REGIONAL PATELLAR TENDON STRAIN IN THE SHORT- AND LONG-TERM AFTER ACLR USING BONE-PATELLAR TENDON-BONE AUTOGRAFT

Abstract

Background:

Anterior cruciate ligament reconstruction (ACLR) often involves harvesting a bone–patellar tendon–bone (BPTB) autograft. How graft harvest affects tendon strain across the three distinct regions (medial, lateral, central) is not known.

Purpose:

In two cohorts of patients after ACLR using BPTB autograft (one group less than 24 months since surgery, and another group greater than 24 months since surgery), we aimed to 1) quantify strain in the three regions of the patellar tendon during 60% of maximum voluntary isometric contraction (MVIC) in 90 degrees of knee flexion, and 2) assess how effort level in two different knee joint angles (60 vs 90 degrees) impacts strain in the medial and lateral regions of the patellar tendon.

Study Design:

Cross-sectional study; Level of evidence, 3.

Methods:

Effort levels ranging from 20–100% MVIC were performed at 90 and 60 degrees of knee flexion on an isokinetic dynamometer, with simultaneous ultrasound imaging of tendon length to calculate regional strain. Linear mixed effects models were used to evaluate regional strain at 60% MVIC in 90 degrees of knee flexion. The fixed effects of %MVIC, joint angles, and time from surgery groups on patellar tendon strain were evaluated using separate models for the medial and lateral region.

Results:

Fourteen participants in the short-term group (7.4 ± 1.9 months from surgery) and 15 participants in the long-term group (71.5 ± 16.5 months from surgery) were included. At 60% MVIC in 90 degrees of knee flexion, the short-term group demonstrated lower patellar tendon strain than the long-term group regardless of region (p < 0.01). The central region also had lower strain than the medial and lateral regions regardless of group (both p < 0.01). Finally, the rate at which strain increased with increasing effect levels differed between time from surgery groups in both medial and lateral regions.

Conclusion:

The short-term group had lower strain in all regions of the patellar tendon following ACLR using BPTB autograft. The medial and lateral regions showed varying strain based on time from surgery and effort level. Graft site rehabilitation may have to consider varying knee joint angles and effort levels to target specific regions of the patellar tendon.

Introduction

Anterior cruciate ligament reconstruction (ACLR) commonly consists of harvesting either a bone–patellar tendon–bone (BPTB) autograft or combined semitendinosus–gracilis tendon (hamstring) autograft.^{10,22,29,34,38,40} Despite BPTB autograft being considered the best graft choice, especially in young athletes,^10,38 harvesting a BPTB graft induces patellar tendon injury, resulting in higher levels of anterior knee pain,^13,21,28 prolonged quadriceps weakness,^13,21,35 and potentially a higher risk of post-traumatic knee osteoarthritis development compared to other graft types.³² Graft site morbidities originate from surgical trauma and subsequent structural changes within the patellar tendon over the course of rehabilitation.^12,36 A recent study also demonstrated that the whole tendon (medial, central, and lateral), and not just the central region of the tendon, undergoes significant changes in morphology and mechanical properties over the course of rehabilitation. ¹⁹

After ACL injury, patellar tendon force is reduced due to activation failure and atrophy of the quadriceps.^8,11,20 A delay between peak quadriceps activity and peak knee extensor moment has also been observed during walking gait after ACLR with BPTB autografts,¹⁶ possibly reflecting the patellar tendon’s loss of ability to transfer force efficiently.^16,18 Early hypertrophy of the healing graft site tendon (i.e., larger cross-section area) one month after graft harvest is associated with higher peak knee extensor torque,²⁰ highlighting the importance of patellar tendon remodeling for recovery after ACLR. Alterations to thickness, length, and mechanical properties, persist years after ACLR with a BPTB autograft.^1,12,19 Nevertheless, limited research exists on the effects of BPTB autograft harvest on regional patellar tendon strain characteristics and recovery of mechanical properties.

Patellar tendon strain can reliably be obtained with B-mode ultrasound imaging in the three distinct regions (medial, lateral, central).^18,19 Tendon strain reflects tendon mechanical properties,²⁶ and may be used to assess tendon health and recovery from injury.^24,27 Changes in patellar tendon strain over time may further inform the impact of BPTB autograft harvest on tendon function and healing. In two cohorts of patients after ACLR using BPTB autograft (one group less than 24 months since surgery, and another group greater than 24 months since surgery), we aimed to 1) quantify strain in the medial, lateral, and central regions of the patellar tendon during 60% of maximum voluntary isometric contraction (MVIC) in 90 degrees of knee flexion, and 2) assess how effort level (expressed as %MVIC) in two different knee joint angles (60 vs 90 degrees) impacts strain in the medial and lateral regions of the patellar tendon. We hypothesized that the time from surgery would affect strain, with the short-term group demonstrating less strain. We also hypothesized that effort level and joint angle would impact strain.

Methods

Study Design

This is a Level 3 cross-sectional study. All participants provided their informed consent prior to inclusion in the study. The study was approved by the Institutional Review Board at the University of Delaware.

Participants

For inclusion, participants had to be between 18 and 45 years old, have a history of ACLR using BPTB autograft, and have been cleared to perform knee extension MVIC tests by a medical provider. Time from surgery varied between 5 to 102 months after ACLR and participants were grouped based on time from surgery into short-term (less than 24 months] and long-term groups (greater than 24 months): (median [range]: Short-term = 7.3 [4.5 – 11.3], long-term = 74.4 [43.8 – 102.7]. This categorization was chosen as majority of tendon healing is expected to take place in the first 2 years following surgery, with less tendon adaptations expected thereafter.³⁷ The Marx Activity Rating Scale,²⁵ Victorian Institute of Sports Assessment - Patellar Tendon (VISA-P),³⁹ and Patient-Reported Outcome Measurement Information System (PROMIS domains: Pain Interference Score, and Physical Function Score),¹⁵ were used to report participant characteristics.

Patellar Tendon Strain Testing Protocol

The testing procedure started with a standardized warm-up, including 10 minutes of stationary bike riding and familiarization with the isokinetic dynamometer (Biodex System 3, Shirley, New York).¹⁸ Only the side with ACLR was tested. For all participants, the protocol was first conducted at 90 degrees of knee flexion, then at 60 degrees of knee flexion and then at again at 90 degrees of knee flexion, as described below (Figure 1). During each voluntary contraction, participants received verbal feedback from the assessor (NI) and visual feedback from the torque-time curve on the screen of the dynamometer.

Figure 1:

Testing protocol for regional patellar tendon strain assessment at 60 and 90 degrees.

Four MVIC trials (two for the medial and two for the lateral) were performed followed by six trials (three for the medial and three for the lateral) at 80%, 60%, 40%, and 20% of their MVIC. After completing the protocol at 90 and 60 degrees of flexion, the knee was adjusted back to 90 degrees to assess central strain at 60% MVIC. The target torque at each % MVIC (beginning with the highest effort level) was then presented as a target on a monitor for the participant to follow. Participants were offered 45 seconds to rest between trials or until they felt prepared for the next trial, to reduce quadriceps fatigability between trials.^2,17 Additionally, five resting length images were collected between the 100% and 80% MVIC trials in both 60 and 90 degrees (Figure 1). Concurrent extended field of view ultrasound images of patellar tendon length were obtained using B-mode ultrasound (GE Healthcare, Logiq e, Frequency: 10MHz, Depth: 2~3 cm) while participants maintained the torque output at the target levels for 5 seconds (Figure 1). Strain was obtained across three distinct regions within the patellar tendon: the medial, lateral, and central regions. The specific locations were determined through palpation and identifying the medial and lateral borders of the tendon on ultrasound as previous described.^17,18 Strain was defined as the change in patellar tendon length (contracted – resting) divided by the resting length and expressed as a percentage (%). Patellar tendon length was measured as the linear distance between the deepest attachment point of the tendon on the tibia and the patella.¹⁸ Good to excellent reliability has been established in these methods.^7,17

Statistics

No a priori power analyses were performed as no empirical data exist.

Comparing three region (medial, lateral, central) strain between groups

A linear mixed effects model (LMM) was used to examine the fixed effects of region (medial, lateral, and central), group (short [less than 24 months] and long-term groups [greater than 24 months]), and the interaction between region and group on patellar tendon strain during 60% MVIC contraction in 90 degrees of knee flexion. Sex was included as a covariate in the model since sex differences are known to influence tendon strain,¹⁷ and participants were modeled as random effects.^9,33 Degrees of freedom were calculated using Satterthwaite’s method. Significant results were further investigated using pairwise comparisons with Tukey honestly significant difference tests (α=0.05).

Since strength differed between time from surgery groups in 90 degrees of flexion (Table 1), and higher quadriceps forces may explain higher levels of strain, a second model including MVIC torque (Nm) was also created. The final model was determined with model comparisons, using fit statistics (Akaike information criteria [AIC] and Bayesian information criteria [BIC]) and log likelihood ratio tests with Chi-squared distributions to derive significance.

Table 1.

Participant demographics

	Total (n = 29)	Short-term (n = 14)	Long-term (n = 15)	P – Value
Time from surgery (months)	40.6 ± 34.6	7.5 ± 1.9	71.5 ± 16.5	< 0.001*
Age (years)	23.7 ± 5.5	23.0 ± 7.3	24.3 ± 3.3	0.560
Sex (Male/Female)	12/17	7/7	5/10	0.678
Body mass (kg)	71.4 ± 11.4	70.7 ± 9.8	71.9 ± 13.0	0.291
Height (cm)	169.2 ± 9.1	171.1 ± 8.8	167.5 ± 9.4	0.778
BMI (kg/m2)	24.8 ± 3.0	24.1 ± 2.1	25.5 ± 3.6	0.195
MVIC at 90 degrees (Nm)	187.3 ± 73.4	147.9 ± 57.1	224.1 ± 68.9	0.003*
MVIC at 60 degrees (Nm)	172.9 ± 48.1	166.4 ± 49.8	179.0 ± 47.3	0.493
MVIC at 90 degrees (Nm/Kg)	2.6 ± 0.8	2.1 ± 0.6	3.1 ± 0.7	< 0.001*
MVIC at 60 degrees (Nm/Kg)	2.4 ± 0.5	2.3 ± 0.4	2.5 ± 0.5	0.317
MARX Score	6.5 ± 4.7	4.2 ± 3.7	8.6 ± 4.7	0.010*
VISA-P	74.3 ± 21.2	62.9 ± 19.6	85.0 ± 17.0	0.003*
PROMIS Score Pain	43.6 ± 4.0	43.7 ± 4.1	40.2 ± 7.8	0.928
PROMIS Score Physical Function	55.0 ± 4.4	53.6 ± 5.6	43.5 ± 4.0	0.112

Mean ± standard deviation; body mass index (BMI); knee extensor torque (Nm) at maximum voluntary isometric contraction (MVIC); knee extensor torque at maximum voluntary isometric contraction normalized by body weight (Nm/Kg), Marx Activity Rating Scale (Marx), Victorian Institute of Sports Assessment - Patellar Tendon (VISA-P), and Patient-Reported Outcome Measurement Information System (PROMIS domains: Pain Interference Score, and Physical Function Score). All comparisons were performed using t-tests, except for sex distribution, which was assessed using a Chi-squared test, as it did not have a normal distribution.

^*Denotes difference between the short-term group compared to the long-term group

Influence of group, effort level, and joint angles on strain

LMMs were used to determine the fixed effects of group (time since surgery), effort level (%MVIC), and joint angle (90 and 60 degrees of flexion) on patellar tendon strain. 2- and 3-way interactions for the time from surgery, effort level, and joint angle were also evaluated. Separate models were created for the medial and lateral regions. Effort level (in % MVIC) was modeled as a continuous variable. Participants were modeled as random effects and models were adjusted for sex (α=0.05). Statistics were performed in R studio (package: lme4, lmerTest).^5,23

Results

Twenty-nine participants were included in this analysis, with 14 in the short-term group and 15 in the long-term group. Participant demographics and modeled variables are reported in Table 1.

Compaing regional (medial, lateral, central) strain between groups

No region by time from surgery group interaction effect was observed (p = 0.16; Figure 2; Supplementary Table 1). The main effect of time from surgery group (p < 0.01) was characterized by the short-term group demonstrating less strain compared to the long-term group (estimated marginal mean [95% confidence interval]: short-term = 5.26 [3.58 – 6.93] %, long-term = 8.86 [7.16 – 10.57] %). A main effect of region was also observed (p < 0.01) characterized by less strain in the central region (5.37 [3.82 – 6.93] %) compared to both the medial (p < 0.01, 9.15 [7.64 – 10.66] %), and lateral (p < 0.01, 6.66 [5.15 – 8.17] %) regions. Inclusion of MVIC torque did not significantly improve overall model fit (p = 0.53; Supplementary Tables 2 and 3). The reported results and estimated marginal means are from the model without torque for parsimony.

Figure 2.

Regional (medial, central, and lateral) patellar tendon strain at 60% of maximal voluntary isometric contraction (MVIC) in 90 degrees of knee flexion grouped by surgery group.

Influence of time from surgery group, effort level, and joint angle on patellar tendon strain

Medial Region

No 3-way interaction effect was observed (p = 0.45; Figure 3; Table 2, Supplementary Tables 4 and 5). Time from surgery group by effort level (p < 0.01), and time from surgery group by joint angle (p = 0.01) interactions were observed. The group by effort level interaction was characterized by a steeper slope (i.e., greater rate of increase [0.126 and 0.151 for 60 and 90 degrees of knee flexion, respectively] in strain with increase in effort) in the long-term, compared to the short-term group regardless of joint angle (Figure 3). The group by joint angle interaction was characterized by the short-term group (6.25 [4.53 – 7.97] %) demonstrating lower strain compared to the long-term group (12.22 [10.48 – 13.95] %) at 90 degrees of knee flexion, but no difference in the strain between the short- (8.30 [6.58 – 10.03] %) and long-term (9.12 [7.39 – 10.86] %) groups at 60 degrees (Supplementary Table 6). No effort level by joint angle interaction was observed (p = 0.23).

Figure 3.

Patellar tendon strain at various effort levels grouped by region (medial and lateral), time from surgery group (short- and long-term), and joint angle (60 and 90 degrees). Each color represents a different participant.

Table 2.

Patellar tendon strain at various effort levels separated by region and time from surgery group

Medial Region	20%MVIC	40%MVIC	60%MVIC	80%MVIC	100%MVIC
Short-term Group Strain (%)	3.56 (1.34, 5.79)	5.49 (3.34, 7.64)	7.41 (5.28, 9.54)	9.34 (7.18, 11.49)	11.26 (9.03, 13.50)
Long-term Group Strain (%)	4.83 (2.57, 7.09)	7.60 (5.41, 9.79)	10.37 (8.20, 12.54)	13.14 (10.95, 15.33)	15.91 (13.65, 18.17)
Lateral Region
Short-term Group Strain (%)	2.02 (0.37, 3.67)	3.81 (2.24, 5.39)	5.61 (4.06, 7.16)	7.40 (5.82, 8.98)	9.20 (7.54, 10.86)
Long-term Group Strain (%)	3.23 (1.55, 4.90)	5.50 (3.89, 7.10)	7.76 (6.18, 9.34)	10.03 (8.43, 11.64)	12.30 (10.63, 13.98)

Values are reported as estimated marginal means from the models (95% confidence interval). MVIC: maximum voluntary isometric contraction.

Lateral Region

No 3-way interaction effect was observed (p = 0.918; Figure 3; Tables 2, 3 and Supplementary Table 6 and 7). Time from surgery group by effort level (p = 0.03), and effort level by joint angle (p = 0.02) interaction effects were observed. Similar to the medial region, the group by effort level interaction was characterized by a steeper slope (0.100 and 0.127 for 60 and 90 degrees of knee flexion, respectively) in the long-term, compared to the short-term group regardless of joint angle (Figure 3). The effort level by joint angle interaction was characterized by a steeper slope in 90 versus 60 degrees of flexion (Table 3 and Figure 3). No time from surgery group by joint angle interaction was observed (p = 0.53).

Table 3.

Strain in the lateral region at various effort levels grouped by joint angle

Lateral Region	20%MVIC	40%MVIC	60%MVIC	80%MVIC	100%MVIC
60 degrees Strain (%)	2.84 (1.58, 4.10)	4.62 (3.45, 5.78)	6.39 (5.26, 7.51)	8.16 (7.00, 9.32)	9.94 (8.68, 11.20)
90 degrees Strain (%)	2.41 (1.15, 3.67)	4.69 (3.54, 5.85)	6.98 (5.86, 8.11)	9.27 (8.11, 10.43)	11.56 (10.29, 12.83)

Values are reported as estimated marginal means from the models (95% confidence interval). MVIC: maximum voluntary isometric contraction.

Discussion

To our knowledge, this is the first study to quantify regional patellar tendon strain at various effort levels of knee extensor torques and knee joint angles in patients after ACLR. Lower patellar tendon strain in the short-term group compared to the long-term group was observed even after accounting for maximum voluntary isometric contraction strength. Furthermore, the long-term group exhibited a greater increase in strain with increasing % effort levels compared to the short-term group. These findings potentially suggest a diminished capacity to achieve higher strain in all regions of the patellar tendon during the earlier post-surgery period. We also found that the central patellar tendon had lower strain, which is consistent with what we previously observed in healthy patellar tendons.¹⁷ Finally, the medial patellar tendon strain was lower at 90 degrees of knee flexion, in comparison to 60 degrees of knee flexion in the short-term group, indicating that joint angle and time from surgery may play a role in regional tendon strain. In summary, our results indicate that the patellar tendon is not a uniform structure that responds similarly to load across regions (medial, lateral, central), and strain is diminished particularly in the short-term following ACLR using BPTB autografts.

Our results demonstrate that the patellar tendon experiences less strain in the short-term following ACLR, which may be detrimental to recovery since mechanical load is critical for tendon healing. Considering that recent research has shown that patellar tendon healing with a larger cross-section area after graft harvest one month after ACLR is associated with greater quadriceps strength,²⁰ efforts to stimulate patellar tendon healing early may be of paramount importance for better long-term results after ACLR. Specifically in the early stages of rehabilitation, it may be crucial to prioritize exercises that promote higher load and longer time under tension to the tendon to effectively generate tendon strain and promote healing. Our findings also demonstrated less medial patellar tendon strain at 90 degrees early after surgery, suggesting the need to perform knee extension exercises at various joint angles to achieve optimal medial patellar tendon strain. Neuromuscular electrical stimulation (NMES) is an essential intervention to improve quadriceps function after ACLR.^4,14 NMES has the ability to overwrite quadriceps inhibition via forced contractions and may also be an important supplement to patellar tendon loading exercises to promote higher strain to maximize tendon load and subsequent healing.⁶

Joint position and time from surgery group influenced the slope of the effort-strain curve in both the medial and lateral regions. Our results indicate that the slope, representing the increase in strain relative to the increase in effort level, was greater in the long-term group than in the short-term group. Our findings may reflect higher patellar tendon stiffness in the long-term group. While this analysis was performed in 2 separate cross-sectional cohorts, the finding may be reflecting the mechanical properties of the tendon changing with time after ACLR. Supporting this assumption, a recent study found that patellar tendon viscosity is lower after 6−9 months after surgery when compared to 1 month after ACLR using BPTB.¹⁹ It is important to note, however, that patellar tendon strain is influenced by both quadriceps muscle force and tendon stiffness.^3,41 Net knee extensor torque, measured on a dynamometer, is determined by quadriceps muscle forces and the opposing forces from hamstring co-contraction, which has been observed to be greater after ACLR.³⁰ Considering that the addition of MVIC torque did not significantly influence model fit for patellar tendon strain, the observed differences in patellar strain between time from surgery group may be attributed to alterations in both tendon stiffness and hamstring co-contraction during our tests.

In line with our hypothesis, the graft site patellar tendon demonstrated nonuniform regional strain. Pearson et al³¹ revealed that anterior proximal, and distal posterior areas exhibited higher strains compared to other regions of the patellar tendon in the longitudinal axis. Ito et al¹⁸ demonstrated that the lateral region of the patellar tendon showed lower viscosity and thickness than the medial region of healthy patellar tendons in the mediolateral axis. The recent study by Ito et al¹⁹ compared the morphology and mechanical properties of the medial, central, and lateral regions of the patellar tendon after ACLR using the BPTB graft in relation to the same regions of the healthy patellar tendon of the uninvolved limb. It was observed that there is tendon thickening and elongation not only in the central third of the patellar tendon after ACLR using a BPTB graft but also in the medial and lateral regions where no graft was harvested.¹⁹ The results of the current study support these findings and suggest that harvesting the central region of the patellar tendon for ACLR may affect the strain the medial and lateral regions of the patellar tendon receive during quadriceps contraction, leading to non-uniform recovery in mechanical properties across regions. There may be a need to consider separate assessments of regional strain induced in the central, medial and lateral regions for exercise selection to optimize patellar tendon healing.

Graft site morbidities after BPTB autograft for ACLR are well-documented and include higher prevalence and persistent anterior knee pain,^{10,13,21,26,28,37} prolonged quadriceps weakness,^13,21,35 and possible elevated risks of post-traumatic knee osteoarthritis³² when compared to those who receive alternative graft types. Appropriate levels of strain are known to influence tendon healing and adaptations.²⁷ The imbalance in strain among the medial, lateral, and central regions, along with an overall lower strain earlier on after ACLR we observed may contribute to the tendon healing process and subsequently influence graft site morbidities.

It is important to acknowledge limitations of this study. Firstly, this is a cross-sectional investigation without a control group. Longitudinal studies with larger samples are necessary to track patellar tendon strain over an extended post-ACLR period. Graft harvest techniques, differences in rehabilitation, and demographic factors such as age, general activity level, joint hypermobility, or postoperative knee stiffness, are potential confounding factors may influence tendon strain and should be considered in the future. Fatigue may have influenced our results. Future studies should consider studying central patellar tendon strain across various effort levels and knee joint angles as this was not performed in this study to limit the effects of fatigue.

Conclusion

Lower strain in all regions of the patellar tendon was observed in the short-term group (<24 months) in relation to the long-term group (>24 months) following ACLR using BPTB autograft. The central patellar tendon exhibited lower strain than the remaining regions of the patellar tendon regardless of group. Finally, the medial and lateral regions of the patellar tendon experienced strain at different rates in relation to effort level, potentially reflecting alterations to regional tendon stiffness. Limited regional patellar tendon strain after BPTB autograft harvest may contribute to graft site morbidities and less than optimal recovery after ACLR. Clinicians should consider the time elapsed since surgery, along with joint angle and regional strain, when recommending patellar tendon loading exercises after ACLR using BPTB autografts.

Clinical Relevance:

Clinicians should consider the time elapsed since surgery and the manipulation of knee angle, as factors that can exert varying levels of strain on different regions of the patellar tendon after ACLR. Accelerating quadriceps strengthening to enhance strain distribution across the patellar tendon to promote tendon healing may be of benefit for optimizing post-surgery rehabilitation.

What is known about the subject:

Using bone–patellar tendon–bone (BPTB) autografts for anterior cruciate ligament reconstruction (ACLR) is associated with higher levels of anterior knee pain, and prolonged quadriceps weakness compared to alternative graft types. Optimal healing of the patellar tendon may influence knee function after ACLR, and mechanical strain is critical for tendon healing. Additionally, the three distinct regions (medial, lateral, central) of the patellar tendon are all influenced by graft harvest and may require independent assessments to quantify healing.

What this study adds to existing knowledge:

This study highlights the potential importance of considering time from surgery and joint angle to enhance regional patellar tendon strain during rehabilitation after ACLR using BPTB autografts to optimize tendon healing.