The effects of shockwave therapy on musculoskeletal conditions based on changes in imaging: a systematic review and meta-analysis with meta-regression

Background Shockwave therapy (SWT) is a commonly used intervention for a number of musculoskeletal conditions with varying clinical outcomes. However, the capacity of SWT to influence pathophysiological processes and the morphology of affected tissues remains unclear. The objective of the current review is to evaluate changes in imaging outcomes of musculoskeletal conditions following SWT. Methods A comprehensive search of Medline, Embase, Cochrane Controlled Trials Register, CINAHL and SportDiscus was conducted from inception to October 2018. Prospective clinical trials evaluating the effectiveness of SWT based on changes in imaging outcomes were eligible for inclusion. Articles were evaluated independently for risk of bias using the Cochrane Risk of Bias list and the Methodological Index for Non-Randomized Studies. Random-effects meta-analysis and meta-regression with a priori determined covariates was conducted for each condition to determine potential predictors of SWT effects. Results Sixty-three studies were included, with data from 27 studies available for effect size pooling. Meta-analyses and meta-regression on imaging outcomes were performed for rotator cuff calcific tendinitis (n = 11), plantar fasciitis (n = 7) and osteonecrosis of the femoral head (n = 9). There was an overall reduction in the size of measured lesion following SWT (MD 8.44 mm (95%CI -4.30, 12.57), p < 0.001) for calcium deposit diameter, (MD 0.92 mm (95%CI -0.03, 1.81), p = 0.04) for plantar fascia thickness and (MD 4.84% (95%CI -0.06, 9.75), p = 0.05) for lesion size in femoral head osteonecrosis. Meta-regression showed no influence of SWT dosage parameters, however, baseline lesion size was an independent predictor for changes in imaging outcomes. Conclusions SWT altered the morphology of musculoskeletal conditions, potentially reflecting changes in underlying pathophysiological processes. The parameters of SWT dosage are not significant predictors of changes in imaging outcomes. Lack of adequate reporting of imaging outcomes limited the conclusions that could be drawn from the current review. Registration number: PROSPERO CRD42018091140.


Background
Shockwave Therapy (SWT) is used to treat a range of musculoskeletal conditions. Focused and radial shockwave (radial pressure waves) are two technically distinct forms of SWT. It has been argued that focused shockwave therapy and radial shockwave therapy should be viewed as distinctly different therapeutic modalities [1]. However, despite the differences in their physical characteristics, method of energy generation and shockwave propagation, focused and radial shockwave types share common clinical indications [2]. SWT is often indicated as a secondary conservative treatment choice for recalcitrant musculoskeletal conditions, unresponsive to standard care [1,3]. These indications include plantar fasciitis, Achilles tendinopathy, patellar tendinopathy, calcific and non-calcific shoulder tendinopathy and lateral epicondylitis. Also, bone and cartilage related disorders such as non-union of fractures, osteonecrosis of the femoral head and knee osteoarthritis related bone marrow edema (BME) are among the range of SWT clinical indications. Research evidence for SWT clinical effectiveness varies across the indicated conditions. Good evidence based on systematic reviews exists to support the use of SWT for calcific tendinopathy of the shoulder [4], Achilles tendinopathy [5,6], knee osteoarthritis [7], early stage osteonecrosis of the femoral head [8] and plantar fasciitis [9]. However, research evidence for the effectiveness of SWT in lateral epicondylitis is variable [1] and is lacking for patellar tendinopathy [3,6].
Focused shockwaves are generated through three mechanisms: electrohydraulic, piezoelectric or electromagnetic methods that convert electrical energy into kinetic energy, whilst radial shockwaves are generated pneumatically [2]. The proposed mechanism of action for SWT is based on mechano-transduction [10]. The delivery of mechanical acoustic energy to the target tissue induces molecular, cellular and tissue responses [11]. Based on animal studies, SWT promotes the expression of various angiogenic and osteogenic growth factors such as vascular endothelial growth factor (VEGF) and bone morphogenetic protein (BMP), which promotes tissue regeneration [12]. In addition, SWT has an anti-inflammatory effect by modulating the expression of interleukins (IL-6 and IL-10) and other cytokines [13].
Clinical outcomes such as pain rating and functional disability scores are commonly used to evaluate the effectiveness of SWT. However, the utilization of objective outcome measures that evaluate changes in the affected musculoskeletal tissues is required to provide evidence of SWT influence on pathophysiological processes in humans. Different imaging techniques are used clinically to establish a diagnosis, guide the delivery of an intervention or to evaluate the effectiveness of an intervention. A range of imaging modalities including magnetic resonance imaging (MRI), ultrasonography (US), Computed Tomography (CT), dual-energy x-ray absorptiometry (DEXA) or plain radiography have been used to support diagnosis and in some cases to evaluate outcomes in studies investigating SWT. Despite lack of consistency in utilization, these imaging modalities are valuable tools for evaluating the morphological characteristics of injured tissues. That can then be used clinically to monitor improvements in the underlying pathophysiological process following an intervention. Therefore, the utilization of suitable imaging tools at appropriate time points as an adjunct to clinical examination may provide a better understanding of the tissue pathophysiology and support the management planning process. Little is known about the capacity of SWT to induce positive improvements in pathophysiological processes in musculoskeletal disorders, as indicated by changes in imaging parameters and no systematic reviews have been conducted on this topic. Therefore, the primary aim of this review was to explore the available evidence from clinical prospective trials with regard to any changes in the morphology of musculoskeletal lesions following SWT, as measured by imaging parameters. The secondary aim was to investigate significant predictors for the SWT effects using meta-regression. We also sought to make recommendations for future research studies.

Protocol and registration
This systematic review and meta-analysis was conducted following the guidelines of the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA). The protocol of the current review was registered on the International Prospective Register of Systematic Reviews (PROSPERO; 2018, CRD42018091140).

Search strategy
A comprehensive search using the electronic databases of MEDLINE, EMBASE, CINAHL, SPORTDiscuss and the Cochrane Controlled Trials Register using the PICOS format (population, intervention/exposure, comparison, outcome and study design) was conducted in October 2018. A range of keywords integrated with subject headings relevant to the review were systematically searched to maximize the search results. Keywords for the population category contained terms related to musculoskeletal conditions such as arthritis, fractures, bone marrow edema and tendinopathies. The intervention category contained keywords related to shockwave therapy. The outcome category contained terms related to different imaging methods (e.g. X-ray, MRI, Ultrasound imaging). No specific comparator was added as the primary aim of the review was to examine post intervention imaging changes. An example of the basic search strategy for databases is presented in Additional file 1.
Eligibility criteria included prospective study designs to avoid any potential selection bias inherent in retrospective studies, adult participants of both genders with an established musculoskeletal diagnosis and reported pre and post imaging measures to facilitate evaluation of changes in affected tissue morphology following SWT of any type (focused and radial). Non-human and non-English language studies were excluded.

Study identification
The principal author conducted the database search. Study screening was carried out using the Covidence platform [14]. Two independent reviewers (HA and SA) screened titles and abstracts to determine eligibility for inclusion. The decision to include or exclude studies, based on the eligibility criteria, was made independently by each author. Full-texts of potentially relevant studies were retrieved for further assessment. Discrepancies in opinions were resolved by a third senior reviewer (AW).

Risk of bias assessment
The quality of studies was independently evaluated by two reviewers (HA and SA). A third senior reviewer (AW) was involved to resolve differences in assessment. Randomized Controlled Trials (RCTs) were evaluated using the Cochrane Risk of Bias list [15]. The tool evaluates the risk of bias in five domains (selection, performance, attrition, reporting and other) for seven elements to be judged as high, unclear or low risk. A three-point scale was used to assign each degree of risk a number value with higher numbers indicating a lower risk of bias (low risk = 2, unclear risk = 1 and high risk = 0). The maximum score of 14 indicates the lowest risk of bias for a given study.
The Methodological Index for Non-Randomized Studies (MINORS) tool [16,17] was used to evaluate the non-randomized studies. The MINORS is a valid and reliable tool that contains 12 methodological items, the first eight items measure attributes specific to noncomparative studies. Four additional items are evaluated only for studies with comparative groups. A higher score indicates a lower risk of bias.
The risk of bias was assessed mainly at the reported measured imaging outcomes level. The score for each risk of bias tool was divided by the total number of points possible to calculate the risk of bias percentage for each study.

Data extraction
One reviewer (HA) extracted relevant data from individual studies and the data were independently checked by a second reviewer (SA) for consistency. Extracted data included study characteristics such as study design, participant numbers and demographics, the condition treated, parameters and dosage of the SWT intervention (level of energy, number of shocks and sessions), followup period, imaging modality description and findings. Authors were contacted when relevant outcomes were incompletely reported and were asked to provide missing information. When the values of standard deviations (SD) were not provided for the intervention or control group, these were calculated from confidence intervals, and p-values for differences in means, or for group means using the RevMan calculator [18].

Primary outcome
Changes in measures derived from imaging methods such as MRI, ultrasonography, CT, DEXA or plain radiography reflecting morphological changes in affected musculoskeletal tissues following SWT were the primary evaluated outcomes for this review. The measures of effect were pre to post-imaging changes, demonstrating presence, grade, signal intensity or size of the tissue lesion. The strategy for data synthesis employed a quantitative method in the form of meta-analysis depending upon the type of data extracted, alongside narrative qualitative synthesis. The method of evaluation for each study is clearly stated in the relevant tables.

Statistical analysis
Quantitative data analysis was performed using RevMan [18]. Meta-analysis was conducted to calculate pooled effect size for included studies. Continuous data were presented as mean differences (MD) and dichotomous data with odds ratios (OR), including the corresponding 95% confidence interval (CI). Pooling was performed using a random effects model to provide summary effect size owing to expected clinical and methodological heterogeneity. Subgroup analyses were performed separately for each musculoskeletal condition based on SWT type or dosage, control group or use of imaging guidance. Heterogeneity was assessed statistically with χ 2 test and I 2 statistics and significance was set at p < 0.05. Meta-regression was carried out to explain the source of heterogeneity based on important mediating covariates such as SWT dose parameters, baseline lesion size, utilization of anesthesia and imaging guidance (determined a priori).

Study selection and characteristics
The database searches resulted in 789 titles. (Fig. 1). Following the removal of duplicates and exclusion of records based on title and abstract screening, a total of 93 studies were available for full-text review. A total of 30 studies were further excluded with inappropriate study design (predominantly retrospective studies) or lack of imaging outcome as the most frequent reasons for exclusion, leaving a final selection of 63 studies meeting the inclusion criteria. Of the included studies, 30 used an RCT design and the remaining 33 studies were prospective cohort trials.

Description of included studies
The total number of participants in the SWT groups (focused and radial SWT) was 3110. The most commonly evaluated musculoskeletal condition was rotator cuff calcific tendinitis (23 studies), followed by plantar fasciitis (13 studies), femoral head necrosis (12 studies) and fracture nonunion (4 studies). There were a few other individual studies evaluating different musculoskeletal conditions (Fig. 2).

Quality assessment
The risk of bias assessment is summarized in Additional file 2 for included RCTs and prospective trials respectively. The mean score (%) ± SD of the Cochrane risk of bias tool among the 31 RCT studies was 59.68 ± 18.95, while the mean score was 73.61 ± 7.65 for non-RCT studies according to the MI-NORS tool. Items of risk of bias rating across all included RCT and non-RCT studies are shown in Additional file 2. Inter-rater reliability of the risk of bias assessment was calculated using the kappa coefficient (k). The overall agreement between the two primary reviewers was 79.84% (k = 0.53) indicating moderate agreement.

Rotator cuff calcific tendinitis
There were 23 studies (8 RCTs) published between 1995 and 2017 evaluating the effect of SWT on rotator cuff calcific tendinitis. The total sample was 1110 (1141 shoulders) participants. The main imaging inclusion criterion for participant recruitment was the presence of symptomatic type I or II calcification of the rotator cuff based on the Gartner and Simons radiographic classification [19], as detected on radiographs. Detection of glenohumeral or acromioclavicular arthritis, rotator cuff tear or type III calcification were the main exclusion criteria.
Focused type SWT was used in 21 studies, while two studies used radial type SWT. The mean ± SD SWT delivered shocks was 2104 ± 990.57 (1000-6000) and the mean energy flux density (EFD) was 0.26 ± 0.15 (0.02-0.6) mJ/ mm 2 . The mean number of SWT sessions was 2.66 ± 1.91 (1-8) with 10.1 ± 7.26  days interval between sessions. Ten studies used anesthesia prior to SWT application and 14 studies used imaging guidance to target calcium deposits. Further details on included studies characteristics and SWT parameters are presented in Table 1.
Plain radiographs obtained at different time points post SWT was the method used to assess imaging outcomes in 21 studies, while two studies used ultrasonography. Four studies reported the change in calcium deposit diameter (mm), 16 studies reported the number of participants demonstrating total calcification resorption and three reported on reduction or fragmentation of the calcium deposit ( The reduction in calcium deposit size favored SWT compared to placebo but did not reach statistical significance (MD − 11.26 mm (95%CI -24.68, 2.17), P = 0.1) [21,29]. However, the effect of SWT on calcium deposit size was less compared to ultrasound-guided needling (MD 4.25 (95%CI 2.27, 6.24), p = 0.006), (Fig. 4) [26,31]. Total calcium resorption was reported in 222/559 (35%) shoulders. Total calcium resorption was greater in the SWT group compared to placebo (OR 6.38 (95%CI 1.33, 30.70, p = 0.02), but not compared to ultrasound guidedneedling (OR 0.27 (95%CI 0.12, 0.64), p = 0.003), (Fig. 5).

Plantar fasciitis and heel spurs
A total of 13 studies (7 RCTs) published between 2008 and 2018 evaluated imaging changes in plantar fasciitis following SWT. The total number of included participants was 359 (365 heels) diagnosed with symptomatic heel pain in addition to imaging features of plantar fascia thickening (> 4 mm) as the main inclusion criteria. The main exclusion criteria were the presence of systemic disease such as rheumatoid arthritis, diabetes mellitus, vascular abnormalities or neurological impairments.
Eight studies utilized focused-SWT, radial-SWT was used in three studies and one study employed a combined focused and radial SWT type stimulus. The type of SWT was not reported in one study [43]. The mean SWT shocks per treatment session was 2426.92 ± 965.36 (1000-4000) with a mean EFD of 0.22 ± 0.14 (0.03-0.42) mJ/mm 2 . The mean number of treatment sessions was 3.04 ± 1.67 (1)(2)(3)(4)(5)(6)(7)(8) sessions with a mean of 24.79 ± 43.56  days between sessions. Only two studies used anesthesia before the application of SWT and four studies employed sonographic guidance to localize the area of application (Table 3).      Ultrasonography was the most common method for evaluating the imaging outcomes as reported in nine studies, followed by MRI in four studies. The change in plantar fascia thickness (PFT) was evaluated in eight studies as the most frequently reported imaging outcome (Table 4).
The bone resorption effect of SWT on calcaneal spurs was evaluated in three studies (1 RCT) with a total of 461 participants. Chronic symptomatic heel pain (> 3 months) and radiological evidence of a calcaneal spur were the main inclusion criteria. Radial SWT was used in two studies and one study employed focused SWT. None of the studies utilized radiological guidance and only one study used anesthesia [50]. The range of delivered shocks was 1500-2000 shocks carried in 1-5 sessions with 7-30 days interval between each session (Table 3).
Morphological changes in calcaneal spurs were evaluated using plain radiographs in two studies, while one study [46] measured bone mineral density (BMD) and bone mineral content (BMC) using DEXA scanning. None of the studies reported calcaneal spur fragmentation or significant reduction of the spur dimensions [50,57]. However, both BMD and BMC demonstrated statistically significant improvement following focused SWT after a 12 week follow-up period, indicating an osteogenic effect [46] (Table 4).

Osteonecrosis of the femoral head
The effects of SWT on osteonecrosis of the femoral head (ONFH) were evaluated in 12 studies (4 RCTs) published between 2001 and 2018. These studies included 325 (404 hips) participants with stage I-III ONFH according to the Association Research Circulation Osseous (ARCO) classification as the main inclusion criterion, except for one study [59] that included ARCO stage I only. The main exclusion criteria were late ARCO stages, infection, advanced arthritis, neoplastic disease or blood coagulation disorders.
All of the included studies employed focused SWT with a mean EFD of 0.57 ± 0.06 (0.47-0.62) mJ/mm 2 and a mean of 4867.86 ± 1469.64 (2000-6000) shocks. All of the included studies delivered the treatment over one session, except Vulpiani et al. [60] who provided treatment over four sessions and D'Agostino et al. [59] over two sessions (2-3 days interval). Anesthesia was used in eight studies and radiological guidance was implemented in nine studies to accurately target the site of lesion (Table 5).  Imaging changes of ONFH were measured utilizing MRI in addition to radiography in all included studies to evaluate the lesion size, femoral head congruency, presence of a crescent sign, BME and degenerative changes of the hip joint. The percentage of change in the osteonecrosis lesion size was reported in eight studies ( Table 6).

Discussion
The aim of the current systematic review and meta-analysis was to evaluate changes in the morphology of musculoskeletal structures as measured by imaging following a SWT (focused and radial) intervention. Overall, there was a tendency for SWT to demonstrate morphological changes among most of the included musculoskeletal conditions based on different quantitative imaging methods. These tissue changes tended to favor SWT over placebo or other comparators except for rotator cuff calcific tendinitis that favored ultrasound guided needling over SWT. Interestingly, SWT type (radial and focused) and the therapeutic dosage parameters did not appear to have a significant influence on the evaluated imaging outcomes according to our subgroup and meta-   regression analyses. Also, the utilization of imaging guidance and use of anesthesia had no clear impact on the evaluated imaging outcomes. The baseline size of the lesion was the only factor that explained the heterogeneity in our findings. However, these results should be interpreted with caution due to the relatively small overall number of studies and several potential sources of heterogeneity such as variation in study design, variation in imaging methods and measures, time period for imaging follow-up, high risk of bias and the small number of trials included in the subgroup and metaregression analyses. In addition, SWT device related factors may contribute to the heterogeneity such as different SWT types. Further research is required to clearly determine whether there are differences in response to focused or radial SWT for different conditions. Our meta-analysis data on the effect of SWT on rotator cuff calcific tendinitis were comparable to a recent meta-analysis [4] that reported total resorption occurred more commonly using high-versus low-energy SWT at 3 months (OR: 3.4 (95% CI 1. 35,8.58); p = 0.009) based on 3 included studies of 163 participants. In addition to reporting the chances of total resorption versus control, the current meta-analysis has also reported within and between groups changes in the size of the calcium deposit diameter that have not been reported previously. A small number of studies (5/23) reported quantification of the change in the size  of calcium deposit diameter, which limited the power of the current meta-analysis. Out of the five included studies in the meta-analysis, Cacchio et al. [21] reported the highest rate of total resorption (86.6%) and reduction in calcification deposit size MD = 20.45 mm (18.23, 22.67) at one week follow-up using a radial SWT device. The authors themselves did not expect this high rate of resorption and attributed it to the feature of radial SWT that insures the whole calcification area is included inside the wave propagation area. The only available included comparable study is by De Boer et al. [25] that also used a radial SWT device with similar treatment parameters demonstrating a total resorption of only 7% at six weeks. Our data comparison could provide an explanation related to the initial size of the calcium deposit that was exceptionally high in the Cacchio et al. [21] study (21.3 ± 7.5 mm), which was the only significant predictor in our meta-regression analysis. Larger deposits being more responsive to treatment. According to our meta-analysis data, SWT demonstrated significant reduction in PFT and associated calcaneal BME. The reduction in PFT was greater when utilizing radiological guidance, which might be due to more consistent targeting of the SWT to the affected tissue area and   avoiding surrounding areas. Although one study [55] using radial SWT showed greater reduction in PFT over focused or combined SWT, these results should be interpreted with caution based on the small number of studies included in the subgroup analysis. Despite the observed overall reduction in PFT in our review that can be correlated with the improvement in chronic plantar pain [9], it remains unclear whether the type of SWT generation device is an important factor for providing the best outcomes [2]. Our meta-analysis data revealed non-significant reduction in the lesion size with high-over medium and lowdosage SWT for ONFH. The SWT dose parameters were fairly consistent among all the included studies. This could be attributed to the same research group implementing similar treatment protocols. There are no previously published reviews or meta-analyses reporting on imaging changes following SWT among patients with ONFH to allow results comparisons. However, the observed modest reduction in the size of the lesion could be correlated with a recently published meta-analysis [8] that reported significant differences between SWT and control groups in pain rating and motor function measures.
Overall, the reported studies support changes in morphological features based on imaging findings that may reflect changes in underlying

Limitations of the study
Our aim was to include studies on all kinds of relevant musculoskeletal conditions that had been treated with any type of SWT and reporting any imaging outcomes. We conducted a comprehensive search strategy by including all possible synonyms to avoid missing any potential relevant trials, hence reducing publication bias. To date, we are not aware of any previous systematic reviews that evaluated changes in musculoskeletal conditions based on imaging outcomes following SWT interventions. Publication bias was not evaluated as there were < 10 trials for each included condition; hence, the power of the test would be very low to distinguish real asymmetry from chance. The risk of bias scores were 60% for the RCT studies and 74% for non-RCT studies indicating relatively high risk of bias. Participant blinding, allocation concealment, F-SWT

SWT + alendronate
Radiographs of the hip joint were used to assess the size of the lesion, congruency of the femoral head, the presence of a crescent sign and degenerative changes of the hip joint. MRI was used to examine the size of the lesion, the congruency of the femoral head, the presence of a crescent sign, BME, and degenerative changes of the hip joint 3,6,12, 24 months The mean size of the lesion (%) over the femoral head pre-SWT was 27.7 ± 15.5 and post-SWT was 25.7 ± 16.2 at 6 months follow-up. In contrast, the pre-SWT+ alendronate group was 32.6 ± 19.9 and post-SWT+ alendronate was 29.32 ± 21.99. Significant reduction in BME was noted in both groups.
=0.679 0.145 Wang (2011) [70] F-SWT None Radiographs in AP and lateral views were used to assess the size of the lesion, congruency of the femoral head, the presence of a crescent sign and degenerative changes. MRI was used to evaluate the size of the lesion, the collapse of femoral head and BME.

6,12 months
The mean size of the lesion (%) over the femoral head pre-SWT was 27.23 ± 18.9 and 27.04 ± 19.17 post-SWT. Significant improvement in BME was noted following SWT > 0.05 for the size of lesion and = 0.04 for reduction of BME NA BME bone marrow edema, F-SWT focused SWT, R-SWT radial SWT, NR not reported, NS not significant, NA not applicable study size calculation and other sources of bias (defined according to imaging assessment accuracy) were the lowest scored items. Of particular interest to this review, the accuracy of imaging measurements was questionable in some cases due to the under-reporting of details pertaining to measurement standardization. It was decided that meeting a minimum of two out of an a priori set of four criteria related to imaging accuracy would be used for judging risk of bias based on imaging accuracy. These criteria were based on providing details on the experience or specialty of the radiologist in musculoskeletal imaging, details of the imaging procedure to insure participant's consistent position during all image acquisition, prior testing or training of the assessor to ensure reliability and reporting the score of measurements based on the average of multiple measurements.

Future research recommendations
Current available research has provided preliminary evidence related to the capacity of SWT to influence underlying pathophysiological processes in various musculoskeletal conditions as demonstrated through changes in imaging. However, considering more standardized and reliable quantitative imaging measures as a primary outcome would be warranted in future research. This can be achieved through improving the imaging outcomes assessment methodology to ensure consistent and valid reporting based on our suggested criteria for imaging assessment accuracy. Adopting such criteria can limit the imaging assessment procedure variations as it is challenging to account for it as a covariate in the intervention effect estimate. Imaging endpoints are recommended to be specified and reported to evaluate short, medium and long term changes. Study sample sizes should be calculated based upon imaging parameters as a primary outcome. It would also be of great value if researchers could reach consensus on the optimal imaging modality and relevant imaging measures for each musculoskeletal disorder. Consistency of approach would significantly improve the quality of research. It was surprising that our comprehensive search strategy did not identify any studies using imaging outcomes for commonly treated tendinopathies such as patellar, proximal hamstring and rotator cuff noncalcific tendinopathies and identified only a few trials for Achilles and wrist extensor tendinopathies. This is an indication of the limited use of imaging as a measure of outcome in addition to the usual clinical outcome measures for this type of condition. Whilst clinical outcomes are clearly of primary importance, imaging does provide a window to assist us in understanding the effects and potential mechanism of action of SWT. Future trials might consider making