Tendon morphology and mechanical properties assessed by ultrasound show change early in recovery and potential prognostic ability for 6 month outcomes

Abstract

Purpose:

Optimizing tendon structural recovery during the first 12 weeks after Achilles tendon rupture is a prime target to improve patient outcomes, but a comprehensive understanding of biomarkers is needed to track early healing. The purpose of this study was to observe healing of tendon structure over time using ultrasound based, tendon-specific measures and to identify relationships between tendon structural characteristics and clinical measures of gait and strength.

Methods:

Twenty-seven participants (21 male, mean(SD) age 39(11) years) were assessed at 4,8, 12, and 24 weeks after injury or surgery using ultrasound imaging techniques. Gait analysis and strength testing were added at the later time points.

Results:

Ruptured tendons had significantly lower dynamic shear modulus (p < 0.001), greater tendon cross-sectional area (p < 0.001), and greater length (p < 0.001) than the uninjured side. Dynamic shear modulus, cross sectional area, and length were found to increase over time (p<0.01). Tendon structure at 4 weeks post-injury [cross sectional area symmetry (r=0.737, p=0.002) and dynamic shear modulus (r=0.518, p=0.040)] related to stance phase walking symmetry at 24 weeks.

Conclusions:

Tendon structure assessed by ultrasound imaging changes over the first 24 weeks of healing after Achilles tendon rupture, suggesting it could be used as a biomarker to track tendon healing early in recovery. Additionally, tendon structure within the first 12 weeks relates to later walking gait and heel-rise symmetry, which may indicate that tendon structure could have prognostic value in the care of these patients. This study’s clinical relevance is in its support for using ultrasound imaging to assess early patient healing and prognosticate later patient outcomes after Achilles tendon rupture.

Level of Evidence:

Level 2, prospective cohort prognostic study

Introduction

Achilles tendon rupture can cause focal, longstanding plantar flexor deficits[5, 18, 20, 23, 29] resulting in altered running and jumping biomechanics in the long term[6, 42]. Evidence suggests that tendon structural changes, such as elongation, develop early after injury [21, 26] and underlie some of these plantar flexor deficits, including triceps surae atrophy[17], impaired calf strength and endurance[6, 17, 36], altered muscle function during gait[38], and asymmetrical ankle biomechanics during running and jumping[6]. As rehabilitation moves from time-based to criterion-based progression[19, 24], a comprehensive understanding of biomarkers will be needed to help track tendon healing within the first 12 weeks following injury. Tendon structural characteristics like tendon morphology[15, 21, 25, 32, 36, 43] and mechanical properties[1, 13, 15, 32, 43], may help fill this need for early biomarkers of tendon recovery and provide important information to the clinician regarding tendon response to treatment.

Measurement of tendon structure has its challenges. For example, prior studies investigating tendon mechanical properties have often depended on voluntary muscle contraction[1, 15, 28, 33], and assume the total force is generated by the triceps surae. This assumption may be acceptable in healthy populations, however, it is likely flawed in the context of tendon injury. Prior studies have suggested that the flexor hallucis longus compensates for dysfunction of the triceps surae in individuals following Achilles tendon rupture[12, 17]. Therefore, in the context of assessing tendon mechanical properties, these techniques likely overestimate the force output of the triceps surae, and consequently overestimate values for tendon stiffness and Young’s modulus in unhealthy tendons.

The advancement of ultrasound imaging techniques may provide improved ability to identify early biomarkers of tendon structural healing. Shear wave elastography techniques have been developed and applied to individuals following Achilles tendon rupture[43–45]. These passive techniques do not require volitional contraction, making them safe to use earlier following rupture and potentially less prone to overestimation of tendon mechanical properties. Additionally, with the development of extended field of view ultrasound imaging[30, 35], tendon length measurement has become cheaper and more clinically viable than MRI[10, 17, 29] and ultrasound-with-motion-capture[36].

The purpose of this study was to track Achilles tendon healing over time using ultrasound based, tendon-specific measures. It was hypothesized that tendon geometry and mechanical properties assessed using ultrasound-based techniques would show side-to-side differences and change over time. In order to investigate the prognostic ability of these structural parameters, we secondarily aimed to identify relationships between early tendon structural characteristics and clinical measures of gait and strength later in recovery. It was hypothesized that improved tendon structural symmetry in the first 12 weeks post-injury would positively relate to improved gait symmetry at 24-weeks post-injury.

Materials and Methods

All procedures performed were in accordance with the ethical standards of the University of Delaware Institutional Review Board under approval ID 784188–9. Informed consent was obtained from all participants included in the study.

Participants within the first month following unilateral Achilles tendon rupture were recruited for this study from local orthopaedic practices. To be included, participants needed to have a unilateral Achilles tendon rupture managed operatively by non-augmented repair or nonoperatively. Individuals were excluded if they were under the age of 18 or had a history of collagen disorders, peripheral neuropathy, or peripheral vascular disorders. Participants were prospectively followed longitudinally, with assessments at 4, 8, 12, and 24 weeks following surgery or injury if managed non-surgically. All testing was performed on the right leg first in order to quasi-randomize injured sides. All assessments were performed by a single examiner. Because this study was somewhat exploratory in its methods, we did not have previously published data to perform an a priori power analysis. Prior studies reported significant change over time using similar methods[45] along with relationships between tendon structure and functional performance[33] were used to estimate sample size. These studies[45, 33] included 20–26 participants, so the present study aimed to include 26 participants.

Standardized treatment was not provided as part of this study, so study participants received care at the discretion of their treating physicians and physical therapists. Participants were interviewed to record the time spent non-weight bearing and time spent immobilized.

Measurement of tendon structure

Tendon structure was assessed at all time points. Tendon morphology was measured using B-mode ultrasound imaging at 10 MHz. Tendon length from the calcaneus to the gastrocnemius myotendinous junction was measured as described by Silbernagel, et al. [ICC = 0.955, standard error of measurement (SEM) = 0.67 cm, minimal detectible change at group level (MDC_95% group) = 0.43 cm] (GE Logiq e, GE Healthcare, WI, USA)[30, 35]. Tendon cross sectional area was measured by placing the ultrasound probe at the site of rupture (or comparable location on the uninjured side) in short axis (ICC = 0.985, SEM = 0.02 cm2, MDC_95% group =0.01 cm²). (Due to tendon lengthening, there were instances that the same exact location could not be used on both sides as it was no longer in the area of the free tendon on the uninjured side.) The distance between the calcaneal notch and the region of interest both marked on the skin was taken using a tape measure in order to ensure the same site was assessed over time.

Tendon mechanical properties were quantified using continuous shear wave elastography (cSWE) as described by Cortes, et al(2015). cSWE is an ultrasound technique that uses an external actuator to propagate a shear wave along the length of the tendon. An ultrasound probe at the region of interest images the linear displacement of the tissue, allowing the speed of wave propagation to be calculated. Using a biomechanical model, shear modulus (ICC = 0.67, SEM = 9.4 kPa, MDC_95% group = 6.0 kPa) and viscosity (ICC = 0.80, SEM = 6.0 Pa*s, MDC_95% group = 3.8 Pa*s) are estimated. The novelty of this technique compared to shear wave elastography is that cSWE overcomes concerns regarding saturation and allows for separating out rate dependent (viscosity) from non-rate dependent (shear modulus) properties[9]. The average of three trials was used in data analysis. This technique has been previously validated[9] and shown to be reliable[39, 43].

Prior studies using cSWE have applied the technique to Achilles tendons following rupture[43, 44], however, this is the first study to use cSWE in ruptured tendons in individuals who are still immobilized. Therefore, the foot position of the participant was modified for safety purposes. At the 4-week time point, the participant was positioned in prone with up to 4, 1.1 cm high wedges that approximated the amount of wedging their healthcare provider recommended using in their walking boot. This same foot position was repeated at all following time points.

To improve the comparison to other work that has used shear wave elastography, a combined measure of dynamic shear modulus (at 400 Hz) (ICC = 0.77, SEM = 23.1 kPa, MDC_95% group = 14.3 kPa) was calculated from the shear modulus and viscosity using the following equation:

$$\text{μ}\left(\omega \right)=\frac{2\left({\text{μ}}_1^2+{\left(2\pi f\right)}^2{\text{μ}}_2^2\right)}{\left({\mu }_1+\sqrt{{\text{μ}}_1^2+{\left(2\pi f\right)}^2{\text{μ}}_2^2}\right)}$$

Where μ(ω) is dynamic shear modulus, μ₁ is shear modulus, μ₂ is viscosity, and f is the frequency (400Hz in this case). Dynamic shear modulus likely captures a more comprehensive picture of tendon mechanical properties as shear modulus may be more associated with fibrous content and viscosity may be more associated with fluid content.

Functional performance assessment

Gait was assessed in the laboratory using laser gaits along with inertial measurement units (IMUs) (MuscleLab™, Ergotest Technology, Norway) strapped to the dorsum of the participant’s shoe. Participants walked at their “usual, comfortable pace” on a 10 meter walk test, of which the middle 6 meters were assessed. Laser gates measured gait speed and IMUs quantified percent of gait cycle spent in stance along with percent of stance time spent in loading response, foot flat, and pre swing. Gait speed was assessed at 8, 12, and 24-week time points. Stance temporal parameters were assessed at 12 and 24-week time points once participants were able to walk in standard street shoes as the IMUs have not been validated when used with an orthopaedic boot. Stance timing asymmetries were defined as the difference between injured and uninjured sides (% stance difference = % stance uninjured − % stance injured). The average of three gait trials was used for analysis.

In order to get a sense of the loads placed on the tendon, physical activity was assessed at the 12-week time point (between days 80–90) using a Fitbit Zip™ step counter. Wear time was not available, so a minimum daily step count of 1500 steps was used to define a valid day. This step count was a conservative estimate based on a previously reported basal daily step count of 2500 steps in healthy adults[40]. Step counts were originally intended to be collected throughout the first 12 weeks of recovery, but reliable data was unable to be captured early in recovery as participants walked with different assistive devices that did not consistently trigger the step counter. Therefore, only step counts of valid days when participants were no longer using assistive devices or orthopaedic boots were included. Physical activity was also self-rated by participants at all time points using a 6-point physical activity scale (PAS)[16]. This scale goes from 1–6, with higher values indicating higher levels of activity.

Calf muscle endurance was assessed using a heel-rise test as described by Silbernagel, et al[34, 36] at the 24-week time point. In this test, participants stand on a 10 degree slant box and perform unilateral heel-rises at a cadence of 30 per minute until fatigue. A linear encoder (MuscleLab™, Ergotest Technology, Norway) taped to the participant’s heel was used to measure the maximum heel-rise height and total heel-rise work (= total linear displacement * body weight). Performance on the heel-rise test has been found to relate to improved functional prognosis[6]. Limb symmetry indexes (LSI (%) = injured/uninjured value *100) were calculated for maximum heel-rise height and total heel-rise work and used in data analysis.

Demographics were collected and participant self-reported function was measured using the Achilles tendon Total Rupture Score (ATRS)[27]. This is a 0 to 100 scale score where higher self-reported function is indicated by a higher score. The ATRS was used for descriptive purposes and is included along with participant age, sex, body mass index (BMI), and side of injury.

Statistical analysis

All statistical analysis was performed using IBM SPSS Statistics software (IBM, Armonk, New York, USA). Descriptive statistics were calculated for all variables. Data was inspected to ensure it fit the assumptions of parametric statistical testing. To determine change in tendon structure and gait parameters over time, within subject Analysis of Variance (ANOVA) was used; post hoc pairwise comparisons were made with a Bonferroni correction. Relationships of tendon elongation and dynamic shear modulus/cross sectional area LSIs with gait/strength outcomes are reported using Pearson correlations. For gait analysis, paired t-tests were used to compare the injured and uninjured sides. Independent samples t-tests were used to identify any differences in relative tendon structural parameters (i.e. tendon cross sectional area LSI, dynamic shear modulus LSI, and elongation) between individuals able and unable to return to prior level of activity on the PAS at 24 weeks.

One interest in this study was to describe the general population of individuals after Achilles tendon rupture, so participants managed both surgically and non-surgically were included. This decision was based on prior studies reporting equivalent functional outcomes in individuals irrespective of initial treatment strategy[37]. To ensure that including both surgically and non-surgically managed participants did not change the outcome of the study the analysis was repeated including only those participants who were managed surgically.

Results

Description of participants

Demographics for the 27 included participants are listed in Table 1. Participants were managed surgically by 10 local surgeons and non-surgically by a single rehabilitation group. Four surgically managed participants had incisions less than 3.4 cm in length suggesting a minimally invasive repair[7], the remainder had incisions greater than 5 cm in length suggesting an open repair. Five participants missed the 4-week time point (2 due to wound healing concerns, 2 due to scheduling difficulties, and 1 was able to attend but was still casted so could only complete questionnaires). One participant missed the 12-week time point due to traveling out of the country. Two participants were lost to follow-up – one after the 4-week and one after the 12-week time point. Therefore, data from 22 participants are included in the 4-week time point, data from 26 participants are in the 8-week time point, and data from 25 participants are in the 12- and 24-week time points.

Table 1.

Participant Demographics

Demographic	Mean(SD)/Frequency
Age	39(11) years
Body mass index	27.9(4.4) kg/m2
Sex	21 male, 6 female
Injured side	13 right, 14 left
Initial treatment	22 surgical, 5 non-surgical
Weight bearing onset	39(22) days
Time of discontinuation of immobilization	69(16) days
ATRS score - 24 weeks	77(18)

Tendon structural change over time

Descriptive statistics for tendon structure and results of the ANOVA are displayed in Table 2. Ruptured tendons had significantly lower shear modulus, viscosity, and dynamic shear modulus compared to uninjured tendons with a difference that exceeded measurement error (Table 2, Figure 1). Change in dynamic shear modulus exceeded the MDC between 8–12-week time points on the ruptured side and between 12–24-week time points on the uninjured side.

Table 2:

Tendon structural change over time. * indicates p < 0.05

	4 weeks		8 weeks		12 weeks		24 weeks		P-values and effect size (partial eta squared) for main effects
Tendon Structural Characteristic	Injured	Uninjured	Injured	Uninjured	Injured	Uninjured	Injured	Uninjured	Time	Side	Time*side
Shear Modulus (kPa)	64.8(13.7)	89.0(19.0)	77.2(17.2)	96.3(19.6)	71.2(18.8)	94.9(17.7)	85.0(18.1)	96.0(17.6)	n.s. pη2 =0.100	<0.001* pη2 =0.635	n.s. pη2 =0.017
Viscosity (Pa*s)	23.1(9.8)	47.4(8.7)	25.5(11.0)	46.2(14.7)	28.7(16.1)	47.4(12.7)	28.1(9.8)	47.1(12.3)	n.s. pη2 =0.088	<0.001* pη2 =0.761	n.s. pη2 =0.035
Dynamic Shear Modulus (kPa)	102.7(32.1)	187.3(33.6)	115.5(36.9)	188.3(49.5)	138.3(40.5)	207.0(32.7)	126.8(33.8)	190.1(44.1)	0.003* pη2 =0.266	<0.001* pη2 =0.844	n.s. pη2 =0.030
Cross-sectional area (cm2)	1.92(0.94)	0.69(0.36)	2.23(1.16)	0.58(0.12)	3.07(1.23)	0.59(0.14)	3.31(0.94)	0.59(0.15)	<0.001* pη2 =0.539	<0.001* pη2 =0.806	<0.001* pη2 =0.526
Length to gastrocnemius (cm)	21.7(2.9)	20.6(2.4)	22.0(2.9)	20.7(2.7)	22.2(2.8)	20.6(2.5)	22.3(2.7)	20.6(2.7)	0.003* pη2 =0.227	<0.001* pη2 =0.626	n.s. pη2 =0.100

Figure 1.

Change of tendon cross sectional area (A), length (B), and dynamic shear modulus (C) between 4 and 24 weeks from injury or surgery.

Ruptured tendons had larger cross sectional area than uninjured tendons, and side to side differences exceeded measurement error at all time points (Figure 1). Change in tendon cross sectional area exceeded the MDC on the ruptured side between each time point. Ruptured tendons were longer than uninjured tendons, which exceeded measurement error, and length increased over time (Figure 1).

Functional performance assessment

Gait speed increased significantly between 8 and 24 weeks (p < 0.001) with significant pairwise comparisons between 8–24 (p = 0.002) and 12–24 (p=0.001) weeks (Table 3). Asymmetries were observed in stance time, loading response, and pre-swing (Table 3). Participants reported a pre-injury PAS score of 5.3(0.9) that was significantly higher than at 24 weeks post-injury (p < 0.001) (Table 3). Ten participants reported a return to the same level of activity as pre-injury on the PAS. Participants walked a mean of 5,789(SD: 2,014, range: 2,682–9,506) steps per day at 12 weeks. There was no relationship observed between step counts and any relative tendon structural parameter at any time point.

Table 3:

Stance phase gait parameters at 12 and 24 weeks. * indicates p <0.05

	4 weeks	8 weeks	12 weeks			24 weeks
Gait Parameter			Injured	Uninjured	p-value	Injured	Uninjured	p-value
Stance Time (% of gait cycle)	--	--	64.5(6.7)	65.9(7.2)	0.009*	64.1(1.8)	65.0(2.1)	0.006*
Loading Response (% stance phase)	--	--	18.9(2.4)	17.8(2.6)	0.038*	20.0(2.2)	18.9(2.0)	0.015*
Foot Flat (% stance phase)	--	--	30.1(6.2)	29.0(6.0)	n.s.	26.8(4.0)	27.3(3.6)	n.s.
Pre-swing (% stance phase)	--	--	15.5(2.8)	19.1(2.7)	< 0.001*	17.2(2.2)	18.7(1.6)	< 0.001*
Speed (m/s)	--	0.92(0.25)	1.1(0.18)			1.3(0.16)
Physical Activity Scale	2.3(0.7)	2.7(1.0)	3.3(1.0)			43(1.1)

Relationship between early tendon structure and later gait function

Greater dynamic shear modulus LSI (r=0.518, p=0.040) and tendon cross sectional area LSI (r=0.737, p=0.002) at 4 weeks positively related to larger stance phase asymmetries at 24 weeks. There were no other statistically significant relationships observed between any relative tendon structural parameter at any time point and gait characteristics at 24 weeks. There were no differences in tendon structure between participants able compared to unable to return to their reported pre-injury level on the PAS.

Relationship between early tendon structure and later strength function

Greater tendon cross sectional area LSI at 8 weeks (r=0.658, p=0.002), 12 weeks (r=0.524, p=0.018), and 24 weeks (r=0.461, p=0.036) related positively to more symmetrical heel-rise height in participants able to perform a unilateral heel-rise at 24 weeks (n=21). Tendon elongation at 8 weeks (r= −0.602, p=0.005) and 12 weeks (r= −0.589, p=0.008) related negatively to heel-rise work LSI in participants able to perform a unilateral heel-rise at 24 weeks.

Surgically managed participants - sensitivity analysis

Change of tendon structural properties over time were similar to those reported for the total cohort for all characteristics except length. For tendon length, there was a statistically significant interaction term when including only the surgically-managed participants. Tendon length significantly increased between 4–24 (p=0.021), 8–12 (p<0.001) and 8–24 weeks (p=0.007). Results of the ANOVA can be seen in table 4.

Table 4:

Results of ANOVA for surgically-managed participants only. * indicates p < 0.05

Tendon Structural Parameter	Main effect for time (p-value/pη2)	Main effect for side (p-value/pη2)	Interaction term (p-value/pη2
Shear modulus	n.s./0.057	0.002*/0.575	n.s./0.024
Viscosity	n.s./0.055	<0.001*/0.713	n.s./0.052
Dynamic shear modulus	0.023*/0.230	<0.001*/0.818	n.s./0.030
Tendon cross sectional area	<0.001*/0.556	<0.001*/0.857	<0.001*/0.556
Tendon length	<0.001*/0.430	0.002*/0.547	0.045/0.184

Gait was similarly affected in the surgically managed participants, the only difference being surgically managed participants also spent less time in foot flat at 24 weeks (p=0.043).

Surgically managed participants demonstrated similar relationships as the whole cohort between tendon properties and stance phase asymmetries at 24 weeks[dynamic shear modulus LSI at 4 weeks (r=0.601, p=0.039); tendon cross sectional area LSI at 4 weeks (r=0.797, p=0.002)]. Similar to the whole group, tendon cross sectional area at 8 weeks was associated with improved heel-rise test height LSI (r=0.556, p=0.025) and tendon elongation at 8 weeks (r= −0.571, p=0.017) and 12 weeks (r= −0.585, p=0.017) was negatively related to heel-rise work LSI at 24 weeks in those individuals able to perform a unilateral heel-rise. No other relationships were observed between tendon structural parameters and heel-rise performance within this subgroup.

Discussion

The most important findings of this study are that tendon structure assessed by ultrasound imaging shows change over the first weeks of recovery after Achilles tendon rupture, and early tendon structure seems to be a potential prognostic indicator of walking gait and heel-rise performance later in recovery. When tracking healing over time, it seems tendon morphology is the more robust measure to gauge patient progress compared to tendon mechanical properties.

From a clinical standpoint, this study provides support for using ultrasound-based measures in the assessment of patient progress after Achilles tendon rupture. Tendon cross sectional area[32] and tendon length[21, 26] have been previously found to be responsive to change over time. Within our group, the majority of tendon lengthening occurred prior to 4 weeks after injury with small changes up to 6 months post injury. This is fairly consistent with previous literature reporting tendon lengthening within the first 6 weeks of recovery[21, 26].

The largest changes in cross sectional area were anticipated to occur with weight bearing onset, but were found to happen later. This period of greatest change aligned better with the timing of discontinuation of the walking boot. This was somewhat unexpected as prior animal studies have related weight bearing, but not immobilization, to tendon callus formation (characterized by increasing tendon cross sectional area)[3, 4, 13]. A potential explanation for the observed timing of cross sectional area increase may be that walking in a boot with wedges places low amounts of load on the Achilles tendon, and it is only when wedges are removed that loads become high enough to stimulate repair and remodeling[41] of the tendon. Immobilization with wedging is aimed at approximating tendon ends[8, 11] and decreasing tendon load[2, 22, 31], but the effect of wedging on Achilles load is debated[14] and may depend on the specific way in which a patient is immobilized. When rehabilitating these patients, it may be that exercise-based treatment during the immobilization phase that safely and gradually increases tendon loads[32] could help progress the tendon toward callus formation while mitigating the risk of overloading and re-rupturing the tendon.

With regard to elastography, it seems that using shear modulus and viscosity measured via cSWE may be more subject to change with subtle changes in the tendon, resulting in increased variability. However, in combination, dynamic shear modulus could potentially be used to track the early healing trajectory. This measure is considered to be more comparable to the shear elasticity measure provided by commercially available shear wave elastography scanners, and the results of this study provide similar values and change over time in ruptured Achilles tendons reported by Zhang, et al[45]. Of interest is that dynamic shear modulus changed similarly on both ruptured and uninjured sides. It may be that as individuals are changing weight bearing status and beginning strength training in rehabilitation that both sides respond to these changes in tendon loading.

Prior studies have found tendon length, cross sectional area, and mechanical properties to relate to a variety of parameters of patient function within a single time point[5, 29, 36, 38, 43]. The results of this study expand on those findings, supporting the notion of tendon structure relating to patient function across time points.

This study is limited by small sample size. Additionally, the population studied is heterogeneous with regard to sex, age, and treatment strategy. While this may result in large variability, it does make the population of individuals included in this study more representative of the variety of patients presenting for post-Achilles rupture rehabilitation. Despite these limitations, this study does provide novel, preliminary data in an emerging body of literature regarding the clinical usage of ultrasound to track soft tissue healing and supports further study.

Clinically, the results of this study suggest that assessment of tendon morphology early in recovery may assist with improving the objectivity of establishing patient prognosis and planning a patient’s rehabilitative course. This study supports the use of tendon structure, particularly tendon morphology, assessed via ultrasound imaging as an early prognostic indicator of recovery in individuals with Achilles tendon rupture.

Conclusion

After Achilles tendon rupture, tendon cross sectional area increases, tendon length increases, and tendon dynamic shear modulus decreases relative to the uninjured side. Within the ruptured tendon, all of these tendon structural characteristics tend to increase over time during the first 24 weeks of recovery. This finding suggests that tendon structure assessed by ultrasound imaging could be a useful biomarker to track early tendon healing. Tendon cross sectional area and dynamic shear modulus in the first 4 weeks also related to 24 week walking gait symmetry and heel-rise performance, supporting the idea that early tendon structure could have prognostic value for 6 month outcomes in individuals early in recovery from Achilles tendon rupture.