- Email: [email protected]
Shear Wave Elastography (SWE) for the Evaluation of Patients with Plantar Fasciitis
ARTICLE IN PRESS
Original Investigation
Shear Wave Elastography (SWE) for the Evaluation of Patients with Plantar Fasciitis Matthias Gatz, MD, Ljudmila Bejder, MD, Valentin Quack, MD, Simone Schrading, MD, Timm Dirrichs, MD, Markus Tingart, MD, Christiane Kuhl, MD, Marcel Betsch, MD
Abbreviations ASE axial strain elastography SWE shear wave elastography PF plantar fascia PFis plantar fasiitis B-US B-Mode ROI region of interest
Rationale and Objectives: The current imaging standard for diagnosing plantar fasciitis is B-Mode ultrasound (B-US). The aim of this study was to determine the diagnostic potential of Shear Wave Elastography (SWE) and the correlation of clinical scores to elastographic parameters. Materials and Methods: Diagnostic case-control study with n = 82 plantar fascia (PF). PF were divided into three subgroups: (1) symptomatic PF (n = 39); (2) control group of unilateral asymptomatic PF (n = 23); (3) bilateral asymptomatic PF (n = 20). Reference standard for positive findings in B-US was a PF thickness greater than 4 mm. For SWE tissue elasticity (Young’s modulus kPa; shear wave speed m/ s) was measured at Location 1: directly at the calcaneus; Location 2: +1 cm distal of the calcaneus and Location 3: central part of the calcaneus. Sensitivity, specificity, and diagnostic accuracy as well as correlation to American Orthopaedic Foot and Ankle Score (AOFAS) and Food Functional Index (FFI) were determined. Results: Symptomatic PF are thicker (4.2 mm, n = 39) than asymptomatic (3.0 mm, n = 43) (p < 0.001). Thickness of the PF (n = 82) correlated poorly to clinical scores (p = 0.001): FFI-pain (r = 0.349); FFI-function (r = 0.381); AOFAS (r = ¡0.387). Cut-off point for positive SWE finding was 51.5 kPa (4.14 m/s). Symptomatic PF (31.9 kPa, 3.26 m/s, n = 39) differ significantly from asymptomatic PF (93.3 kPa, 5.58 m/s, n = 43) with significant differences at L1 between all groups (p < 0.001). Correlation between Young’s modulus (n = 82) and clinical scores was strong (p < 0.001): FFI-pain (r = ¡0.595); FFI-function (r = ¡0.567); AOFAS (r = 0.623,). B-US: sensitivity (61%), specificity (95%); SWE sensitivity (85%), specificity (83%). The combination of SWE and B-US increases the sensitivity (100%) with a diagnostic accuracy of 90%. Conclusion: Based on our results, we could show that SWE can improve the diagnostic accuracy in patients with plantar fasciitis compared to B-US. Level of evidence: II Key Words: Shear wave elastography; Plantar fasciitis; AOFAS Score; FFI; ultrasound. © 2019 Published by Elsevier Inc. on behalf of The Association of University Radiologists.
INTRODUCTION
P
lantar fasciitis (PFis) is a common cause for heel pain and in the United States alone almost 1 million patients per year require treatment (1). Imaging, eg with ultrasound, is useful in unclear cases to verify the
Acad Radiol 2019; &:1–8 From the Department of Orthopedics, University Hospital RWTH Aachen, Pauwelsstraße 30, Aachen 52074, Germany (M.G., L.B., V.Q., M.T., M.B.); Department of Diagnostic and Interventional Radiology, University Hospital RWTH Aachen, Pauwelsstraße 30, Aachen 52074, Germany (S.S., T.D., C.K.). Received March 3, 2019; revised April 8, 2019; accepted April 30, 2019. Declarations of interest: none Address correspondence to: M.G. e-mail: [email protected] © 2019 Published by Elsevier Inc. on behalf of The Association of University Radiologists. https://doi.org/10.1016/j.acra.2019.04.009
diagnosis (2). Contemporary ultrasound imaging of plantar fasciitis (PFis) is mainly based on gray-scale imaging (B-US) finding hypoechoic areas, border irregularities of the plantar fascia (PF) and calcifications (3,4). Current studies showed that PF thickness of greater than 4 mm is the most important diagnostic parameter for diagnosing PFis using conventional B-US (3,4). However, recently Hansen et al found that the PF of patients with PFis can also decrease in thickness over time (5) and that even after a follow-up of more than 10 years hypoechoic and thickened areas in the PF can still be found in 75% of asymptomatic individuals (5). Conventional B-US elastographic modalities, such as axial strain elastography (ASE) and shear wave elastography (SWE), have been evaluated in the diagnosis of PFis in the
1
ARTICLE IN PRESS GATZ ET AL
recent years. ASE is based on the external compression of the tissue by a probe, and therefore it offers only semi-quantitative and user-dependent information of tissue elasticity (6 8). In contrast, using SWE an impulse is generated by a probe and elastographic parameters are quantitatively measured in Kilopascal (kPa, m/s) as an elastic module (Young’s modulus, kPa) or as shear wave speed progression (m/s) (6 8). Therefore, the method of SWE is considered to be user-independent. Studies using ASE showed that symptomatic PF are “softer” than asymptomatic fascia and that older individuals might have a “softer” PF than younger individuals (9). Furthermore, ASE improves the diagnostic accuracy to diagnose symptomatic PF, in cases where no morphologic changes on B-US were found (10,11,12). Zhang et al found “softer” PF in symptomatic patients, at the site of the pain, while the stiffness of the fascia increased 1 cm distal of its insertion (13). Besides diagnosing PFis, elastography might be more helpful to monitor treatment effects than conventional B-US. Using ASE, Kim et al found no significant reduction in PF thickness, hypoechogenicity, or perifascial oedema three months after a collagen injection. However, the stiffness of the fascia increased, which might be caused by a slight reduction of the edema in the PF, that cannot be detected using B-US (14). In summary, ultrasound elastography seems to be beneficial in diagnosing and monitoring PFis, however most current studies have used ASE rather than SWE. Therefore, aim of this prospective diagnostic study was to comparatively analyze fascia stiffness and thickness using B-US and SWE in asymptomatic and symptomatic individuals. Furthermore, purpose of this study was to determine the sensitivity, specificity, and diagnostic accuracy of B-Mode, SWE and the combination of both modalities in the diagnosis of PFis. METHODS Study Design
This is a diagnostic case-control study using SWE and B-US to examine PFis. The actual status of symptoms was evaluated by the validated Foot Functional Index (FFI) Pain and Function Scale as well as by the American Orthopaedic Foot and Ankle Score (AOFAS) (15,16). The study was approved by the local ethics committee (EK 060/17) and all participants provided written informed consent.
Academic Radiology, Vol &, No &&, && 2019
big toe (Windlass Test). In the clinical history there was no suspicion for differential diagnosis eg stress fractures or nerve compression syndromes. If available, previous image modalities such as X-ray were reevaluated. There was no minimum symptom duration for inclusion in the trial, but FFI- Pain and AOFAS-Scores were required to be over 10 and under 90, respectively. Individuals with a rheumatic disease, overand underweight (BMI (kg/m2): <17; >34), previous injection therapy in the last six months, previous surgery and trauma as well as pregnancy were excluded from the study. According to the participants symptoms, we divided them into three subgroups (Fig 1): Group 1: Symptomatic individuals suffering from a clinically diagnosed unilateral or bilateral PFis. Group 2: Asymptomatic PF of individuals with unilateral PFis from group 1 with no heel pain or discomfort at any time prior to the study. Group 3: Asymptomatic individuals with no pain or discomfort in both PF at any time and the respective score points (FFI: zero; AOFAS: 100). Data Acquisition with B-US, SWE
Every participant underwent a standardized multi-modal ultrasound protocol consisting of B-US and SWE of both PF. Measurements were conducted using an ultrasound system (Aixplorer, Supersonic Imagine, Aix-en-Provence, France) with a high-resolution linear 18 MHz transducer, (SuperLinear SL 18-4, Supersonic Imagine, Aix-en-Provence, France) in longitudinal planes, carefully avoiding over-compression of the PF. All individuals underwent a standardized examination protocol, always starting with the left foot and conducting B-US before SWE. US examinations were always performed in the same air-conditioned room with a standardized temperature of 20°C. Individuals were lying in the prone position with the foot hanging relaxed over the examination table. The measured foot was kept in the neutral position without any active or passive dorsiflexion of the big toe, which might have preloaded the PF similar to the Windlass Test. Nevertheless, a recent study using SWE demonstrated, that preloading does not influence measurements of the PF directly at the insertion site at the calcaneus (17). All examinations were performed by one physician with four years of experience in ultrasound and shear wave elastography, who was blinded to the clinical symptoms of the study participants.
Inclusion and Exclusion Criteria
Individuals were recruited from offices of primary care physicians from 07/2017 until 03/2018. All individuals have been previously diagnosed and treated for PFis (Secondary Health Care) and were referred by the treating clinicians. We included participants older than 18 years with a clinical diagnosis of PFis. During clinical examination, by an experienced orthopedic surgeon, all individuals had pain at the insertion of the PF, which increased by dorsiflexion of the 2
B-US Measurements and PF Thickness>4 mm as Reference Standard
According to previous studies, PF thickness over 4 mm was considered a validated parameter for an acute state of PFis, and the reference standard for sonographic diagnosis (3,4). Therefore, a PF thickness greater than 4 mm was regarded as a positive finding for PFis, while hypoechogenicity and loss of PF borders were only noted, but not considered for
ARTICLE IN PRESS Academic Radiology, Vol &, No &&, && 2019
EVALUATION OF PATIENTS WITH PLANTAR FASCIITIS
Figure 1. Study design and overview of the groups. Based on symptomatic and asymptomatic participants three groups were formed. (Color version of figure is available online.)
calculating sensitivity or specificity. The measurements of the PF were conducted strictly vertically in a longitudinal plane over the calcaneus at the thickest part of the fascia, which is mainly at the medial tubercle of the calcaneus.
SWE-measurements
The size of the SWE-measurement window was 2 cm2. Quantitative SWE-measurements were performed by placing
Figure 2. Acquisition of Youngs moduli at three different locations within the PF. L1: directly at the calcaneus, L2: 1 cm distal from the calcaneus, L3: central part of the PF. All images demonstrate the semi-quantitative SWE color mapping, while quantitative values were measured in the Q-Box (white circle). Blue color demonstrates areas with lower Youngs modulus while red depict areas with the highest Youngs Modulus (blue
3
ARTICLE IN PRESS GATZ ET AL
a region of interest of 2-mm-diameter in the most rigid part of the PF in the area of the greatest PF thickness. The acquired SWE-information was evaluated quantitatively as Young’s modulus (kPa) and shear wave speed (m/s). Mean value and standard deviations of the measurements were calculated. Every PF was systematically analyzed in the longitudinal plane at three locations: L1: insertion directly at the calcaneus, L2: +1 cm distally from the calcaneus insertion and L3: at the central part of the PF at least 2 cm from the calcaneal insertion (Fig 2). Representative SWE-measurements were performed three times in each location (= nine images in total per PF) and then were averaged. Since there is no cut-off value in the current literature for diagnosing PFis, a ROC analysis was used to establish a cut-off value for measurements at the calcaneus. The “SWE-penetration mode” has been used as a setting to reduce so called “signal void areas,” which are parts without any SWE-information (18). Nevertheless, the SWE measurement window does not always allow to display a complete SWE-signal color map of the area of the PF or the subcutaneous fat pad. In these cases, the quality and completeness of the color map and thus SWE acquisition was rated in four grades based on the amount of colored area within the PF and the subcutaneous fat pad: grade 1: 100%, grade 2: 99% 75%, grade 3:74% 50%, and grade 4: 50% (Fig 3). Statistical Analysis
For all analyses, SPSS Statistics, version 24.0 (IBM Corp.) was used. SWE-values were analysed with descriptive statistics. Additionally, normal distribution of the data was tested by the Shapiro Wilk-test. To examine for significant differences between the three independent groups, a Kruskal Wallis test was used with a posthoc Dunn Bonferroni test. A Wilcoxon
Academic Radiology, Vol &, No &&, && 2019
signed-rank test analysed intraindividual differences of group 2. The correlation between SWE-values and the clinical scores or age was assessed using Pearson’s correlation test. The level of significance was set at p < 0.05. Effect size was calculated with G*Power 3.1.9.4 (D€ usseldorf, Germany) (19). RESULTS In total, 82 PF have been included in this study. They consisted of 31 patients with a total of 39 symptomatic PF (23 unilateral, 16 bilateral), 23 asymptomatic PF of patients with unilateral PFis. Furthermore, 10 bilateral asymptomatic individuals with 20 bilateral PF of a healthy control have been included. G*Power revealed an effect size of 0.44. Figure 1 provides an overview of the different groups and depicts the group constellations in detail. Table 1 demonstrates the demographic data as well as their clinical scores: patients in group 1 (age: p = 0.016; BMI: p = 0.010) and group 2 (age: p = 0.009, BMI: p = 0.016) were significantly older and with a higher BMI, than patients in group 3. However, there was no difference for hours of sport activities between the groups. In symptomatic patients, we found a moderate and significant correlation between BMI and the FFI-function score, as well as pain measured by the FFI-pain score and for the AOFAS score: FFI pain (r = 0.534, p = 0.002, n = 31); FFI function (r = 0.511, p = 0.003, n = 31); AOFAS (¡0.457, p = 0.010, n = 31). However, there was no correlation between age and the respective clinical scores (FFI-pain r = 0.132, p = 0.480; FFI-function r = 0.051, p = 0.786; AOFAS r = ¡0.046, p = 0.806, n = 31). B-US Findings
Figure 4 shows the thickness of the PF at the calcaneus in all three groups. Symptomatic PF of group 1 are on average
Figure 3. Due to “signal void areas” the quality of the received SWE signal was rated in four grades depending on the area of the PF which was without signal: a) grade 1: 100% b) grade 2: 99% 75% c) grade 3: 74% 50% d) grade 4: <50%. (Color version of figure is available online.)
4
ARTICLE IN PRESS Academic Radiology, Vol &, No &&, && 2019
EVALUATION OF PATIENTS WITH PLANTAR FASCIITIS
TABLE 1. Demographics of the Different Groups
Age (a) min max Sex BMI (kg/m2) min max Sport (h/week) min max Symptom duration (month) min max FFI Pain min max FFI Functional min max 3AOFAS min max Therapy - Orthotics - Analgesics - Physiotherapy - Kryotherapy - Deep x-ray therapy
Significant Difference (Kruskal Wallis)
Group 1
Group 2
Group 3
48.9 (§9.6) (26 66) Male:8 female: 23 26.7 (§4.6) (21 34) 3.8 (§3.6) (0 12.5) 14.3 (§16.7) (2 84) 36.4 (§17.7) (11 75) 31.6 (§17.9) (12 69) 74 (§11) (47 85)
50.0 (§9.5) (33 66) Male: 8 female:15 26.6 (§4.1) (21 38) 3.8 (§3.9) (0 12.5) 0
30.4 (§11.1) (21 60) Male:5 female:5 22.8 (§1.0) (21 24) 2.1 (§2.1) (0 6.0) 0
1 vs. 2/3 (p < 0.001)
0
0
1 vs. 2/3 (p < 0.001)
0
0
1 vs. 2/3 (p < 0.001)
100
100
1 vs. 2/3 (p < 0.001)
1 vs. 3 (p = 0.016) 2 vs. 3 (p = 0.009)
3 vs. 1 (p = 0.010) 3 vs. 2 (p = 0.016) 1 vs. 2 vs. 3. (p = 0.726)
12 9 10 6 4
4.2 mm (§1.2) thick, while asymptomatic PF of group 3 have a PF thickness of only 2.8 mm (§0.5). Further detailed information is shown in Table 2. The Kruskal Wallis test revealed, that there is a significant difference between group 1 and group 3 (p < 0.001) as well as group 1 and group 2 (p = 0.030), showing that symptomatic PF are thicker than asymptomatic ones. Moreover, thickness is significantly (p < 0.001) reduced in all asymptomatic PF (group 2/3: 3.0 mm (§0.6)) compared to symptomatic PF (group 1: 4.2 mm (§1.2)). Thickness of the PF measured by B-US correlated to clinical scores (FFI-pain, FFI-function, AOFAS as well as to age, but not to BMI (Table 3). SWE Findings
The ROC analysis of 82 PF showed x< 51.5 kPa (4.14 m/s) as an ideal cut-off value to diagnose PFis using SWE. Table 2 and Figure 4 show Young’s moduli in every group and every location measured within the PF. In all three measured locations (L1 L3), group 3 has significantly higher elastic Young’s moduli than both other cohorts (p < 0.001 p = 0.002, Table 2). In particular, at L1, SWE elastic moduli differed significantly between all groups (Table 2). At L1 symptomatic PF (31.9 kPa, 3.26 m/s, n = 39; L1; group 1) had significant (p < 0.001) lower Young’s moduli than asymptomatic PF in a pooled analysis of group 2, 3 (93.3 kPa, 5.56 m/s, n = 43). Moreover, a Wilcoxon signed rank test analyzing the intraindividual differences of patients with unilateral symptoms (group 2) revealed a
significant difference between the symptomatic (18.5 kPa, 2.48 m/s) and asymptomatic (36.1 kPa, 3.47 m/s) side (z = ¡1,964, p = 0.05, n = 23). We found a moderate but significant correlation between the Young’s moduli measured by SWE (L1) and clinical scores (FFI-pain, FFI-function, AOFAS) as well as age and BMI (Table 3). Additionally, there was a correlation between SWE values and the thickness measured in B-US (r = ¡0.421, p < 0.001, n = 82). In total, analysing the quantity of the received SWE signal of asymptomatic PF (group 3) had a mean grade of 1.7 (§0.47) at L1, 1.7 (§0.47) at L2 and grade 1 (§0) at L3. In symptomatic patients PF (group 1) the quality of the SWE signal was significantly lower compared to all asymptomatic patients (group 2/3): at L1 grade for signal mapping was 2.4 (§0.9) (U = 501, p = 0.001); at L2 mean grade was 2.4 (§1.1) (U = 546, p = 0.009); and at L3 1.8 (§1.1) (U = 642, p = 0.056).
Sensitivity, Specificity, and Diagnostic Accuracy
In this study, B-US alone had the lowest sensitivity for diagnosing PFis with 61% and a specificity of 95%, while SWE alone was more sensitive (85%) but less specific (83%). Nevertheless, the combination of both modalities, SWE and B-US, increased the sensitivity to 100% and the diagnostic accuracy to 90%. A comparison of diagnostic accuracy of the ultrasound techniques is given in Table 4. 5
ARTICLE IN PRESS GATZ ET AL
Academic Radiology, Vol &, No &&, && 2019
DISCUSSION
Figure 4. PF thickness (mm) in B-US and Youngs moduli (kPa) in SWE. (Color version of figure is available online.)
6
The present study is the first, which determines the diagnostic accuracy of SWE as a tool for the diagnosis of PFis in comparison to B-US. Furthermore, we evaluated the correlation of PF thickness (cm) measured in B-US and Young’s moduli (kPa) to clinical scores (FFI-Pain, FFI-Function, AOFAS), which has not been investigated yet in the current literature. Some of the results presented in our study are in line with previous studies using ASE and SWE: Symptomatic PF have significant lower Young’s moduli than asymptomatic ones (31.9 kPa vs. 93.3 kPa; 3.26 m/s vs. 5.56 m/s; p < 0.001), this means that they are “softer” or “less elastic.” Furthermore, PF do not have only lower Young’s modulus values when they are symptomatic, but these values also decrease with age (r = ¡0.352, p = 0.001). Three studies using ASE confirmed the reduction of elasticity due to symptoms or age (9,12,20). Zhang et al could demonstrate that these findings shall apply to SWE as well (13). At the calcaneal insertion (L1) symptomatic PF and PF of older participants have significant lower Young’s elastic moduli (13). Furthermore, elastic moduli increase, when measured +1 cm from the calcaneal insertion (13). Our findings showed that in every group Young’s moduli increase the further away we measured from the calcaneus. However, there was only a significant difference found between L1 vs. L3 and L2 vs. L3 in groups 1 and 3 (Table 2). The optimum location for measuring the PF is therefore directly at the location of symptoms (L1), where SWE elastic modulus differ significantly between all groups except group 2 vs. group 3. So far, the diagnostic accuracy for diagnosing PFis has only been evaluated using ASE. By combining B-US and ASE the diagnostic accuracy was, increased from 90.0% to 95.4% (10). The present study showed for the first time, that SWE has an additive diagnostic value for diagnosing PFis with a sensitivity of 100% for the combined usage of SWE and B-US. Additionally, the diagnostic accuracy increased from 79% using BUS to 84% using SWE. The combination of both modalities has a diagnostic accuracy of 90%. Compared to the present study, Sconfienza et al reported higher diagnostic accuracies for B-US (90%) and ASE (95.4%) (10), which might be caused by differences between the study cohorts, a different elastographic setting and the fact that hypoechoic area and blurring of fascia borders were also taken into account. In contrast, our present study only evaluated quantitative parameters (mm, kPa). In previous studies, parameters of elastography have not been correlated with clinical scores. The results of our study showed a strong correlation between Young’s moduli and clinical scores (r = 0.6; p < 0.001), which was higher than the correlation between B-US and clinical scores (r = 0.35; p < 0.001). Our study has a few limitations according to the used study design and methods. Participants in group 3 were younger and lighter than patients in group 1 and 2. Nevertheless, there was no significant age difference between group 1 and 2,
ARTICLE IN PRESS Academic Radiology, Vol &, No &&, && 2019
EVALUATION OF PATIENTS WITH PLANTAR FASCIITIS
TABLE 2. Imaging Findings of SWE and B-US in Comparison to Different Groups and Locations of Measurement Youngs Modulus (kPa) Shear Wave Speed (m/s)
Group 1
Group 2
Group 3
Sign. Difference (Kruskal Wallis) 1 vs. 2 p = 0.006 1 vs. 3 p< 0.001 2 vs. 3 p = 0.002 1 vs. 2 p = 0.167 1 vs. 3 p< 0.001 2 vs. 3 p< 0.001 1 vs. 3 p = 0.362 1 vs. 3 p< 0.001 2 vs. 3 p< 0.001
L1: Calcaneus
Mean (§SD)
31.9 kPa (§16.2) 3.26 m/s
65.3 kPa (§42.4) 4.66 m/s
124.1 kPa (+/-28.5) 6.43 m/s
L2: Calcaneus +1cm
Mean (§SD)
46.3 kPa (§37.5) 3.93 m/s
75.3 kPa (§59.2) 5.01 m/s
158,4 kPa (§55.7) 7.26 m/s
L3: Central part
Mean (§SD)
88.9 kPa (§66.5) 5.46 m/s
109.1 kPa (§72.5) 6.03 m/s
230.6 kPa (§25.9) 8.76 m/s
L3 vs. L1 p< 0.001 L3 vs. L2 p = 0.023
¡
L3 vs. L1 p = 0.001 L3 vs. L2 p = 0.001
Sign. difference (Kruskal¡Wallis)
Group 1
Group 2
Group 3
Sign. Difference (p < 0.05)
Hypoechogenicity
Mean (§SD) median Positive findings
4.2 mm (§1,2) 4.6 mm 35/39
3.19 mm (§0.7) 3.1 mm 12/23
2.8 mm (§0.5) 2.6 mm 1/20
Border irregularities
Positive findings
32/39
5/23
0/20
1 vs. 3 p < 0.001 1 vs. 2 p = 0.030 1 vs. 3 p < 0.001 1 vs. 2 p = 0.001 1 vs. 3 p < 0.001 1 vs. 2 p < 0.001
B-US: thickness of PF (mm) Calcaneus
with group 2 being an asymptomatic control having higher SWE-values than group 1. Additionally, we scattered the influence of age and BMI, by an intraindividual comparison of group 2 showing that symptomatic PF have significantly lower Young’s modulus than asymptomatic ones (p = 0.05). A further limitation is that, all individuals in group 1 received different treatments before, which might be a potential confounding factor influencing the Young’s modulus. Dirrichs et al showed that excessive load in semi-professional runners might lead to higher Young’s modulus in Achilles tendons (21). Therefore, physical therapy before recruitment might have influenced elasticity of PF. Furthermore, all image acquisitions were conducted by one physician, thus it was not possible to evaluate interclass observer correlation. Additionally, in the current literature there are no standard values for symptomatic or asymptomatic PF, so that other studies might reveal other cut-off values than ours. Furthermore, the present study is a pilot study with a small sample size being more a case-control study than a diagnostic study and further
studies need to revaluate the results with larger cohorts. Using 7% as the general prevalence in individuals over 65 years (2) and a change of sensitivity from 60% (B-US) to 80% (SWE) a larger control group would have been necessary (22). Nevertheless, our study compares to sample sizes of previous investigations using ASE (10,11,13). To our knowledge, previous studies using ASE or SWE have not reported difficulties in receiving a full elastographic signal in the ROI. However, in our study measurements at L1 in group 1, which is actually the most relevant group and location for measurements, we found that in up to 25% of the PF no SWE signal could be measured. As this was not seen that strong in asymptomatic PF (group 3; mean grade 1.7) or at measurements at locations without symptoms (L3 symptomatic: 1.8; L3 asymptomatic: one) inadequate acquisition of SWE signal might be due to tissue remodeling at the insertion of PF due to matrix calcification, chondroid metaplasia or collagen necrosis being common findings of histopathological adaptation (23). Bouchet et al stated in their recent review that chondroid areas as well
TABLE 3. Correlation in B-US and SWE to Clinical Scores and Demographic Parameters of the Whole Study Population (n = 82)
B-US thickness (group 1,2,3 n = 82) SWE Youngs modulus (L1) (group 1,2,3 n = 82)
FFI-pain
FFI-function
AOFAS
Age
BMI
r = 0.349
r = 0.381
r = ¡0.387
r = 0.293
r = 0.192
p = 0.001 r = ¡0.595
p < 0.001 r = ¡0.567
p < 0.001 r = 0.623
p = 0.008 r = ¡0.442
p = 0.085 r = ¡0.485
p< 0.001
p < 0.001
p < 0.001
p < 0.001
p < 0.001
7
ARTICLE IN PRESS Academic Radiology, Vol &, No &&, && 2019
GATZ ET AL
TABLE 4. Sensitivity, Specificity and Diagnostic Accuracy of B-Mode, SWE and the Combination of SWE+ B-Mode B -Mode Calcaneus
True + False + Sensitivity Pos. predictive value Diagnostic accuracy
True ¡ False¡ Specificity Neg. predictive value
as intralesional necrosis can be “signal void areas,” which are not coded on the color map (18). Therefore, in very severe cases of PF with a large area of tissue remodeling the informative value of SWE might be diminished. Moreover, micro calcification being not that apparent in B-US could have influenced SWE measurements in symptomatic PF (group 1) and lead to reduced Young’s modulus (24). The results of this study showed that SWE (84%) improves the diagnostic accuracy compared to B-US (79%) and that the combination of SWE and B-US (90%) is a strong diagnostic tool. SWE might be the method of choice for providing information, which correlate stronger to clinical scores than B-US. SWE provides a quantitative assessment of PF integrity and can distinguish between symptomatic and asymptomatic patients better than B-US. Therefore, further studies need to reevaluate if B-US is still the gold standard in the diagnosis of PFis.
REFERENCES 1.
2. 3.
4.
5.
6. 7.
8
Riddle DL, Schappert SM. Volume of ambulatory care visits and patterns of care for patients diagnosed with plantar fasciitis: a national study of medical doctors. Foot Ankle Int 2004; 25:303–310. Goff JD, Crawford R. Diagnosis and treatment of plantar fasciitis. Am Fam Physician 2011; 84:676–682. McMillan AM, Landorf KB, Barrett JT, et al. Diagnostic imaging for chronic plantar heel pain: a systematic review and meta-analysis. J Foot Ankle Res 2009; 2:32. Mohseni-Bandpei MA, Nakhaee M, Mousavi ME, et al. Application of ultrasound in the assessment of plantar fascia in patients with plantar fasciitis: a systematic review. Ultrasound Med Biol 2014; 40:1737–1754. Hansen L, Krogh TP, Ellingsen T, et al. Long-term prognosis of plantar fasciitis: A 5- to 15-year follow-up study of 174 patients with ultrasound examination. Orthop J Sports Med 2018; 6:2325967118757983. Ryu J, Jeong WK. Current status of musculoskeletal application of shear wave elastography. Ultrasonography 2017; 36:185–197. Taljanovic MS, Gimber LH, Becker GW, et al. Shear-wave elastography: basic physics and musculoskeletal applications. Radiographics 2017; 37:855–870.
24 2 61% 0.92 79%
41 15 95% 0.73
SWE 33 7 85% 0.83 84%
SWE+ B-Mode 36 6 83% 0.86
39 8 100% 0.83 90%
35 0 81% 1.00
8. Fusini F, Langella F, Busilacchi A, et al. Real-time sonoelastography: principles and clinical applications in tendon disorders. A systematic review. Muscles Ligaments Tendons J. 2017; 7:467–477. 9. Wu CH, Chang KV, Mio S, et al. Sonoelastography of the plantar fascia. Radiology 2011; 259:502–507. 10. Sconfienza LM, Silvestri E, Orlandi D, et al. Real-time sonoelastography of the plantar fascia: comparison between patients with plantar fasciitis and healthy control subjects. Radiology 2013; 267:195–200. 11. Wu CH, Chen WS, Wang TG. Plantar fascia softening in plantar fasciitis with normal B-mode sonography. Skeletal Radiol 2015; 44:1603–1607. 12. Lee SY, Park HJ, Kwag HJ, et al. Ultrasound elastography in the early diagnosis of plantar fasciitis. Clin Imaging 2014; 38:715–718. 13. Zhang L, Wan W, Zhang L, et al. [Assessment of plantar fasciitis using shear wave elastography]. Nan Fang Yi Ke Da Xue Xue Bao 2014; 34:206–209. 14. Kim M, Choi YS, You MW, et al. Sonoelastography in the evaluation of plantar fasciitis treatment: 3-Month follow-up after collagen injection. Ultrasound Q 2016; 32:327–332. 15. Kostuj T, Krummenauer F, Schaper K, et al. Analysis of agreement between the German translation of the American Foot and Ankle Society's Ankle and Hindfoot Scale (AOFAS-AHS) and the Foot Function Index in its validated German translation by Naal et al. (FFI-D). Arch Orthop Trauma Surg 2014; 134:1205–1210. 16. Naal FD, Impellizzeri FM, Huber M, et al. Cross-cultural adaptation and validation of the Foot Function Index for use in German-speaking patients with foot complaints. Foot Ankle Int 2008; 29:1222–1228. 17. Chino K, Lacourpaille L, Sasahara J, et al. Effect of toe dorsiflexion on the regional distribution of plantar fascia shear wave velocity. Clin Biomech 2018; 61:11–15. 18. Bouchet P, Gennisson JL, Podda A, et al. Artifacts and technical restrictions in 2D shear wave elastography. Ultraschall Med 2018. 19. Faul F, Erdfelder E, Lang AG, et al. G*Power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Methods 2007; 39:175–191. 20. Rios-Diaz J, Martinez-Paya JJ, del Bano-Aledo ME, et al. Sonoelastography of plantar fascia: reproducibility and pattern description in healthy subjects and symptomatic subjects. Ultrasound Med Biol 2015; 41:2605–2613. 21. Dirrichs T, Schrading S, Gatz M, et al. Shear Wave Elastography (SWE) of asymptomatic achilles tendons: a comparison between semiprofessional athletes and the nonathletic general population. Acad Radiol 2019. 22. Bujang MA, Adnan TH. Requirements for minimum sample size for sensitivity and specificity analysis. J Clin Diagn Res 2016; 10: YE01-YE6. 23. Snider MP, Clancy WG, McBeath AA. Plantar fascia release for chronic plantar fasciitis in runners. Am J Sports Med 1983; 11:215–219. 24. Gregory A, Bayat M, Denis M, et al. An experimental phantom study on the effect of calcifications on ultrasound shear wave elastography. Conf Proc IEEE Eng Med Biol Soc 2015: 3843–3846. 2015.
Original Investigation
Shear Wave Elastography (SWE) for the Evaluation of Patients with Plantar Fasciitis Matthias Gatz, MD, Ljudmila Bejder, MD, Valentin Quack, MD, Simone Schrading, MD, Timm Dirrichs, MD, Markus Tingart, MD, Christiane Kuhl, MD, Marcel Betsch, MD
Abbreviations ASE axial strain elastography SWE shear wave elastography PF plantar fascia PFis plantar fasiitis B-US B-Mode ROI region of interest
Rationale and Objectives: The current imaging standard for diagnosing plantar fasciitis is B-Mode ultrasound (B-US). The aim of this study was to determine the diagnostic potential of Shear Wave Elastography (SWE) and the correlation of clinical scores to elastographic parameters. Materials and Methods: Diagnostic case-control study with n = 82 plantar fascia (PF). PF were divided into three subgroups: (1) symptomatic PF (n = 39); (2) control group of unilateral asymptomatic PF (n = 23); (3) bilateral asymptomatic PF (n = 20). Reference standard for positive findings in B-US was a PF thickness greater than 4 mm. For SWE tissue elasticity (Young’s modulus kPa; shear wave speed m/ s) was measured at Location 1: directly at the calcaneus; Location 2: +1 cm distal of the calcaneus and Location 3: central part of the calcaneus. Sensitivity, specificity, and diagnostic accuracy as well as correlation to American Orthopaedic Foot and Ankle Score (AOFAS) and Food Functional Index (FFI) were determined. Results: Symptomatic PF are thicker (4.2 mm, n = 39) than asymptomatic (3.0 mm, n = 43) (p < 0.001). Thickness of the PF (n = 82) correlated poorly to clinical scores (p = 0.001): FFI-pain (r = 0.349); FFI-function (r = 0.381); AOFAS (r = ¡0.387). Cut-off point for positive SWE finding was 51.5 kPa (4.14 m/s). Symptomatic PF (31.9 kPa, 3.26 m/s, n = 39) differ significantly from asymptomatic PF (93.3 kPa, 5.58 m/s, n = 43) with significant differences at L1 between all groups (p < 0.001). Correlation between Young’s modulus (n = 82) and clinical scores was strong (p < 0.001): FFI-pain (r = ¡0.595); FFI-function (r = ¡0.567); AOFAS (r = 0.623,). B-US: sensitivity (61%), specificity (95%); SWE sensitivity (85%), specificity (83%). The combination of SWE and B-US increases the sensitivity (100%) with a diagnostic accuracy of 90%. Conclusion: Based on our results, we could show that SWE can improve the diagnostic accuracy in patients with plantar fasciitis compared to B-US. Level of evidence: II Key Words: Shear wave elastography; Plantar fasciitis; AOFAS Score; FFI; ultrasound. © 2019 Published by Elsevier Inc. on behalf of The Association of University Radiologists.
INTRODUCTION
P
lantar fasciitis (PFis) is a common cause for heel pain and in the United States alone almost 1 million patients per year require treatment (1). Imaging, eg with ultrasound, is useful in unclear cases to verify the
Acad Radiol 2019; &:1–8 From the Department of Orthopedics, University Hospital RWTH Aachen, Pauwelsstraße 30, Aachen 52074, Germany (M.G., L.B., V.Q., M.T., M.B.); Department of Diagnostic and Interventional Radiology, University Hospital RWTH Aachen, Pauwelsstraße 30, Aachen 52074, Germany (S.S., T.D., C.K.). Received March 3, 2019; revised April 8, 2019; accepted April 30, 2019. Declarations of interest: none Address correspondence to: M.G. e-mail: [email protected] © 2019 Published by Elsevier Inc. on behalf of The Association of University Radiologists. https://doi.org/10.1016/j.acra.2019.04.009
diagnosis (2). Contemporary ultrasound imaging of plantar fasciitis (PFis) is mainly based on gray-scale imaging (B-US) finding hypoechoic areas, border irregularities of the plantar fascia (PF) and calcifications (3,4). Current studies showed that PF thickness of greater than 4 mm is the most important diagnostic parameter for diagnosing PFis using conventional B-US (3,4). However, recently Hansen et al found that the PF of patients with PFis can also decrease in thickness over time (5) and that even after a follow-up of more than 10 years hypoechoic and thickened areas in the PF can still be found in 75% of asymptomatic individuals (5). Conventional B-US elastographic modalities, such as axial strain elastography (ASE) and shear wave elastography (SWE), have been evaluated in the diagnosis of PFis in the
1
ARTICLE IN PRESS GATZ ET AL
recent years. ASE is based on the external compression of the tissue by a probe, and therefore it offers only semi-quantitative and user-dependent information of tissue elasticity (6 8). In contrast, using SWE an impulse is generated by a probe and elastographic parameters are quantitatively measured in Kilopascal (kPa, m/s) as an elastic module (Young’s modulus, kPa) or as shear wave speed progression (m/s) (6 8). Therefore, the method of SWE is considered to be user-independent. Studies using ASE showed that symptomatic PF are “softer” than asymptomatic fascia and that older individuals might have a “softer” PF than younger individuals (9). Furthermore, ASE improves the diagnostic accuracy to diagnose symptomatic PF, in cases where no morphologic changes on B-US were found (10,11,12). Zhang et al found “softer” PF in symptomatic patients, at the site of the pain, while the stiffness of the fascia increased 1 cm distal of its insertion (13). Besides diagnosing PFis, elastography might be more helpful to monitor treatment effects than conventional B-US. Using ASE, Kim et al found no significant reduction in PF thickness, hypoechogenicity, or perifascial oedema three months after a collagen injection. However, the stiffness of the fascia increased, which might be caused by a slight reduction of the edema in the PF, that cannot be detected using B-US (14). In summary, ultrasound elastography seems to be beneficial in diagnosing and monitoring PFis, however most current studies have used ASE rather than SWE. Therefore, aim of this prospective diagnostic study was to comparatively analyze fascia stiffness and thickness using B-US and SWE in asymptomatic and symptomatic individuals. Furthermore, purpose of this study was to determine the sensitivity, specificity, and diagnostic accuracy of B-Mode, SWE and the combination of both modalities in the diagnosis of PFis. METHODS Study Design
This is a diagnostic case-control study using SWE and B-US to examine PFis. The actual status of symptoms was evaluated by the validated Foot Functional Index (FFI) Pain and Function Scale as well as by the American Orthopaedic Foot and Ankle Score (AOFAS) (15,16). The study was approved by the local ethics committee (EK 060/17) and all participants provided written informed consent.
Academic Radiology, Vol &, No &&, && 2019
big toe (Windlass Test). In the clinical history there was no suspicion for differential diagnosis eg stress fractures or nerve compression syndromes. If available, previous image modalities such as X-ray were reevaluated. There was no minimum symptom duration for inclusion in the trial, but FFI- Pain and AOFAS-Scores were required to be over 10 and under 90, respectively. Individuals with a rheumatic disease, overand underweight (BMI (kg/m2): <17; >34), previous injection therapy in the last six months, previous surgery and trauma as well as pregnancy were excluded from the study. According to the participants symptoms, we divided them into three subgroups (Fig 1): Group 1: Symptomatic individuals suffering from a clinically diagnosed unilateral or bilateral PFis. Group 2: Asymptomatic PF of individuals with unilateral PFis from group 1 with no heel pain or discomfort at any time prior to the study. Group 3: Asymptomatic individuals with no pain or discomfort in both PF at any time and the respective score points (FFI: zero; AOFAS: 100). Data Acquisition with B-US, SWE
Every participant underwent a standardized multi-modal ultrasound protocol consisting of B-US and SWE of both PF. Measurements were conducted using an ultrasound system (Aixplorer, Supersonic Imagine, Aix-en-Provence, France) with a high-resolution linear 18 MHz transducer, (SuperLinear SL 18-4, Supersonic Imagine, Aix-en-Provence, France) in longitudinal planes, carefully avoiding over-compression of the PF. All individuals underwent a standardized examination protocol, always starting with the left foot and conducting B-US before SWE. US examinations were always performed in the same air-conditioned room with a standardized temperature of 20°C. Individuals were lying in the prone position with the foot hanging relaxed over the examination table. The measured foot was kept in the neutral position without any active or passive dorsiflexion of the big toe, which might have preloaded the PF similar to the Windlass Test. Nevertheless, a recent study using SWE demonstrated, that preloading does not influence measurements of the PF directly at the insertion site at the calcaneus (17). All examinations were performed by one physician with four years of experience in ultrasound and shear wave elastography, who was blinded to the clinical symptoms of the study participants.
Inclusion and Exclusion Criteria
Individuals were recruited from offices of primary care physicians from 07/2017 until 03/2018. All individuals have been previously diagnosed and treated for PFis (Secondary Health Care) and were referred by the treating clinicians. We included participants older than 18 years with a clinical diagnosis of PFis. During clinical examination, by an experienced orthopedic surgeon, all individuals had pain at the insertion of the PF, which increased by dorsiflexion of the 2
B-US Measurements and PF Thickness>4 mm as Reference Standard
According to previous studies, PF thickness over 4 mm was considered a validated parameter for an acute state of PFis, and the reference standard for sonographic diagnosis (3,4). Therefore, a PF thickness greater than 4 mm was regarded as a positive finding for PFis, while hypoechogenicity and loss of PF borders were only noted, but not considered for
ARTICLE IN PRESS Academic Radiology, Vol &, No &&, && 2019
EVALUATION OF PATIENTS WITH PLANTAR FASCIITIS
Figure 1. Study design and overview of the groups. Based on symptomatic and asymptomatic participants three groups were formed. (Color version of figure is available online.)
calculating sensitivity or specificity. The measurements of the PF were conducted strictly vertically in a longitudinal plane over the calcaneus at the thickest part of the fascia, which is mainly at the medial tubercle of the calcaneus.
SWE-measurements
The size of the SWE-measurement window was 2 cm2. Quantitative SWE-measurements were performed by placing
Figure 2. Acquisition of Youngs moduli at three different locations within the PF. L1: directly at the calcaneus, L2: 1 cm distal from the calcaneus, L3: central part of the PF. All images demonstrate the semi-quantitative SWE color mapping, while quantitative values were measured in the Q-Box (white circle). Blue color demonstrates areas with lower Youngs modulus while red depict areas with the highest Youngs Modulus (blue
3
ARTICLE IN PRESS GATZ ET AL
a region of interest of 2-mm-diameter in the most rigid part of the PF in the area of the greatest PF thickness. The acquired SWE-information was evaluated quantitatively as Young’s modulus (kPa) and shear wave speed (m/s). Mean value and standard deviations of the measurements were calculated. Every PF was systematically analyzed in the longitudinal plane at three locations: L1: insertion directly at the calcaneus, L2: +1 cm distally from the calcaneus insertion and L3: at the central part of the PF at least 2 cm from the calcaneal insertion (Fig 2). Representative SWE-measurements were performed three times in each location (= nine images in total per PF) and then were averaged. Since there is no cut-off value in the current literature for diagnosing PFis, a ROC analysis was used to establish a cut-off value for measurements at the calcaneus. The “SWE-penetration mode” has been used as a setting to reduce so called “signal void areas,” which are parts without any SWE-information (18). Nevertheless, the SWE measurement window does not always allow to display a complete SWE-signal color map of the area of the PF or the subcutaneous fat pad. In these cases, the quality and completeness of the color map and thus SWE acquisition was rated in four grades based on the amount of colored area within the PF and the subcutaneous fat pad: grade 1: 100%, grade 2: 99% 75%, grade 3:74% 50%, and grade 4: 50% (Fig 3). Statistical Analysis
For all analyses, SPSS Statistics, version 24.0 (IBM Corp.) was used. SWE-values were analysed with descriptive statistics. Additionally, normal distribution of the data was tested by the Shapiro Wilk-test. To examine for significant differences between the three independent groups, a Kruskal Wallis test was used with a posthoc Dunn Bonferroni test. A Wilcoxon
Academic Radiology, Vol &, No &&, && 2019
signed-rank test analysed intraindividual differences of group 2. The correlation between SWE-values and the clinical scores or age was assessed using Pearson’s correlation test. The level of significance was set at p < 0.05. Effect size was calculated with G*Power 3.1.9.4 (D€ usseldorf, Germany) (19). RESULTS In total, 82 PF have been included in this study. They consisted of 31 patients with a total of 39 symptomatic PF (23 unilateral, 16 bilateral), 23 asymptomatic PF of patients with unilateral PFis. Furthermore, 10 bilateral asymptomatic individuals with 20 bilateral PF of a healthy control have been included. G*Power revealed an effect size of 0.44. Figure 1 provides an overview of the different groups and depicts the group constellations in detail. Table 1 demonstrates the demographic data as well as their clinical scores: patients in group 1 (age: p = 0.016; BMI: p = 0.010) and group 2 (age: p = 0.009, BMI: p = 0.016) were significantly older and with a higher BMI, than patients in group 3. However, there was no difference for hours of sport activities between the groups. In symptomatic patients, we found a moderate and significant correlation between BMI and the FFI-function score, as well as pain measured by the FFI-pain score and for the AOFAS score: FFI pain (r = 0.534, p = 0.002, n = 31); FFI function (r = 0.511, p = 0.003, n = 31); AOFAS (¡0.457, p = 0.010, n = 31). However, there was no correlation between age and the respective clinical scores (FFI-pain r = 0.132, p = 0.480; FFI-function r = 0.051, p = 0.786; AOFAS r = ¡0.046, p = 0.806, n = 31). B-US Findings
Figure 4 shows the thickness of the PF at the calcaneus in all three groups. Symptomatic PF of group 1 are on average
Figure 3. Due to “signal void areas” the quality of the received SWE signal was rated in four grades depending on the area of the PF which was without signal: a) grade 1: 100% b) grade 2: 99% 75% c) grade 3: 74% 50% d) grade 4: <50%. (Color version of figure is available online.)
4
ARTICLE IN PRESS Academic Radiology, Vol &, No &&, && 2019
EVALUATION OF PATIENTS WITH PLANTAR FASCIITIS
TABLE 1. Demographics of the Different Groups
Age (a) min max Sex BMI (kg/m2) min max Sport (h/week) min max Symptom duration (month) min max FFI Pain min max FFI Functional min max 3AOFAS min max Therapy - Orthotics - Analgesics - Physiotherapy - Kryotherapy - Deep x-ray therapy
Significant Difference (Kruskal Wallis)
Group 1
Group 2
Group 3
48.9 (§9.6) (26 66) Male:8 female: 23 26.7 (§4.6) (21 34) 3.8 (§3.6) (0 12.5) 14.3 (§16.7) (2 84) 36.4 (§17.7) (11 75) 31.6 (§17.9) (12 69) 74 (§11) (47 85)
50.0 (§9.5) (33 66) Male: 8 female:15 26.6 (§4.1) (21 38) 3.8 (§3.9) (0 12.5) 0
30.4 (§11.1) (21 60) Male:5 female:5 22.8 (§1.0) (21 24) 2.1 (§2.1) (0 6.0) 0
1 vs. 2/3 (p < 0.001)
0
0
1 vs. 2/3 (p < 0.001)
0
0
1 vs. 2/3 (p < 0.001)
100
100
1 vs. 2/3 (p < 0.001)
1 vs. 3 (p = 0.016) 2 vs. 3 (p = 0.009)
3 vs. 1 (p = 0.010) 3 vs. 2 (p = 0.016) 1 vs. 2 vs. 3. (p = 0.726)
12 9 10 6 4
4.2 mm (§1.2) thick, while asymptomatic PF of group 3 have a PF thickness of only 2.8 mm (§0.5). Further detailed information is shown in Table 2. The Kruskal Wallis test revealed, that there is a significant difference between group 1 and group 3 (p < 0.001) as well as group 1 and group 2 (p = 0.030), showing that symptomatic PF are thicker than asymptomatic ones. Moreover, thickness is significantly (p < 0.001) reduced in all asymptomatic PF (group 2/3: 3.0 mm (§0.6)) compared to symptomatic PF (group 1: 4.2 mm (§1.2)). Thickness of the PF measured by B-US correlated to clinical scores (FFI-pain, FFI-function, AOFAS as well as to age, but not to BMI (Table 3). SWE Findings
The ROC analysis of 82 PF showed x< 51.5 kPa (4.14 m/s) as an ideal cut-off value to diagnose PFis using SWE. Table 2 and Figure 4 show Young’s moduli in every group and every location measured within the PF. In all three measured locations (L1 L3), group 3 has significantly higher elastic Young’s moduli than both other cohorts (p < 0.001 p = 0.002, Table 2). In particular, at L1, SWE elastic moduli differed significantly between all groups (Table 2). At L1 symptomatic PF (31.9 kPa, 3.26 m/s, n = 39; L1; group 1) had significant (p < 0.001) lower Young’s moduli than asymptomatic PF in a pooled analysis of group 2, 3 (93.3 kPa, 5.56 m/s, n = 43). Moreover, a Wilcoxon signed rank test analyzing the intraindividual differences of patients with unilateral symptoms (group 2) revealed a
significant difference between the symptomatic (18.5 kPa, 2.48 m/s) and asymptomatic (36.1 kPa, 3.47 m/s) side (z = ¡1,964, p = 0.05, n = 23). We found a moderate but significant correlation between the Young’s moduli measured by SWE (L1) and clinical scores (FFI-pain, FFI-function, AOFAS) as well as age and BMI (Table 3). Additionally, there was a correlation between SWE values and the thickness measured in B-US (r = ¡0.421, p < 0.001, n = 82). In total, analysing the quantity of the received SWE signal of asymptomatic PF (group 3) had a mean grade of 1.7 (§0.47) at L1, 1.7 (§0.47) at L2 and grade 1 (§0) at L3. In symptomatic patients PF (group 1) the quality of the SWE signal was significantly lower compared to all asymptomatic patients (group 2/3): at L1 grade for signal mapping was 2.4 (§0.9) (U = 501, p = 0.001); at L2 mean grade was 2.4 (§1.1) (U = 546, p = 0.009); and at L3 1.8 (§1.1) (U = 642, p = 0.056).
Sensitivity, Specificity, and Diagnostic Accuracy
In this study, B-US alone had the lowest sensitivity for diagnosing PFis with 61% and a specificity of 95%, while SWE alone was more sensitive (85%) but less specific (83%). Nevertheless, the combination of both modalities, SWE and B-US, increased the sensitivity to 100% and the diagnostic accuracy to 90%. A comparison of diagnostic accuracy of the ultrasound techniques is given in Table 4. 5
ARTICLE IN PRESS GATZ ET AL
Academic Radiology, Vol &, No &&, && 2019
DISCUSSION
Figure 4. PF thickness (mm) in B-US and Youngs moduli (kPa) in SWE. (Color version of figure is available online.)
6
The present study is the first, which determines the diagnostic accuracy of SWE as a tool for the diagnosis of PFis in comparison to B-US. Furthermore, we evaluated the correlation of PF thickness (cm) measured in B-US and Young’s moduli (kPa) to clinical scores (FFI-Pain, FFI-Function, AOFAS), which has not been investigated yet in the current literature. Some of the results presented in our study are in line with previous studies using ASE and SWE: Symptomatic PF have significant lower Young’s moduli than asymptomatic ones (31.9 kPa vs. 93.3 kPa; 3.26 m/s vs. 5.56 m/s; p < 0.001), this means that they are “softer” or “less elastic.” Furthermore, PF do not have only lower Young’s modulus values when they are symptomatic, but these values also decrease with age (r = ¡0.352, p = 0.001). Three studies using ASE confirmed the reduction of elasticity due to symptoms or age (9,12,20). Zhang et al could demonstrate that these findings shall apply to SWE as well (13). At the calcaneal insertion (L1) symptomatic PF and PF of older participants have significant lower Young’s elastic moduli (13). Furthermore, elastic moduli increase, when measured +1 cm from the calcaneal insertion (13). Our findings showed that in every group Young’s moduli increase the further away we measured from the calcaneus. However, there was only a significant difference found between L1 vs. L3 and L2 vs. L3 in groups 1 and 3 (Table 2). The optimum location for measuring the PF is therefore directly at the location of symptoms (L1), where SWE elastic modulus differ significantly between all groups except group 2 vs. group 3. So far, the diagnostic accuracy for diagnosing PFis has only been evaluated using ASE. By combining B-US and ASE the diagnostic accuracy was, increased from 90.0% to 95.4% (10). The present study showed for the first time, that SWE has an additive diagnostic value for diagnosing PFis with a sensitivity of 100% for the combined usage of SWE and B-US. Additionally, the diagnostic accuracy increased from 79% using BUS to 84% using SWE. The combination of both modalities has a diagnostic accuracy of 90%. Compared to the present study, Sconfienza et al reported higher diagnostic accuracies for B-US (90%) and ASE (95.4%) (10), which might be caused by differences between the study cohorts, a different elastographic setting and the fact that hypoechoic area and blurring of fascia borders were also taken into account. In contrast, our present study only evaluated quantitative parameters (mm, kPa). In previous studies, parameters of elastography have not been correlated with clinical scores. The results of our study showed a strong correlation between Young’s moduli and clinical scores (r = 0.6; p < 0.001), which was higher than the correlation between B-US and clinical scores (r = 0.35; p < 0.001). Our study has a few limitations according to the used study design and methods. Participants in group 3 were younger and lighter than patients in group 1 and 2. Nevertheless, there was no significant age difference between group 1 and 2,
ARTICLE IN PRESS Academic Radiology, Vol &, No &&, && 2019
EVALUATION OF PATIENTS WITH PLANTAR FASCIITIS
TABLE 2. Imaging Findings of SWE and B-US in Comparison to Different Groups and Locations of Measurement Youngs Modulus (kPa) Shear Wave Speed (m/s)
Group 1
Group 2
Group 3
Sign. Difference (Kruskal Wallis) 1 vs. 2 p = 0.006 1 vs. 3 p< 0.001 2 vs. 3 p = 0.002 1 vs. 2 p = 0.167 1 vs. 3 p< 0.001 2 vs. 3 p< 0.001 1 vs. 3 p = 0.362 1 vs. 3 p< 0.001 2 vs. 3 p< 0.001
L1: Calcaneus
Mean (§SD)
31.9 kPa (§16.2) 3.26 m/s
65.3 kPa (§42.4) 4.66 m/s
124.1 kPa (+/-28.5) 6.43 m/s
L2: Calcaneus +1cm
Mean (§SD)
46.3 kPa (§37.5) 3.93 m/s
75.3 kPa (§59.2) 5.01 m/s
158,4 kPa (§55.7) 7.26 m/s
L3: Central part
Mean (§SD)
88.9 kPa (§66.5) 5.46 m/s
109.1 kPa (§72.5) 6.03 m/s
230.6 kPa (§25.9) 8.76 m/s
L3 vs. L1 p< 0.001 L3 vs. L2 p = 0.023
¡
L3 vs. L1 p = 0.001 L3 vs. L2 p = 0.001
Sign. difference (Kruskal¡Wallis)
Group 1
Group 2
Group 3
Sign. Difference (p < 0.05)
Hypoechogenicity
Mean (§SD) median Positive findings
4.2 mm (§1,2) 4.6 mm 35/39
3.19 mm (§0.7) 3.1 mm 12/23
2.8 mm (§0.5) 2.6 mm 1/20
Border irregularities
Positive findings
32/39
5/23
0/20
1 vs. 3 p < 0.001 1 vs. 2 p = 0.030 1 vs. 3 p < 0.001 1 vs. 2 p = 0.001 1 vs. 3 p < 0.001 1 vs. 2 p < 0.001
B-US: thickness of PF (mm) Calcaneus
with group 2 being an asymptomatic control having higher SWE-values than group 1. Additionally, we scattered the influence of age and BMI, by an intraindividual comparison of group 2 showing that symptomatic PF have significantly lower Young’s modulus than asymptomatic ones (p = 0.05). A further limitation is that, all individuals in group 1 received different treatments before, which might be a potential confounding factor influencing the Young’s modulus. Dirrichs et al showed that excessive load in semi-professional runners might lead to higher Young’s modulus in Achilles tendons (21). Therefore, physical therapy before recruitment might have influenced elasticity of PF. Furthermore, all image acquisitions were conducted by one physician, thus it was not possible to evaluate interclass observer correlation. Additionally, in the current literature there are no standard values for symptomatic or asymptomatic PF, so that other studies might reveal other cut-off values than ours. Furthermore, the present study is a pilot study with a small sample size being more a case-control study than a diagnostic study and further
studies need to revaluate the results with larger cohorts. Using 7% as the general prevalence in individuals over 65 years (2) and a change of sensitivity from 60% (B-US) to 80% (SWE) a larger control group would have been necessary (22). Nevertheless, our study compares to sample sizes of previous investigations using ASE (10,11,13). To our knowledge, previous studies using ASE or SWE have not reported difficulties in receiving a full elastographic signal in the ROI. However, in our study measurements at L1 in group 1, which is actually the most relevant group and location for measurements, we found that in up to 25% of the PF no SWE signal could be measured. As this was not seen that strong in asymptomatic PF (group 3; mean grade 1.7) or at measurements at locations without symptoms (L3 symptomatic: 1.8; L3 asymptomatic: one) inadequate acquisition of SWE signal might be due to tissue remodeling at the insertion of PF due to matrix calcification, chondroid metaplasia or collagen necrosis being common findings of histopathological adaptation (23). Bouchet et al stated in their recent review that chondroid areas as well
TABLE 3. Correlation in B-US and SWE to Clinical Scores and Demographic Parameters of the Whole Study Population (n = 82)
B-US thickness (group 1,2,3 n = 82) SWE Youngs modulus (L1) (group 1,2,3 n = 82)
FFI-pain
FFI-function
AOFAS
Age
BMI
r = 0.349
r = 0.381
r = ¡0.387
r = 0.293
r = 0.192
p = 0.001 r = ¡0.595
p < 0.001 r = ¡0.567
p < 0.001 r = 0.623
p = 0.008 r = ¡0.442
p = 0.085 r = ¡0.485
p< 0.001
p < 0.001
p < 0.001
p < 0.001
p < 0.001
7
ARTICLE IN PRESS Academic Radiology, Vol &, No &&, && 2019
GATZ ET AL
TABLE 4. Sensitivity, Specificity and Diagnostic Accuracy of B-Mode, SWE and the Combination of SWE+ B-Mode B -Mode Calcaneus
True + False + Sensitivity Pos. predictive value Diagnostic accuracy
True ¡ False¡ Specificity Neg. predictive value
as intralesional necrosis can be “signal void areas,” which are not coded on the color map (18). Therefore, in very severe cases of PF with a large area of tissue remodeling the informative value of SWE might be diminished. Moreover, micro calcification being not that apparent in B-US could have influenced SWE measurements in symptomatic PF (group 1) and lead to reduced Young’s modulus (24). The results of this study showed that SWE (84%) improves the diagnostic accuracy compared to B-US (79%) and that the combination of SWE and B-US (90%) is a strong diagnostic tool. SWE might be the method of choice for providing information, which correlate stronger to clinical scores than B-US. SWE provides a quantitative assessment of PF integrity and can distinguish between symptomatic and asymptomatic patients better than B-US. Therefore, further studies need to reevaluate if B-US is still the gold standard in the diagnosis of PFis.
REFERENCES 1.
2. 3.
4.
5.
6. 7.
8
Riddle DL, Schappert SM. Volume of ambulatory care visits and patterns of care for patients diagnosed with plantar fasciitis: a national study of medical doctors. Foot Ankle Int 2004; 25:303–310. Goff JD, Crawford R. Diagnosis and treatment of plantar fasciitis. Am Fam Physician 2011; 84:676–682. McMillan AM, Landorf KB, Barrett JT, et al. Diagnostic imaging for chronic plantar heel pain: a systematic review and meta-analysis. J Foot Ankle Res 2009; 2:32. Mohseni-Bandpei MA, Nakhaee M, Mousavi ME, et al. Application of ultrasound in the assessment of plantar fascia in patients with plantar fasciitis: a systematic review. Ultrasound Med Biol 2014; 40:1737–1754. Hansen L, Krogh TP, Ellingsen T, et al. Long-term prognosis of plantar fasciitis: A 5- to 15-year follow-up study of 174 patients with ultrasound examination. Orthop J Sports Med 2018; 6:2325967118757983. Ryu J, Jeong WK. Current status of musculoskeletal application of shear wave elastography. Ultrasonography 2017; 36:185–197. Taljanovic MS, Gimber LH, Becker GW, et al. Shear-wave elastography: basic physics and musculoskeletal applications. Radiographics 2017; 37:855–870.
24 2 61% 0.92 79%
41 15 95% 0.73
SWE 33 7 85% 0.83 84%
SWE+ B-Mode 36 6 83% 0.86
39 8 100% 0.83 90%
35 0 81% 1.00
8. Fusini F, Langella F, Busilacchi A, et al. Real-time sonoelastography: principles and clinical applications in tendon disorders. A systematic review. Muscles Ligaments Tendons J. 2017; 7:467–477. 9. Wu CH, Chang KV, Mio S, et al. Sonoelastography of the plantar fascia. Radiology 2011; 259:502–507. 10. Sconfienza LM, Silvestri E, Orlandi D, et al. Real-time sonoelastography of the plantar fascia: comparison between patients with plantar fasciitis and healthy control subjects. Radiology 2013; 267:195–200. 11. Wu CH, Chen WS, Wang TG. Plantar fascia softening in plantar fasciitis with normal B-mode sonography. Skeletal Radiol 2015; 44:1603–1607. 12. Lee SY, Park HJ, Kwag HJ, et al. Ultrasound elastography in the early diagnosis of plantar fasciitis. Clin Imaging 2014; 38:715–718. 13. Zhang L, Wan W, Zhang L, et al. [Assessment of plantar fasciitis using shear wave elastography]. Nan Fang Yi Ke Da Xue Xue Bao 2014; 34:206–209. 14. Kim M, Choi YS, You MW, et al. Sonoelastography in the evaluation of plantar fasciitis treatment: 3-Month follow-up after collagen injection. Ultrasound Q 2016; 32:327–332. 15. Kostuj T, Krummenauer F, Schaper K, et al. Analysis of agreement between the German translation of the American Foot and Ankle Society's Ankle and Hindfoot Scale (AOFAS-AHS) and the Foot Function Index in its validated German translation by Naal et al. (FFI-D). Arch Orthop Trauma Surg 2014; 134:1205–1210. 16. Naal FD, Impellizzeri FM, Huber M, et al. Cross-cultural adaptation and validation of the Foot Function Index for use in German-speaking patients with foot complaints. Foot Ankle Int 2008; 29:1222–1228. 17. Chino K, Lacourpaille L, Sasahara J, et al. Effect of toe dorsiflexion on the regional distribution of plantar fascia shear wave velocity. Clin Biomech 2018; 61:11–15. 18. Bouchet P, Gennisson JL, Podda A, et al. Artifacts and technical restrictions in 2D shear wave elastography. Ultraschall Med 2018. 19. Faul F, Erdfelder E, Lang AG, et al. G*Power 3: a flexible statistical power analysis program for the social, behavioral, and biomedical sciences. Behav Res Methods 2007; 39:175–191. 20. Rios-Diaz J, Martinez-Paya JJ, del Bano-Aledo ME, et al. Sonoelastography of plantar fascia: reproducibility and pattern description in healthy subjects and symptomatic subjects. Ultrasound Med Biol 2015; 41:2605–2613. 21. Dirrichs T, Schrading S, Gatz M, et al. Shear Wave Elastography (SWE) of asymptomatic achilles tendons: a comparison between semiprofessional athletes and the nonathletic general population. Acad Radiol 2019. 22. Bujang MA, Adnan TH. Requirements for minimum sample size for sensitivity and specificity analysis. J Clin Diagn Res 2016; 10: YE01-YE6. 23. Snider MP, Clancy WG, McBeath AA. Plantar fascia release for chronic plantar fasciitis in runners. Am J Sports Med 1983; 11:215–219. 24. Gregory A, Bayat M, Denis M, et al. An experimental phantom study on the effect of calcifications on ultrasound shear wave elastography. Conf Proc IEEE Eng Med Biol Soc 2015: 3843–3846. 2015.