Nonlinear Interactions of Lower Limb Clinical Measures Associated With Asymptomatic Achilles Tendon Abnormality in Ballet Dancers

Key Points

• Hip strength and ankle dorsiflexion range of motion should be investigated in combination for an effective understanding of tendon load in ballet dancers.

• Our results reinforce the importance of kinetic chain assessment to better comprehend tendon abnormality complexity. Clinicians may consider interactions between local and nonlocal measures when evaluating and monitoring ballet dancers.

• This study uses a cross-sectional design, which limits the ability to draw cause-effect conclusions. Therefore, extrapolations suggesting causal relationships between clinical measures and tendon pain should be avoided.

Tendon abnormalities have been defined as changes in thickness (increased anteroposterior diameter) and disorganized structure on imaging and may represent overload or load-related change and adaptation.¹ Although those imaging changes are found in individuals with clinical presentation of tendinopathy, they can also be present in asymptomatic people.² Achilles tendon abnormality increases the risk of developing Achilles tendinopathy 7-fold.³

Ankle and Achilles load are extremely high in dance training.⁴ Ballet movements require different calf muscle capacities across extreme ankle ranges. This demand places the Achilles tendon under different force types, such as compression (plié, ankle dorsiflexion), tensile (jumping/running), a combination of both (sauté, going from dorsiflexion to plantarflexion during a ballet jump), and shear forces (tendu, movements with full ankle range of motion [ROM]). The structure and curricula of ballet classes are designed to build the capacities dancers need; however, the load on the Achilles tendon likely makes dancers susceptible to abnormality.^5,6

Several modifiable clinical factors have been associated with Achilles tendinopathy, for instance, lower strength, power, and endurance, impaired ROM and foot pronation, etc.^7,8 However, no research has explored the association of those modifiable factors with tendon abnormality in the absence of pain. Beyond the linear investigations, the use of nonlinear analysis can offer a combined exploration of the independent variables that better capture complexity.⁹ Given that the presence of tendon abnormality can increase the likelihood of future symptoms, understanding the relationship between these modifiable clinical factors and the presence of asymptomatic abnormality may provide additional prevention opportunities.³ Therefore, the purpose of this study was to investigate the predictive value of clinical measures that have been associated with tendinopathy to identify Achilles tendon abnormality in asymptomatic ballet dancers through classification and regression tree (CART) analysis.

METHODS

This cross-sectional study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines and received ethics approval from the Federal University of São Carlos Human Ethics Committee (project number 24257219.4.0000.5504).¹⁰ All participants/legal guardians provided informed consent, and their rights were protected.

Procedures and Assessments

Eligibility criteria were checked by a physiotherapist. Data were collected by 2 investigators, both registered physiotherapists. All participants had both lower limbs evaluated in a random order, and all tests were performed barefoot. Intrarater reliability was assessed through a test-retest study conducted twice, 3 to 7 days apart, by the same examiner on 5 participants (age = 26.4 ± 4.1 years, weight = 61.36 ± 8.4 kg, height = 1.63 ± 0.1 m). Intraclass correlation coefficient (ICC) was calculated using 2-way mixed effects, consistency, and average measures, and the standard error of the measurement (SEM) was calculated using the formula SEM = SD × √1 − ICC (Table 1).¹²

Table 1.Reliability Data for All Variables and the Lever Arm Defined for Each Strength Test

View Fullscreen

Participants Descriptive Data

Demographic, anthropometric, and dance-related details were gathered. The sample was characterized regarding Achilles tendon symptoms and function through the Brazilian Portuguese version of the Victorian Institute of Sport Assessment-Achilles questionnaire.¹³

Achilles Tendon Scans

Scans were collected by a trained physiotherapist using portable ultrasound equipment (LOGIQ V2, GE Medical Systems, China Co, Ltd), with a linear probe (12L-RS, frequency 12 MHz) and 2 cm depth.¹⁴ Participants assumed a prone position with feet off the plinth and in contact with the wall for ankle positioning (Figure 1A).¹⁴ Tendons were longitudinally scanned from insertion to the musculotendinous junction in search of hypoechoic areas. If identified, confirmation was sought through transverse axis rescanning. Tendon abnormality was defined as the presence of hypoechoic areas (focal collagen fiber disorganization) and/or fusiform tendon shape.¹⁵ Tendon thickness was measured using ImageJ Software (National Institutes of Health).¹⁵ The maximum anteroposterior thickness at the enthesis and midportion was measured in 3 images of each portion, and the average was calculated. The midportion thickness was then divided by enthesis thickness. Tendons with a ratio exceeding 1.2 or 20% were classified as having a fusiform shape (tendon abnormality).¹⁵ For analysis, tendons were grouped into the following 2 categories: tendons with abnormality (fusiform shape and/or hypoechoic areas) and tendons without abnormality.

View Figure

• Download as Powerpoint
• Download Figure

Foot Pronation

Foot pronation was evaluated using the Navicular Drop Test.¹⁶ While seated, participants had the height of the most medial aspect of the navicular tuberosity measured in relation to the ground using a card fixed in a right-angle metal bracket. After, in a single-legged standing position, the height was remeasured (Figure 1B, C). Vertical displacement was documented in centimeters, and the average of 3 trials was used for analysis.

Ankle Dorsiflexion ROM

Weight-bearing ankle dorsiflexion ROM was assessed using the Lunge test.¹⁷ Participants stood in front of a wall. A tape line was placed on the floor extending to the wall. They were instructed to position the evaluated foot on the tape line, bend the knee forward until the patella touched the line on the front wall, and maintain heel contact with the ground. To control for subtalar hyperpronation, the second toe and the mid-heel were aligned with the tape line. When the participant reached maximum ankle dorsiflexion, the examiner placed an inclinometer (smartphone application) 15 cm below the tibial tuberosity to measure the angle (Figure 1D). The angle between the tibia and the vertical line was recorded 3 times, and the average was used for analysis.

Calf Endurance

Calf raise endurance was tested using the Ankle Measure for Endurance and Strength device through the Single Leg Heel Raise test.¹⁸ The heel raise height was controlled by an elastic band attached to the device, adjusted by the examiner at maximum plantar flexion. At the start position (ankle plantigrade, nontested leg lifted in the air, 1 hand support), a metronome set at 46 beats per minute was started. At the first beat, the participant rose the heel, clearing the elastic band, and at the second beat lowered the heel down (Figure 1E, F). The test concluded if the participant decided to stop or if any of the specified criteria occurred twice in a row: (1) not reaching the maximum plantar flexion controlled by the elastic, (2) leaning too much on the wall with clear hip flexion, (3) flexing the knee during the movement, or (4) losing the pace. The maximum number of repetitions for each side was documented.

Isometric Strength Testing

A handheld dynamometer (Lafayette Manual Muscle Test System, Lafayette Instruments) was used to measure hip, knee, and ankle torque. Inelastic straps stabilized participants and the dynamometer, eliminating the examiner’s strength effect.¹⁹ After 1 submaximal and 1 maximal contraction for familiarization, 3 valid trials with maximal voluntary isometric contraction held for 5 seconds were recorded. If there was over a 10% variation between contractions, a fourth trial was performed.²⁰ Limbs were assessed alternately, with at least 1 minute of rest between contractions on the same side. The same examiner provided vigorous and standardized verbal encouragement for all tests.²¹

Peak-force values were obtained in kilograms and transformed to Newtons (kg × 9.81). Newtons were converted to torque by multiplying the value by the specific lever arm (Table 1) for each muscle group ([force (N) × lever arm (m)]). Then, torque was normalized by the participant’s body mass and multiplied by 100 to achieve the percentage of body weight (BW) (Nm/kg × 100 = % BW). The average of 3 trials was used for analysis.

Knee extensor torque was obtained with participants in supine-lying position and the tested knee flexed at 30°.²¹ The dynamometer was placed between the medial and lateral malleoli (Figure 2A). Participants were asked to push, extending the knee.²¹

View Figure

• Download as Powerpoint
• Download Figure

Ankle plantar flexor torque was evaluated in prone-lying position, with their feet off the plinth in a neutral position.²¹ Two straps were used for participant stabilization: on the pelvis and on the distal third of the thigh. The dynamometer was stabilized at the plantar region of the metatarsophalangeal joints (Figure 2B).²¹ Participants were instructed to push, trying to move the foot upward.²¹

Hip external rotator torque was measured with participants seated, with their hips and knees at 90°. To prevent substitution from adductors, a towel roll was placed between the knees. The dynamometer was placed 5 cm proximal to the medial malleoli, with a strap pulled through it and secured around the plinth leg (Figure 2C).²² Participants were asked to push the foot inward.

Hip extensor torque was measured with participants lying in prone position. The knee of the tested leg was flexed at 90°, with the pelvis stabilized by a strap.²¹ The dynamometer, positioned proximal to the popliteal fossa, was further secured by a second strap (Figure 2D). Participants were instructed to push the foot toward the ceiling.²¹

Hip abductor torque was measured with participants in side-lying position with their pelvis stabilized by a strap.²⁰ The dynamometer was placed 5 cm proximal to the lateral knee joint line (Figure 2E). Participants were instructed to push the leg upward.²²

RESULTS

Of 37 recruited dancers, 2 were not included (metabolic disease and musculoskeletal condition), leaving 35 included dancers. Descriptive data are presented in Table 2 and Supplemental Table 1. Two participants had only 1 lower limb assessed due to surgery at the contralateral limb (anterior cruciate ligament reconstruction).

Table 2.Participant Descriptive Data (Mean ± SD) and Tendon Abnormality Information (Frequency/Percentage)^a

View Fullscreen

From the 68 tendons assessed by ultrasonography, 42 tendons presented with abnormality and 26 were tendons without abnormality. Tendon abnormality was found in 80% of participants: 16 dancers had bilateral tendon abnormality, 12 showed unilateral (6 right/6 left) tendon changes, and 7 participants had bilateral tendons without abnormality. The most common abnormality type found was fusiform shape (Table 2).

Among the included variables, the CART model identified hip abductor and extensor torque, ankle dorsiflexion ROM, hip external rotator torque, and calf endurance as hierarchical and inter-related predictors of Achilles tendon abnormality (Figure 3). The model accurately classified 88.1% of tendons with abnormality and 76.9% of tendon without abnormality, resulting in an overall classification percentage of 83.8% (PR [95% CI] = 4.302 [1.946–9.509]). The area under the ROC curve was 0.84 (95% CI, 0.74–0.95; P < .0001).

View Figure

• Download as Powerpoint
• Download Figure

The model identified 3 classification rules indicating Achilles abnormality (risk profiles). Hip abductor torque of ≤198.6% BW, hip extensor torque of ≤186.8% BW, and ankle dorsiflexion ROM of ≤40.5°, when combined, were associated with the presence of Achilles abnormality (node 5; n = 20; PR [95% CI] = 1.779 [1.268–2.494]). Interestingly, node 7 suggested that, even with greater ankle dorsiflexion ROM (>40.5°), there is still a risk of having abnormality when combined with an overall below cutoff hip torque (abductor, extensor, and external rotator; n = 12; PR [95% CI] = 1.543 [1.12–2.125]). When hip abductor and extensor torque were below the cutoff and ankle dorsiflexion ROM, hip external rotator torque, and calf endurance were above cutoff values, there was an association with Achilles abnormality; however, this was not statistically significant (node 10; n = 5; PR [95% CI] = 1.396 [0.09247–2.109]).

The model also showed 3 classification rules for tendons without abnormality (protective factors). Hip abductor torque of >198.6% BW directly identified tendons without abnormality (protective factor against Achilles abnormality; node 2; n = 10; PR [95% CI] = 0.406 [0.1742–0.9462]). At node 4, an interaction between lower hip abductor torque and greater hip extensor torque (>186.8% BW) classified 3 tendons without abnormality, but there was no statistical significance (node 4; n = 3; PR [95% CI] = 0.1941 [0.01443–2.61]). The combination of poor hip abductor and extensor torque, when having greater ankle dorsiflexion ROM, increased hip external rotator torque, and reduced calf endurance was associated with tendons without abnormality, also with no statistical significance (node 9; n = 7; PR [95% CI] = 0.182 [0.02901–1.153]).

DISCUSSION

This study was the first to explore how the interaction between clinical measures can identify Achilles abnormality in the absence of pain in a population with a high prevalence of tendon changes, using a statistical analysis that includes nonlinear interactions.^5,6,23 Our findings showed the significance of considering proximal strength as an influencing factor for Achilles abnormality and highlights the impact of a combined assessment, including the whole kinetic chain, to better understand Achilles tendon load in asymptomatic ballet dancers. Eighty percent of the population had Achilles tendon abnormality in 1 or both tendons, supporting the known disconnection between imaging findings and symptoms. This also supports research demonstrating that tendon changes may reflect adaptation to load.^1,26

The strongest interaction for having Achilles abnormality was node 5, suggesting that a combination of poor hip abductor and extensor torque along with a stiff ankle is associated with a 77% higher probability of having an Achilles abnormality. This finding provides a logical clinical rationale. During landing, in the presence of poor hip torque and a stiff ankle, impact forces may be poorly dissipated and absorbed across lower limb joints and muscles, potentially leading to chronic overload on the Achilles tendon over time, thus contributing to abnormality development.^27–29 In a cohort of male runners with Achilles tendinopathy, Sancho et al found that hip extensor and abductor isometric strength were reduced in the Achilles tendinopathy group.³⁰ Despite population variations, our findings relate to the cited results, suggesting that reduced hip torque may be a modifiable risk factor preceding Achilles tendinopathy onset. Our model also demonstrates that poor hip torque can compromise the Achilles tendon, and an increased value can protect it. Node 2 reinforces the importance of hip torque on tendon load, where higher hip abductor torque was associated with a 59% reduction in the probability of Achilles abnormality. This highlights the effectiveness of hip strength assessment to comprehensively understand Achilles abnormality.

Based on node 7, even with greater ankle dorsiflexion ROM, there was still a 54% probability of having an Achilles abnormality if associated with overall poor hip torque. Currently, evidence on ankle dorsiflexion ROM and its association with Achilles tendinopathy is inconclusive, with studies reporting both increased and decreased ROM as risk factors.^31,32 Our findings suggest that discrepancies in the literature may arise from the isolated assessment of ankle dorsiflexion. When interpreting results from nodes 5 and 7 together, it becomes evident that hip strength is a key factor influencing tendon structure, and clinical factors can mutually influence each other, leading to different risk profiles. Our study indicates that solely assessing ankle dorsiflexion ROM may not adequately capture the complexities of Achilles tendon load, emphasizing the importance of a more comprehensive lower limb evaluation.

Among all 5 muscle groups evaluated, only hip torque was selected by the CART model. Surprisingly, ankle plantar flexor torque, despite its recognized importance in the literature for Achilles tendon structure and pain, was not identified as a predictor for abnormality in this sample.^33,34 A prospective study in a military population identified ankle plantar flexor torque as a predictor for Achilles tendinopathy, with greater risk if torque was lower than 50.0 Nm.³³ Also, Nunes et al cross-sectionally investigated the relationship between plantar flexor function and architecture with Achilles tendon morphology in asymptomatic ballet dancers.³⁴ They found that ankle plantar flexor torque and gastrocnemius architecture explained 24% of tendon thickness variance.³⁴ Despite literature expectations, the CART model selected hip torque over ankle plantar flexor torque.

This study has limitations. The lack of an a priori sample size calculation may be viewed as a limitation; however, the high accuracy of the CART model indicates that our sample size was sufficient to ensure the robustness of the analyses. The studied population is niche and challenging to access, leading us to use strategies to achieve a reasonable sample size, despite it being relatively small. The inclusion of both tendons independently per participant, although a common practice in the literature, is acknowledged as a study limitation. However, this approach permits an individualized interpretation and application of results in clinical practice, considering ballet dancers present with important interlimb asymmetries. Another limitation is the inclusion of both professional and amateur dancers to increase the sample size. Despite technical level distinctions, we found no differences between the categories for any characteristic except for weekly training volume, which was expected and is not supported by the literature as a factor related to the occurrence of tendon abnormality.³⁵ The similar total years of dance experience for both categories (P = .147) ensures a homogeneous cumulative load, a key factor influencing tendon abnormality, which is supported by Lieberthal et al. The authors investigated factors associated with asymptomatic tendon abnormality in long-distance runners. Among total years of running, number of weekly running sessions, and weekly mileage, only total years of running was associated with tendon abnormality, showing that acute load does not influence tendon structure analyzed by ultrasound.³⁵ Last, although palpation aligned with existing literature for ruling out Achilles tendon symptoms, a more precise method, such as single-leg hops or progressive load testing, could enhance symptom investigation.³⁶ Yet, during assessment, we asked participants to self-report any tendon pain experienced during ballet movements that potentially loaded the Achilles tendon.

In conclusion, we found that asymptomatic Achilles abnormality was highly frequent among the evaluated dancers. Interestingly, only hip abductor torque was indicated in isolation with a higher likelihood of not having an abnormality. When assessing dancers, clinicians should consider the entire kinetic chain, as a combination of hip torque and ankle dorsiflexion ROM accurately identified the presence of tendon abnormality. Addressing these combined clinical measures into the screening process of asymptomatic dancers may reduce the risk of future tendon pain by identifying individuals with structural change.