Design
The study design was a single-assessor-blinded (i.e., data analysis) controlled clinical parallel-arm trial with an intervention phase of 12 weeks, registered under ClinicalTrials.gov (NCT02732782). The study was conducted according to the CONSORT (Consolidated Standards of Reporting Trials) guidelines [42] (Fig. 1). We recruited patients by physician referral or advertisement (from 05/16 to 12/17). The pre-screening (i.e., initial medical assessment and diagnosis) by medical doctors took place at the sports medicine outpatient clinic of the Charité University Hospital, Berlin. Except for magnetic resonance imaging (MRI), all assessments and evaluations at baseline (PRE) and after completion of the intervention phase (POST) were conducted within the lab facilities of the Department of Training and Movement Sciences at the Humboldt-Universität zu Berlin (G.R., S.B. and K.P.) (Fig. 2). The assessment 6 months after POST (FOLLOW-UP) was performed online.
Eligibility criteria were assessed by researchers (G.R., S.B. and K.P.) within the laboratory facilities and orthopedic physicians from the sports medicine outpatient clinic (Charité University Hospital, Berlin) (Fig. 2). Inclusion criteria were male, aged between 20 and 55 years, and chronic condition (> 3 months) of Achilles tendinopathy. Pathology had to be confirmed via ultrasound (at least discrete hypo-echogenic areas within the tendon) and clinical assessment by a medical doctor. The threshold level of severity was defined by the Victorian Institute of Sport Assessment for Achilles tendinopathy (VISA-A) score of less than 80 points [43]. Exclusion criteria were corticosteroid infiltration of the tendon or any intake of antibiotics (such as Fluoroquinolone, Levofloxacin, Ciprofloxacin) [44] within the past 12 months, any leg surgery, tendon rupture or signs of partial rupture, any systemic inflammatory condition (e.g., rheumatoid arthritis, diabetes) and any spondyloarthropathies (e.g., spondylitis ankylosans). In cases of bilateral symptoms, the leg with a lower clinical (i.e., VISA-A) score and higher pain level was chosen. As there have not been any studies assessing tendon stiffness and Young’s modulus, particularly in Achilles tendinopathy using our high-loading tendon protocol, we relied on prior statistical power analysis in two studies with healthy subjects [26, 27]. This power analysis calculated a sample size of at least n = 12 per group to achieve high statistical power (α = 0.05, power = 0.95, correlation = 0, effect size: stiffness 1.6, Young’s modulus 1.2) [28]. Anticipating a dropout of approximately 20%, we decided to aim for 15 patients per group.
Fifty-five males registered interest and seven of them were excluded during the enrollment process (Fig. 1). Forty-eight participants met the eligibility criteria. They were enrolled in the study and after full completion of the PRE measurements (G.R., S.B. and K.P.) assigned by G.R. to one of the three treatment groups by an ABC pattern based on the order of date of pre-screening: The passive therapy group obtaining passive therapy sessions (i.e., no lower-limb mechanical loading), the standard exercise treatment (i.e., Alfredson group performing home-based eccentric exercise) and the High-load group conducting home-based high-loading tendon exercise. Four participants dropped out during the intervention phase (dropout rate of 8.33%). The remaining 44 participants were allocated as follows: Passive therapy group (n = 14), Alfredson group (n = 15) and High-load group (n = 15) (Table 1).
Allocation and Blinding
The allocation sequence list was generated, possessed, and hidden only by the study organizing researcher (G.R.). Except for him (G.R.), the allocation sequence was concealed for every other person involved in the enrollment, allocation, and baseline assessment process (i.e., medical doctors, researchers, assessors, data analysts and patients). Only after having finished baseline assessments PRE T1–T3, the respective assessor was informed by the researcher (G.R.) about the forthcoming allocation (Fig. 2). The chronological order of the patients at PRE-SCREENING did not correspond to the chronological order of PRE T1-3. Thus, allocation-to-group was not predictable by chronological order at any timepoint from PRE T1 -3. Allocation to groups, PRE measurements, supervision and POST measurements were carried out with strict adherence to standardized assessment procedures to all groups equally without disclosure of our study hypotheses. All PRE and POST assessed and reported data were gathered anonymously and without allocation information, and thus all image processing and data analysis was blinded.
Intervention
All patients equally received the option for 12 therapeutic appointments (i.e., manual therapy, tissue and/or joint mobilization) as a prescription according to the national medical guidelines with free choice of location. The prescription stated the clinical trial involvement with recommendations for the physiotherapists to apply passive treatments and to refrain from active plantar flexor strength training, especially excluding eccentric exercises during the intervention phase. These recommendations were additionally forwarded to the participants by a letter to the physio. Patients were thoroughly informed about the intervention modalities by an experienced physiotherapist (G.R.) in addition to receiving a hand-out with clear and detailed instructions of the intervention protocol. During the intervention phase, patients were monitored and supervised on week 1, week 2, week 4, week 8 and week 11 via telephone call and/or email to ensure compliance (G.R. and K.P.). Moreover, all patients received a training diary for daily recording intervention training frequency, training load and its progression, pain on an NRS scale from 0 to 10 as a mean per day, use of medication, physiotherapy frequency and treatment content, further exercise activities (in km or in min) and additional comments. The activity was assessed in hours per week and based on verbal information (i.e., at PRE T1) as the baseline value (average of the two weeks before PRE T1) and based on the training diary (for PRE and POST). In case of ambiguity (i.e., distance report instead of time within the diary), we converted 10 km of running and 30 km of biking to 1-h activity. Activities like slow strolling, swimming, hiking, skiing, and slow commuting by bike were excluded from activity quantification. Compliance was defined as the percentage of the prescribed intervention.
All patients were allowed to continue with their training habits unless it induced Achilles tendon pain with a level of > 3/10 (NRS scale) up to 24 h later. No additional strength training of the plantar flexors and no implementation of any new sort of lower body training was permitted.
Passive therapy group: Patients were asked to adhere to a maximum of 12 passive therapeutic and manual treatment sessions, while refraining from any explicit plantar flexor strength training or Alfredson eccentric exercise protocol for the time of intervention. Thus, full adherence was defined as having had 12 appointments.
Alfredson group: According to the commonly known and frequently published protocol [33], patients in the Alfredson group performed eccentric exercises in an upright standing position with only the forefoot of the injured leg on the edge of a stair lowering the heel with an eccentric phase of three seconds [45]. We ensured eccentric-only contractions of the plantar flexors by using the healthy leg to return to the start position. Moreover, the use of a full ankle angle of motion in the eccentric phase was encouraged. One session was defined as three sets of 15 repetitions with extended knees followed by another three sets of 15 repetitions with bended knees and 1-min rest in between sets. According to Alfredson, our protocol consisted of two sessions per day every day with no warm-up. The optional level of load progression was defined by a 5 kg additional load per week.
High-load group: Patients in the High-load group obtained a feedback-fitted sling (displaying the applied force due to an integrated strain gauge) for home-based application [41] of the high-loading protocol reported by Arampatzis et al. [26], which provides an efficient stimulus for tendon adaptation [27, 28, 46, 47]. Patients were instructed to sit on the floor with extended knees. The forefoot (with shoes) was placed in the foot plate. The ratchet was individually set and fixed as tightly as possible, to allow for maximal isometric plantar flexor contractions at a standardized ankle angle position (90°). For warm-up, the patients performed three sets of five isometric submaximal plantar flexor contractions with 3 s under tension followed by three sec of rest each and with a rest of 1-min in between sets. After the warm-up in the first supervised session, five isometric MVCs of the plantar flexors were executed. The individual training load was then calculated based on 90% of the mean of the five MVCs. After 10 min of rest, the first high-loading intervention exercise was conducted under supervision with five sets of four repetitions of 90% isometric MVC plantar flexor contractions with 3 s under tension followed by 3-s rest between repetitions and 1-min rest between sets. This training session was repeated four times per week for 12 weeks. The level of load progression was defined by ~ 5% of the individual training load per week.
Alfredson group and High-load group: Only the injured leg was trained. The following instructions were given referring to pain and load progression: No progression of training load within the first two weeks. The load can only be progressed once a week. Adapted by the pain-monitoring model from Silbernagel and Crossley [48], load progression was allowed when the pain level was < 6/10 (NRS scale) [40, 49] and individual rating of perceived exertion was < 3/10 (NRS scale) [48]. Load reduction was recommended when either the pain level was > 5/10 (NRS scale) [40, 49] or the individual rating of perceived exertion was > 5/10 (NRS scale) [48] and should be maintained for a whole week. In case of limited (impossible) load reduction, the repetition number and/or frequency should be reduced. The first session was supervised.
Primary Outcomes
Primary outcomes were defined as the mechanical, material, and morphological Achilles tendon properties including plantar flexor muscle strength as well as the VISA-A score and pain.
Mechanical and Material Properties
Tendon mechanical and material properties (i.e., stiffness, CSA, Young’s modulus) were analyzed using dynamometry, electromyography (EMG), motion capture, ultrasonography and MRI.
For tendon stiffness assessment, patients were seated on a dynamometer (Biodex System 3, Biodex Medical Systems Inc., USA) with a fixed ankle angle in a neutral position (foot sole 90° perpendicular to the tibia), extended knee, hip angle of ~ 110° and the pelvis fixed with a rigid belt. After a standardized warm-up of up to ten moderate to submaximal voluntary isometric plantar flexor contractions and 1–3 MVCs [50], the patients conducted five ramped MVCs in order to achieve high reliability on tendon stiffness measurement [51] with a duration of five seconds each and 2-min rest between repetitions followed by 2–3 isometric plantar flexor MVCs with 2-min rest between repetitions. During all MVCs, standardized verbal encouragement during each attempt was given.
Stiffness was calculated based on the tendon force to tendon elongation ratio. To calculate Achilles tendon force, the plantar flexion moment was divided by the tendon lever arm. We measured the Achilles tendon lever arm with the tendon excursion method [52] by relating the displacement of the m. gastrocnemius medialis myotendinous junction (MTJ) assessed by B-mode ultrasound (7.5 MHz, My Lab60, Esaote, Genova, Italy) to the corresponding angular ankle joint excursion [53]. Changes in lever arm length during the contraction when compared to resting state were considered by including a corrective factor in our calculation [54].
The plantar flexor moment was calculated using an inverse dynamic approach taking the misalignment of the ankle joint axis to the dynamometer axis into consideration [55]. The inverse dynamic calculation was based on kinematic data from an infrared motion capture system (Vicon Nexus, version 1.7.1, Vicon Motion Systems, UK) integrating seven cameras operating at 250 Hz. The contribution of the antagonistic muscle to the plantar flexor moment was considered by determining the m. tibialis anterior activity during plantar flexor MVC with one pair of bipolar surface EMG electrodes (Myon m320RX, Myon AG, Switzerland, 1000 Hz) [56]. The antagonistic moment was estimated based on the relationship of the m. tibialis anterior EMG activity and the exerted moments during two submaximal isometric m. tibialis anterior contractions with slightly lower and higher activity than the m. tibialis anterior ramp contraction activity [56].
Achilles tendon elongation was assessed by placing a B-mode ultrasound probe within a custom-built foam cast that was fixed on the lower leg recording the MTJ displacement during the ramped MVCs. MTJ displacement was traced manually frame-by-frame within a custom-written MATLAB script (The MathWorks, version 2012, USA). To consider the effects of ankle joint motion on tendon elongation measurements, the passive displacement of the MTJ in relation to the ankle angle [57] was determined with five trials of slow passive (no muscle contraction) ankle joint motion over the full range of motion. Force and elongation data from five measurements each were averaged.
We calculated Achilles tendon stiffness as the slope of a linear regression of tendon force to tendon elongation between 50 and 100% of the maximum tendon force. Achilles tendon rest length was measured at 20° plantar flexion with extended knee [58] from the proximal posterior part of the tuber calcanei to the MTJ. We calculated the Young’s modulus of the Achilles tendon by multiplying tendon stiffness with the quotient of tendon rest length and tendon CSA. The stress of the Achilles tendon was determined as the ratio of tendon force and averaged Achilles tendon CSA (see chapter 2.3.2). Maximum tendon strain was calculated as the ratio of maximum tendon elongation (obtained during the ramp MVCs) to rest length.
Morphological Properties
We assessed the CSA of the free Achilles tendon either with a 0.25 T magnetic resonance imaging (MRI) scanner (G-Scan, Esaote, Italy) [3D hybrid contrast enhancement (HYCE) sequence, repetition time (TR) 10 ms, echo time (TE) 5 ms, flip angle 80°, slice thickness 3 mm, space between slices 0.4 mm] at Oscar-Helene-Heim Foundation, Helios clinic Emil von Behring, Berlin, Germany, or a 1.5 T MRI scanner (Siemens Avanto, Siemens, Germany) at the Institute for radiology, Charité – University Medicine, Berlin) [T1- weighted sequence, TR 460 ms, TE 20 ms, slice thickness 2 mm, space between slices 0.4 mm] using transversal and sagittal Achilles tendon scans. Every PRE-POST pair was analyzed with the same scanner. During MRI measurement, the standardized patient position was the supine position with extended hips and knees and the ankle joint fixed at 90°. The transversal scans were positioned perpendicular to the direction of the tibia and manually and assessor-blinded segmented using the software OsiriX (Pixmeo SARL, version 2.5.1, Switzerland) [59]. The sagittal scans were used to precisely determine the proximal (i.e., m. soleus–Achilles tendon junction) and distal (calcaneus bone insertion) end of the free Achilles tendon and its length. The length of the free Achilles tendon was calculated as a curved line through every digitalized transversal scan using a Delaunay triangulation [28]. Tendon cross-sectional area was determined in 10% increments across the whole free tendon length. Average Achilles tendon CSA was calculated as a mean of all assessed CSAs of the free Achilles tendon.
Clinical Outcomes
As a patient-reported outcome measurement (PROM), we used the reliable and valid VISA-A score [60] to determine clinical severity. The VISA-A score was evaluated PRE (in-person), POST (in-person), and FOLLOW-UP (online). According to Stevens and Tan [61], a minimum clinically important difference of 15 points (pts.) was considered clinically significant. As another clinical PROM, the pain was assessed based on daily numerous rating scale (NRS) (0–10 pts.) recordings in the patient diary. We calculated PRE values from the mean of the first 14 days after baseline and POST values from the mean of the last 14 days of the intervention phase.
Secondary Outcomes
Secondary outcomes were defined as functional parameters (i.e., vertical jump height) and tendon vascularity.
Functional Properties
Functional properties were assessed by estimating drop jump (DJ) and countermovement jump (CMJ) height as described previously [41]. Briefly, after a warm-up with up to 12 jumps of low to moderate intensity, five maximum effort CMJs and five DJs were performed (bare feet, hands akimbo, 1-min rest between repetitions). DJs were performed from a 15 cm box. Ground reaction forces were measured with two separate force plates at a rate of 1000 Hz (Kistler, Type 9260AA, 600 × 500 × 50 mm, Switzerland) linked to an analog digital converter (DAQ-System, USB 2.0, Type 5691A1). Data were recorded (BioWare Software, Type 2812A) and jump height was calculated based on the impulse–momentum method [62] for the CMJ and the flight-time method [63] for the DJ, using a custom-written MATLAB interface (version R2012a; MathWorks, Natick, MA, USA). For further CMJ and DJ analysis, the mean of the highest three jumps out of five attempts was used.
Vascularity
Intratendinous vascularity was assessed in the sagittal plane with pulsed-wave power Doppler ultrasonography (7.1 MHz, My Lab60, Esaote, Genova, Italy) using the following settings: Wall filter 1, Density 1, Persistence 3, Pulse repetition frequency 750 Hz [36, 64]. Power and color gain were manually adjusted just below random noise level per participant [65] and recorded at PRE measurement to reapply for POST measurement. The transducer was aligned parallel to the tendon and applied with no pressure. The power Doppler transducer was positioned so that it visualized the proximal portion of the calcaneal bone and the Achilles tendon (Fig. 3). During measurement, participants were in the prone position with the knees extended and the ankle joint passively stabilized at 90° (i.e., tibia perpendicular to the foot) with relaxed plantar flexor muscles. Prior to the measurement, participants were advised to refrain from any intense exercise activity for 2 h [66]. For optimized visualization of tendon vascularity, three sets of 15 unilateral single-leg heel raises were performed before the measurement [67]. Three scans with a duration of 4 s each were recorded. The frame with both the highest signal activity and without any artifacts [68] was chosen for analysis. Analysis was performed using a custom-written MATLAB script (The MathWorks, version 2012, USA) quantifying the number of intratendinous colored pixels and converting them to mm2 (Fig. 3).
Statistical Analysis
We examined the normality of data distribution with the Kolmogorov–Smirnov test. For all baseline between-group comparisons, we performed a one-way analysis of variance (ANOVA) (factor: group). For all PRE to POST comparisons except tendon CSA, a repeated measures ANOVA was conducted (within-subject factor: time; between-subject factor: group). For the PRE to POST comparisons of tendon CSA, we used a two-factor repeated measure ANOVA (factor 1: time; factor 2: localization, considering the 10% steps of the full free Achilles tendon length; between-subject factor: group. In case of a significant effect of interaction, a Bonferroni post hoc analysis was conducted, and adjusted p values were reported. The effect size concerning the effect of training was based on partial eta squared and calculated by Cohen’s f [69] and defined as follows: values of f = 0.10, f = 0.25, and f = 0.40 represent small, medium, and large effect sizes, respectively. The relationship of the dominant to the injured side was established with an adjusted Pearson’s contingency coefficient. For all statistics, the significance level was set at α = 0.05 and the software SPSS Statistics (IBM, version 21, USA) was used.