Microvascular Imaging as a Novel Tool for the Assessment of Blood Flow Velocity in Patients with Systemic Sclerosis: A Single-Center Feasibility Study

Featured Application

Microvascular Imaging (MVI) is a novel ultrasound-based imaging technique used for areas with low blood flow velocities. This is the first report on the feasibility of MVI in blood flow velocity assessment in systemic sclerosis patients.

Abstract

Systemic sclerosis is an autoimmune disease characterized by organ fibrosis and vasculopathy. Almost all patients suffer from Raynaud’s phenomenon. Nailfold video capillaroscopy is the most widely imaging technique available, but flow quantification is impossible. Therefore, novel imaging techniques are of interest. We performed a single-center feasibility study using Microvascular Imaging (MVI) for flow quantification of small fingertip vessels. We compared 20 healthy controls (HCs) with 20 systemic sclerosis (SSc) patients. In HCs, measurements were, on average, statistically significantly higher when combined for all fingers (median 10.68 vs. 6 cm/s, Δ = 4.68 cm/s, p < 0.0001) and for individual fingers. An optimal cut-off value of peak systolic (PS) velocity of <6.13 cm/s and end-diastolic (ED) velocity of <2.13 cm/s discriminated HCs from SSc. Test characteristics for PS showed excellent sensitivity (0.90, 95% CI 0.70–0.98) and specificity (0.85, 95% CI 0.64–0.95; LR + 6.0). For ED velocity, sensitivity was 0.85 (95% CI 0.64–0.95) and specificity was 0.80 (95% CI 0.58–0.92, LR + 4.25). Here, we present the first study on the use of MVI to assess blood flow in the fingertips with high sensitivity and specificity in SSc. Future studies are needed to investigate correlations with the risk of organ complications, such as digital ulcers or pulmonary arterial hypertension.

1. Introduction

Systemic sclerosis (SSc) is a systemic autoimmune disorder characterized by skin and organ fibrosis and vasculopathy [1]. The latter affects about 90–100% of patients; clinically, this is evident by the presence of Raynaud’s phenomenon (RP) and potential complications, including digital ulcers (DU) [2]. The Raynaud Condition Score (RCS), based on subjective information documented by patients in a diary, has been developed to assess RP in clinical trials [3].

In clinical practice, nailfold video capillaroscopy (NVC) is routinely used to assess morphologic changes to distinguish primary RP from secondary forms and follow SSc patients over time [4]. In NVC, it is, however, impossible to measure quantitative changes in the blood flow of affected fingers.

Raynaud’s phenomenon is widespread in SSc patients, and treatment options, including vasodilating agents with different modes of action, are ineffective in a sizeable number of patients [5]. Several non-invasive methods, including NVC, thermal imaging, and various Doppler ultrasound (US)-based techniques, have been described to assess the microvasculature [6]. However, these are cumbersome to perform, have relatively high costs, or require specialized equipment not readily available at most centers. Therefore, novel methods and applications that are widely available and easy to perform are required to assess functional blood flow changes in affected fingers.

Novel Flow Imaging Techniques

Microvascular Imaging (MVI) is a relatively novel US modality for flow imaging. In contrast to conventional Power Doppler (PD), MVI uses adaptive image analysis to achieve an increased low-velocity blood flow stability that is less dependent on motion artifacts [7]. In addition, MVI generates a high-resolution constant flow mapping of small vessels and related branches via an algorithm-driven suppression of interfering signals [8].

In rheumatology, MVI has been investigated to detect synovitis in patients with inflammatory arthritis, such as rheumatoid arthritis and juvenile idiopathic arthritis. In these diseases, MVI was more sensitive than conventional PD US in preliminary reports [9,10]. Potential advances of MVI may include a better follow-up of patients with minimal disease activity and the prediction or early recognition of disease flares [11]. Nevertheless, more extensive studies are lacking.

In this single-center feasibility trial, we prospectively analyzed a cohort of SSc patients using MVI as a method to detect and quantify digital blood flow compared to HCs.

2. Materials and Methods

2.1. Patient Population

We consecutively recruited a convenience sample of ambulatory or hospitalized SSc patients from our center and healthy volunteers who served as a control group. Exclusion criteria for HCs were a history of primary or secondary RP, peripheral arterial occlusive disease, thromboembolism, diabetes mellitus, or cardiovascular events. All participants provided written informed consent according to the declaration of Helsinki. The ethics committee of the University Medical Center Göttingen, Göttingen, Germany approved the study protocol (protocol number 27/7/20).

2.2. Clinical Data Capture

General patient characteristics, such as age, sex, SSc type (VEDOSS (Very Early Diagnosis of Systemic Sclerosis), limited or diffuse cutaneous) were documented. In addition, we recorded routine laboratory values and SSc-associated antibodies. Finally, we assessed all patients for the presence of SSc-related organ manifestations.

2.3. Ultrasound Technique and Settings

All examinations were performed on a General Electric logiq E10 (GE Healthcare GmbH (Germany), Solingen, Germany) ultrasound machine equipped with a hockey stick probe (frequency 8–18 MHz). During the US scan, the individuals were seated upright and had the supinated, dominant hand on an examination pad in front of their body (Figure 1). The height of this pad was adjusted individually so that the arm was flexed about 90–120° at the elbow joint. The room temperature was constant at about 18–20° Celsius, and no participant had an apparent blood perfusion disturbance (e.g., RP attack) during the measurement.

We performed MVI scans with a frequency of 12.5 MHz in all patients. For flow measurements, automated angle correction was used. For better visualization, Radiantflow™, an advanced visualization technology which adds height and depth to color flow signals, leading to a three-dimensional appearance of blood vessels, was set to the maximum (Figure 2). In addition, a standardized preset was created to provide the same measurement technique for all patients and controls. The assessment included the peak systolic flow (PS), the end-diastolic flow (ED)—both reported in cm/s—and the resistance index (RI) of the examined vessels at the second to fifth fingers (DII–DV) of the dominant hand. Two different blood vessels were investigated on each finger, and the averages of the two measurements were used for the statistical analysis. A video loop demonstrating PD and MVI imaging is shown in the supplementary files S1 (PD) and S2 (MVI).

2.4. Statistics

Demographic data were analyzed using descriptive statistics (median, range, and proportions). For flow velocity analysis, only MVI PS and ED values were used. Comparisons between groups were performed using a Welch’s t-test, Mann–Whitney test, or Chi-square test. Effect sizes are reported with 95% confidence intervals (CI). p-values < 0.05 were considered statistically significant. Agreement between different raters was calculated with intraclass correlation coefficients (ICC).

To establish an optimal cut-off value for each parameter to discriminate between HCs vs. SSc patients, receiver operator characteristics (ROC) curves were created, and the area under the curve (AUC) was calculated. Sensitivity and specificity were reported with 95% CI based on the cut-off point (Youden’s J). Furthermore, the positive likelihood ratios (LR +) were reported.

Associations of PS and ED flow between fingers were analyzed by Pearson’s correlation coefficient. Dependency of measurement variables in patients with SSc was assessed with linear regression, and the variance inflation factor (VIF) was calculated to test for multicollinearity. Values greater than four, which are equivalent to an R² value of 0.75, were considered as evidence of multicollinearity, indicating redundant information.

All data analyses were performed with GraphPad Prism (version 9.2.0 for macOS, GraphPad Software, San Diego, CA, USA) and STATA (version 17.0 for Windows, Stata Corp LLC, College Station, TX, USA).

3. Results

3.1. Baseline Characteristics of Systemic Sclerosis Patients and Healthy Controls

Twenty healthy participants were examined. The median age was 26 years (range 19–56), and 13 (65%) were female. All were Caucasian; 19 (95%) participants were right-handed. All healthy individuals were nonsmokers. Two had arterial hypertension (10%) and were taking antihypertensive medication.

Twenty female SSc patients were included. The median age was 60 years (range 24–79). Nineteen patients (95%) were Caucasian, and one patient was of Asian descent (5%). All SSc patients were right-handed (100%). Figure 3 shows the disposition of patients and HCs.

SSc patients were significantly older than HCs (p < 0.0001) and included more female participants (p = 0.0083). Some of the recruited SSc patients were past (n = 4, 20%) or current (n = 2, 10%) smokers, while all HCs were non-smokers (p = 0.0293). No differences were found regarding ethnicity or dexterity (p > 0.9999). Four (20%) SSc patients had type 2 diabetes mellitus, and three (15%) had chronic kidney disease.

Thirteen SSc patients were categorized as having limited cutaneous SSc (lcSSc) and five with diffuse cutaneous SSc (dcSSc); one patient was diagnosed with Very Early Diagnosis of Systemic Sclerosis (VEDOSS). At the same time, one exhibited no skin involvement (sine scleroderma). Eighteen patients (90%) reported the presence of RP. Nailfold video capillaroscopy was available in 15 patients. Here, an “early pattern” was present in three patients, while ten patients demonstrated an “active pattern” and two patients a “late pattern.” Fourteen patients (70%) took vasoactive medication: ten used calcium-channel blockers (CCB) and seven patients received intravenous iloprost. In addition, a phosphodiesterase 5-inhibitor (PDE5-i) was prescribed in two SSc patients, and one patient was taking an endothelin receptor antagonist (ERA). All patient characteristics are shown in

Table 1. Baseline characteristics of the study population.

	Systemic Sclerosis n = 20	Healthy Controls n = 20	p-Value
Demographic Data
Age; years, median (range)	60 (24–79)	26 (19–56)	<0.0001
Female Sex, n (%)	20 (100%)	13 (65%)	0.0083
Right-handed	20 (100%)	19 (95%)
Disease duration; months, median (range)	50 (1–288)	-	>0.9999
Ethnicity			-
- Caucasian, n (%)	19 (100%)	20 (100%)
- Asian, n (%)	1 (5%)	0 (0%)	>0.9999
- Smoking status
- Current, n (%)	2 (10%)	0 (0%)	0.0293
- Past, n (%)	4 (20%)	0 (0%)
- Never, n (%)	14 (70%)	20 (100%)
Comorbidities
- Diabetes mellitus	4 (20%)	0 (0%)	0.083
- Chronic kidney disease	3 (15%)	0 (0%)
Systemic Sclerosis-related characteristics
SSc type
- VEDOSS, n (%)	1 (6.3%)	-
- Limited, n (%)	13 (68.8%)	-
- Diffuse, n (%)	5(18.8%)	-
- Sine scleroderma, n (%)	1 (6.3%)	-
Antibody status
- Anti-Scl70., n (%)	2/19 (10.5%)	-
- Anti-CENP-B, n (%)	12/19 (63.2%)	-
- Anti-RNA Pol III, n (%)	3/15 (20%)	-
- Anti-Fibrillarin pos., n (%)	2/17 (11.8%%)	-
Organ manifestations
- Raynaud’s phenomenon, n (%)	18 (90%)	-
- ILD, n (%)	6 (30%)	-
- PAH, n (%)	3 (15%)	-
- Gastrointestinal, n (%)	15 (75%)	-
- Musculoskeletal, n (%)	19 (95%)	-
- Renal, n (%)	4 (20%)	-
- Digital ulcers, n (%)	0 (0%)	-
Nailfold video capillaroscopy
- Early pattern, n (%)	3/15 (20%)	-
- Active pattern, n (%)	10/15 (66.6%)	-
- Late pattern, n (%)	2/15 (13.3%)	-
- Normal findings, n (%)	0 (0%)	-
Current treatment
- Prednisolon, n (%)	7/19 (36.8%)	-
- MMF, n (%)	3/19 (15.8%)	-
- MTX, n (%)	2/18 (11.1%)	-
Vasoactive medications
- CCB	10/18 (55.6%)	-
- Iloprost	7/19 (36.9%)	-
- ERA	1/19 (5.2%)	-
- PDE-5i	2/19 (10.5%)	-

CCB, calcium channel blocker; CENP-B, centromere protein B; ERA, endothelin receptor antagonist; ILD, interstitial lung disease; MMF, mycophenolate mofetil; MTX, methotrexate; PAH, pulmonary arterial hypertension; PDE-5i, phosphodiesterase-5 inhibitor; RNA Pol, ribonucleic acid polymerase; SSc, systemic sclerosis; VEDOSS, very early diagnosis of systemic sclerosis. Bold indicates statistically significant p-values.

.

3.2. Comparison of Flow Velocity in Patients with Systemic Sclerosis and Healthy Controls

Microvascular imaging was performed in all SSc patients and HCs on DII-V, respectively. Since SSc patients were significantly older than HCs, we analyzed the PS and ED flow velocities in relation to age (Figure 4). No correlation with age was found, neither in HCs (Figure 4A, r² = 0.03 for PS flow, r² = 0.18 for ED flow) nor in SSc patients (Figure 4B, r² = 0.08 for PS flow, r² = 0.01 for ED flow).

Next, flow velocities of all fingers (Figure 5) and each finger separately were compared between the two groups (Figure 6). It was shown that both PS flow velocity and ED flow velocity were significantly different between the two groups. For all fingers measured, the difference in the medians (Δ) for PS velocity was 4.675 cm/s (95% CI 3.3–5.25, p < 0.0001) and for ED velocity was 1.375 cm/s (95% CI 0.85–1.65, p < 0.0001) (Figure 5).

For the individual fingers, the Δ of medians for PS velocity were 5.225 cm/s (95% CI 2.95–7.5, p < 0.0001) and 2.325 (95% CI 1.1–3.1, p < 0.0001) for ED flow velocity for DII (Figure 6A); for PS flow velocity, 4.975 cm/s (95% CI 1.15–6.15, p < 0.01), and for ED flow velocity, 1.4 cm/s (95% CI 0.15–1.9, p < 0.05) at DIII (Figure 6B); for DIV, Δ of the median for PS velocity was 5.275 cm/s (95% CI 2.5–6.8, p < 0.0001) and 1.05 cm/s (95% CI 0.4–2.4, p < 0.01) for ED flow velocity (Figure 6C). Finally, Δ of the median for PS velocity at DV was 3.325 cm/s (95% CI 2.2–5, p < 0.001) and 0.875 cm/s (95% CI 0.6–1.6, p < 0.001) for ED flow velocity (Figure 6D).

3.3. Interrater Agreement

The interrater agreement was assessed using data from three healthy controls assessed by a total of three raters. The ICC was 0.91 (95% CI 0.80–0.96) for individual absolute agreement and 0.95 (95% CI 0.89–0.98) for average agreement, indicating an excellent agreement between raters.

3.4. Determination of Cut-Off Values to Discriminate Healthy Controls from Systemic Sclerosis

To determine which values of flow velocity are discriminative between HCs and SSc patients, receiver operating characteristics (ROC) curves were created. Different measurements were calculated, which are summarized in

Table 2. Test characteristics and optimal cut-off points for all fingers.

	All Fingers Combined	Minimum	Maximum	Sum of Values
Sens PS (95% CI)	0.69 (0.58–0.78)	0.90 (0.70–0.98)	0.90 (0.70–0.98)	0.85 (0.64–0.95)
Spec PS (95% CI)	0.88 (0.79–0.93)	0.85 (0.64–0.95)	0.70 (0.48–0.85)	0.90 (0.70–0.98)
Sens ED (95% CI)	0.65 (0.54–0.75)	0.85 (0.64–0.95)	0.90 (0.70–0.98)	0.80 (0.58–0.92)
Spec ED (95% CI)	0.80 (0.70–0.87)	0.80 (0.58–0.92)	0.95 (0.76–1.0)	0.85 (0.64–0.95)
AUC PS (95% CI)	0.84 (0.77–0.90)	0.93 (0.85–1.0)	0.85 (0.73–0.96)	0.91 (0.82–1.0)
AUC ED (95% CI)	0.80 (0.73–0.87)	0.90 (0.80–0.99)	0.92 (0.82–1.0)	0.89 (0.79–0.99)
LR+ PS	5.5	6.0	3.0	8.5
LR + ED	3.3	4.24	18.0	5.3
Youden’s J PS	<6.9 cm/s	<6.13 cm/s	<12.28 cm/s	<32.1 cm/s
Youden’s J ED	<2.68 cm/s	<2.13 cm/s	<4.45 cm/s	<12.65 cm/s

CI, confidence interval; ED, end-diastolic; LR+, positive likelihood ratio; PS, peak systolic, Sens, sensitivity; Spec, specificity. Bold indicates the best values obtained.

: Cut-off points and test characteristics were calculated for all fingers combined, each finger, minimum and maximum values, and the sum of measured values. Overall, the best sensitivity, specificity, and LR+ were obtained for the minimum value measured per finger (Figure 7). For PS flow velocity, an optimal cut-off point of <6.13 cm/s showed excellent sensitivity (0.90, 95% CI 0.70–0.98) and specificity (0.85, 95% CI 0.64–0.95), corresponding to an LR+ of 6.0. For ED flow velocity, similar test characteristics were obtained. The optimal cut-off point was estimated at <2.13 cm/s, sensitivity was 0.85 (95% CI 0.64–0.95), and specificity was 0.80 (95% CI 0.58–0.92), corresponding to a LR+ of 4.25.

Of note, for the end-diastolic flow measurements, the best test properties were observed for the maximum value measured (Table 2). Nevertheless, the best overall area under the curve (AUC) was observed for the minimum values of each finger.

3.5. Correlation of Flow Velocity in Different Fingers

To test for the correlation of the measured values in each finger in HCs and SSc patients, we performed a correlation analysis of fingers DII–DV (Figure 8). Correlation between the different fingers was, despite being statistically significant at D II/III in SSc patients and D II/IV and D IV/V in HCs, at best, moderate. The highest correlation in SSc patients for PS and ED flow was observed for the second and third finger (correlation coefficient of 0.58 and 0.65, respectively). In HCs, the highest correlation was observed for PS flow between the second and fourth finger and between the fourth and fifth finger (correlation coefficients of 0.57 and 0.54, respectively).

3.6. Linear Regression of Individual Fingers in Systemic Sclerosis

Lastly, we performed a linear regression and calculated the variance inflation factor (VIF) to test whether one or more fingers could be omitted during the US exam.

Table 3. Variance inflation factor for individual fingers in SSc patients after linear regression.

	D II	D III	D IV	D V
Peak Systolic
VIF	1.69	1.18	1.49	1.09
End-diastolic
VIF	2.58	2.66	1.48	1.54

D, digit; VIF, variance inflation factor.

shows that neither values for PS flow velocity nor ED flow velocity explained the values of other fingers assessed. All R² values with each finger as a dependent variable were below 0.7.

4. Discussion

The present study is the first to investigate digital MVI in healthy subjects and patients with SSc. It was designed as a feasibility study without any prespecified hypotheses. Therefore, no sample size calculation was performed for this study.

The main findings are that findings in peak systolic and end diastolic flow velocity differ between SSc patients and controls. It should be noted that MVI was not used and was never intended to be used as a diagnostic tool to detect SSc, as all patients had a known diagnosis of SSc. One patient was classified as “sine scleroderma”. However, we do not believe that type of skin involvement has a great influence on the microvasculature, since microangiopathy (i.e., Raynaud’s phenomenon) is almost universally present within any type of SSc patient [2].

Since our healthy control group exhibited almost no confounding comorbidities, it was well suited to define normal values of MVI and determine cut-off values to discern them from SSc patients. In addition, although HCs were significantly younger than the SSc patients, age did not have a relevant influence on the digital flow velocities based on our results. To test whether age has a strong influence on the measurements, additional patient cohorts with a more comparable age distribution will be required. In NVC, no correlation of changes with age alone has been shown. Thus, we are confident that microvascular changes are more closely related to the disease rather than age alone. One advantage of the control group is the absence of comorbidities that could have influenced the measurements. Another potential influencing factor was that about 30% of SSc patients were past or current smokers. Nevertheless, we do not think that this has an important effect in the absence of overt digital ulcers or necrosis because there were no outliers in the lower measurement ranges (Figure 6A–D) in the SSc cohort. If smoking was indeed an important influencing variable, we would expect the measurements to be even lower in this subgroup.

Further, SSc patients had a higher use of vasodilating agents, potentially influencing flow measurements. Drugs used in SSc for the purpose of improving blood flow include calcium channel blockers, endothelin receptor antagonists, phosphodiesterase-5 inhibitors, and intravenous prostacyclin [12]. Based on their mechanism of action, it would be expected that the use of these drugs would artificially improve MVI (i.e., increase flow velocities). However, despite using these drugs, SSc patients had consistently and statistically significantly lower values than HCs patients.

Our results further indicate that all four fingers should be assessed, since there was only a weak to moderate correlation between individual fingers. As this was also true for HCs, it seems to be an inherent characteristic of MVI measurements. There were measurements in some fingers that correlated moderately well and yielded statistically significant results. However, the regression models obtained did not provide evidence of multicollinearity, indicating that all fingers need to be examined. Nevertheless, this must be tested in a larger sample. Finally, the MVI with flow measurements performed on four fingers took less than 15 min, which is, in our view, feasible in clinical practice.

Currently, NVC is the standard imaging modality for the assessment of the microvasculature in SSc and has been incorporated into the classification criteria [13]. In NVC, the only parameter that has been consistently associated with organ manifestations and disease progression over time is the loss of capillaries [14]. With NVC it is, however, impossible to directly measure flow velocities; rather, NVC relies on patterns (early, active, late) [4]. While this study did not have the purpose of testing the diagnostic properties of MVI in SSc, it has to be regarded as complementary to NVC. Further, longitudinal data are required for an association of MVI (or changes thereof) with organ manifestations or disease progression.

Overall, due to the excellent visualization of microvascular tissue and organ perfusion, MVI has the potential to avoid invasive or radiation-assisted examinations. Still, more extensive studies in rheumatic conditions are not available. In our experience, visualization with MVI was better suited for superficial blood vessels than conventional PD US. Some examination modalities, such as thermography and others, are already available for SSc [6]. However, these are not available everywhere; some are cumbersome to perform, expensive, or require additional equipment. The advantage of MVI is that rheumatologists are already used to performing US examinations and would need only an additional software application to use MVI.

Our study has several limitations: The study cohort included relatively few individuals. However, it included a representative sample of SSc patients. Currently, we cannot ascertain if there are differences between different diseases, such as mixed connective tissue disease or systemic lupus erythematosus, which frequently show RP. We will address this in a further analysis of the method. Finally, based on our results, we cannot recommend estimating flow velocity in less than all four fingers. This is in line with what is frequently found in NVC, where different findings are also present in different fingers of the same individual. Nevertheless, the interrater agreement in healthy control patients was excellent.

The strengths of our study are the feasibility of MVI in clinical practice, the relatively short examination time required, and the novelty of the presented data. Thus, MVI may offer potential applications in assessing microvascular alterations in patients with inflammatory rheumatic diseases, such as SSc and others.

5. Conclusions

We present the first study on the use of MVI as a novel imaging technique to measure blood flow velocity in patients with SSc. For the first time, we report the method’s feasibility and cut-off points to discriminate healthy controls from SSc patients. Whether the results of MVI correlate with organ manifestations or vascular complications of SSc needs to be tested prospectively in a larger cohort. In addition, MVI should be performed in other rheumatic diseases with and without RP to investigate whether the findings presented here are specific for SSc or not.