripheral area. Additionally, evaluating the skin blood flow in the plantar area using LDF might provide diagnostic value. If the patient on dialysis had an SPP value of 40.0 mmHg, then it was too late to prevent PAD resection. Fontaine grade II is the first classification involving positive symptoms. Deterioration in blood flow should be predicted during the early stage of PAD and should be diagnosed before the Fontaine classification reaches grade II. LDF-Plantar-Qb indicated a decrease from Fontaine classification grade I. The LDF-Plantar-Qb values were 24.5 ± 18.0 and 17.3 ± 0.1 mL/min for groups II and III, respectively. Therefore, if LDF-Plantar-Qb is less than 20 mL/min, then we should consider initiating therapeutic intervention, including medication (Fig.
4). Impaired endothelial function might be reflected in the PAD group, but this value was consistent for Qb, and PA might be indicated by Qb. Further studies are necessary to assess this association. Overall, there were significant differences in PA between the non-PAD and PAD groups (
p < 0.0019) (Table
2).
The sensitivity of LDF enables the convenient detection of PAD during an early stage in the clinic, but this sensitivity also causes variations in the measurement results. The LDF measurement values are affected by the temperature on any examination day. It is necessary to adjust the temperature in the room throughout the seasons when LDF is conducted. The temperature in our dialysis room was well-regulated at 24 °C, and most of the study was performed during April and May. Regardless of the season during which the study was conducted, patients waited 40 min to 1 h before the examination in a hospital with a well-regulated temperature system. Therefore, variations in the results because of differences in temperature were limited.
In the healthy volunteer group, the average SPP values were 88.4 ± 11.8 and 96.3 ± 16.6 mmHg in the dorsal and plantar areas, respectively. These values are comparable with those reported by Castronuovo [
16]. In the present investigation using healthy controls, SPP-Dorsal Area in the PAD group was relatively higher in the healthy control group than in the non-PAD group (Additional file
1: Figure S1b). In addition, the LDF-Plantar-Qb value of the control group (23.7 ± 20.3 mL/min) was relatively lower than that of the non-PAD group undergoing dialysis (32.7 ± 15.5 mL/min) (Additional file
1: Figure S1c). One reason for this could have been that 53.1% of dialysis patients were treated with calcium antagonists and 27.3% of patients were treated with angiotensin-converting enzyme (ACE) inhibitors/angiotensin receptor blocker (ARB) drugs (Table
1). Debbabi et al. [
35] reported that skin blood flow using a laser Doppler device was increased in patients with hypertension who were treated with ACE inhibitors compared with that in the control group and in patients with hypertension who were treated with other drugs. Our results showed that LDF-Plantar-Qb of dialysis patients in the non-PAD group was increased compared with that of the healthy control group; however, it is unknown whether ACE/ARB drug administration affected this result. Another potential reason is that small capillary shunting in the peripheral skin area would be disrupted by aging, uremia, hypertension, diabetes, dyslipidemia, and atherosclerosis. If the terminal arteriole and postcapillary venule shunt at the peripheral skin area are disrupted by these factors, then blood flow would increase compared with that of a younger, healthy patient. Another reason for speculation is that LDF measurements were performed for dialysis patients, and it was suspected that the systemic body fluid increased more than that in the non-dialysis control group. In terms of the fluid volume, it was reported that skin blood flow was affected by hemodialysis [
15]. According to the development of fluid removal, LDF blood flow in the plantar area of the foot decreased when systolic blood pressure decreased, suggesting that LDF blood flow is reflective of systemic blood flow or fluid volume.
We performed blood flow LDF before dialysis and after starting dialysis (30 min) for the non-PAD group. Results indicated that the LDF-Planter-Qb significantly decreased after the start of dialysis (Additional file
1: Figure S1 g, Additional file
2: Table S1); however, this did not occur in the dorsal area. This suggested that LDF-Plantar-Qb is mainly affected by the total fluid volume. Together, these factors may affect ACE/ARB, aging, and the uremic state. Furthermore, the value resulting from the second examination using LDF (During Dialysis: LDF-Plantar-Qb 22.5 (20.1, 32.0) mL/min (Additional file
1: Figure S1 g) in the before-and-during study would be close to that in healthy controls (23.7 ± 20.3 mL/min) (Additional file
2: Table S3) according to the first 30 min of fluid removal. Therefore, this deviation requires further investigation.
This sensitive system was also affected by the smoking status of those with DM [
35‐
37]. We compared all dialysis patients with or without DM, including those with lower limb ischemia, and obtained values of 32.6 mL/min and 27.1 mL/min for the non-DM and DM groups, respectively. The prevalence of DM was significantly higher in the PAD group (Table
1) because of the disruption of skin micro-shunting. This result suggested that the Qb value was affected by vasoconstriction and microvascular or endothelial dysfunction, also LDF-Plantar-PA was significantly decreased in PAD (+) group suggested that PA might reflect endothelial dysfunction or atherosclerosis.
This study had several limitations. First, this was an observational study, and the PAD and non-PAD groups were not randomly divided. However, these groups were useful for making immediate judgments regarding the diagnosis. Second, a before-and-during study was designed for repeated LDF measurements 1 year after the first examination. However, the season and temperature were matched. Third, the prescription frequency for calcium antagonists was higher for the PAD group, but there was no evidence of a relationship between the frequency of calcium antagonist administration and skin blood flow. Fourth, the inter-dialysis period was inconsistent, with SPP and LDF examinations performed only after 1 or 2 days. Finally, information regarding comorbidities could only be obtained for DM and ESRD; therefore, other comorbidities were adjusted according to medications.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.