Early MRI-based quantitative outcomes are associated with a positive functional performance trajectory from 6- to 24-months post-ACL surgery

Abstract

Purpose:

Quantitative magnetic resonance imaging (qMRI) has been used to determine the failure properties of ACL grafts and native ACL repairs and/or restorations. How these properties relate to future clinical, functional, and patient-reported outcomes remain unknown. The study objective was to investigate the relationship between non-contemporaneous qMRI measures and traditional outcome measures following Bridge-Enhanced ACL Restoration (BEAR). It was hypothesized that qMRI parameters at 6 months would be associated with clinical, functional, and/or patient-reported outcomes at 6 months, 24 months, and changes from 6 to 24 months post-surgery.

Methods:

Data of BEAR patients (n=65) from a randomized control trial of BEAR versus ACL reconstruction (BEAR II Trial; NCT02664545) were utilized retrospectively for the present analysis. Images were acquired using the Constructive Interference in Steady State (CISS) sequence at 6 months post-surgery. Single-leg hop test ratios, arthrometric knee laxity values, and International Knee Documentation Committee (IKDC) subjective scores were determined at 6- and 24-months post-surgery. The associations between traditional outcomes and MRI measures of normalized signal intensity, mean cross-sectional area (CSA), volume, and estimated failure load of the healing ACL were evaluated based on bivariate correlations and multivariable regression analyses, which considered the potential effects of age, sex, and body mass index.

Results:

CSA (r=0.44, p=0.01), volume (r=0.44, p=0.01), and estimated failure load (r=0.48, p=0.01) at 6-months were predictive of the change in single-leg hop ratio from 6-24 months in bivariate analysis. CSA (β_standardized=0.42, p=0.01), volume (β_standardized=0.42, p=0.01), and estimated failure load (β_standardized=0.48, p=0.01) remained significant predictors when considering the demographic variables. No significant associations were observed between MRI variables and either knee laxity or IKDC when adjusting for demographic variables. Signal intensity was also not significant at any timepoint.

Conclusion:

The qMRI-based measures of CSA, volume, and estimated failure load were predictive of a positive functional outcome trajectory from 6-24 months post-surgery. These variables measured using qMRI at 6 months post-surgery could serve as prospective markers of the functional outcome trajectory from 6-24 months post-surgery, aiding in rehabilitation programming and return-to-sport decisions to improve surgical outcomes and reduce the risk of reinjury.

Level of Evidence:

Level II

INTRODUCTION

Primary traditional outcome measures following anterior cruciate ligament (ACL) surgery typically involve gross measures of joint functional performance (e.g., single-leg hop test), clinical examinations of knee laxity (e.g., arthrometer laxity test), and/or patient-reported outcomes (e.g., International Knee Documentation Committee subjective score).[2, 10] However, none of these outcome measures directly interrogate the structural properties of the healing ligament or graft. In contrast, quantitative magnetic resonance imaging (qMRI) allows for noninvasive, objective assessment of soft tissue organization and the structural properties of ligaments and grafts.[4-7, 9, 22, 44] Furthermore, there is evidence that qMRI can detect early soft tissue changes,[8, 15] even before they manifest clinically.[44]

Previous investigations have suggested that the quantitative graft changes observed on MRI post-ACL reconstruction (ACLR) are related to contemporaneous clinical, functional, and patient-reported outcomes.[4, 21, 24, 29, 35, 40, 45] While these studies highlight the utility of qMRI outcomes in tracking patients’ response to ACL surgery, there is a lack of evidence as to whether qMRI measures of the ACL can be used to predict future clinical, functional, and patient-reported outcomes after ACL surgery. Such predictions could be used to individualize postoperative care, by identifying patients at risk for adverse outcomes at earlier timepoints. For example, these predictions could be useful to inform return-to-sport decisions and for tailoring rehabilitation programs to improve surgical outcomes and to reduce the risk of reinjury.

The objective of the present analysis was to investigate whether qMRI measures can predict future traditional outcome measures following Bridge-Enhanced ACL Restoration (BEAR). The study was designed to identify whether qMRI measures obtained at 6 months post- surgery relate to functional, clinical, and patient-reported outcomes at 24 months post-surgery, as well as to the change in these outcome measures between 6 and 24 months. Three qMRI parameters, which have been previously used to develop estimation models for ACL failure load,[4, 5] were evaluated: 1) normalized ligament signal intensity (SI), 2) ligament mean cross-sectional area (CSA), and 3) ligament volume. The estimated ACL failure load based on these qMRI parameters was also examined.[5] It was hypothesized that early qMRI parameters would be associated with clinical, functional, and/or patient-reported outcomes at 6- and 24-months post-surgery, and predict the change in these outcome measures from 6 to 24 months post-surgery.

MATERIALS AND METHODS

Data were acquired from the IRB-approved “BEAR II Trial” at Boston Children’s Hospital (Boston, MA, USA; NCT02664545, IDE G150268, IRB-P00021470).[31] All patients granted their informed consent. A total of 100 patients were enrolled and randomly assigned with a 2:1 ratio to receive either an ACL reconstruction with a tendon autograft (n=35) or scaffold enhanced restoration of the native ACL (n=65). The BEAR implant used for this procedure is an FDA-authorized device designed to restore the native ACL, both anatomically and compositionally,[3, 17-19, 30, 31, 34, 36] without some of the limitations associated with ACL reconstruction (e.g. donor site morbidity[3]) or primary repair (e.g. high failure rate[1, 16]).

Inclusion criteria for the trial were male and female patients (42% male), who were 14-35 years of age (mean age 19.4 ± 5.1 years), presenting with a complete ACL mid-substance tear within 45 days of surgery. Patients were excluded if they had a history of prior knee surgery or knee infection or potentially adverse risk factors including a history of nicotine use, corticosteroid use, chemotherapy, diabetes, inflammatory arthritis, sickle cell anemia, or anaphylaxis. Patients with concomitant injury to the posterolateral corner, grade III medial collateral ligament injury, or complete patellar dislocation were also excluded.[30] For the current study, only the BEAR patients were included. Any patients that underwent revision surgery or tore their contralateral ACL within 24 months post-surgery were excluded. Additionally, patients that did not have data for both the 6- or 24- month timepoints for a specific outcome were excluded from the analyses of only that outcome measure. Figure 1 details the patient data available for the MRI analysis for each traditional outcome measure.

Figure 1.

CONSORT diagram depicting the flow of patients through the study.

Surgical Procedures

The BEAR procedure was performed arthroscopically, with 4 mm tunnels drilled in the femur and tibia. First, the tibial stump was secured with a #2 absorbable suture (Vicryl; Ethicon) using a whip stitch. A cortical button (Endobutton; Smith & Nephew) with #2 nonabsorbable sutures (Ethibond; Ethicon) and the previous #2 absorbable sutures were passed through the femoral tunnel and secured to the proximal femoral cortex. The BEAR implant, composed of bovine-derived extracellular matrix proteins, was delivered via mini-arthrotomy, and secured by the #2 nonabsorbable sutures. The implant was then saturated with 5-10mL of autologous blood, and the ends of the torn ACL were passed through the implant.[30, 32]

MR Imaging

MR imaging was performed at 6 months post-surgery, using a 3T magnet (Tim Trio; Siemens) and a 15-channel transmit/receive knee coil (Siemens). Three-dimensional Constructive Interference in Steady State (CISS) sequence (FA=35°; TR=12.78ms; TE=6.39ms; FOV=140mm; 384×384 acquisition matrix with voxel size 0.365mm × 0.365mm × 1.5mm) was acquired of the surgical limb. The ACL was then segmented from the MR image stack by one observer with more than 5 years of experience in ACL segmentation (AMK; inter- and intra- segmenter ICC≥0.93) using commercial imaging software (Mimics; Materialise, Leuven, Belgium).

Quantitative MRI Outcomes

The segmented ligaments were used to calculate ligament volume, mean CSA (volume divided by the length), and normalized SI (mean ACL grayscale value normalized to cortical bone grayscale value).[17, 18] The qMRI prediction model of ACL failure load was previously trained on porcine data for which the tensile structural property data were available.[5] The ligament volume parameter for the model was subsequently replaced by the mean CSA, to account for variations in subject size. Thus, the updated model utilized normalized SI and mean CSA as independent variables to predict the failure load of the restored ACL (F_max) (Eq. 1). Quantitative MRI outcome values are summarized in Table 1.

Table 1.

Summary of MRI values and traditional outcome at the analyzed timepoints.

Variable Type	Variable	Timepoint	Mean	Standard Deviation
MRI	Cross-Sectional Area	6mo	56.42	11.94
	Signal Intensity	6mo	1.34	0.21
	Volume	6mo	2338.24	581.24
	Estimated Failure Load	6mo	632.47	156.90
Traditional Outcome	Single-Leg Hop Ratio	6mo	0.86	0.17
		24mo	0.94	0.13
		Δ(24mo - 6mo)	0.08	0.16
	Arthrometer SSD	6mo	2.7	2.9
		24mo	1.6	3.1
		Δ(24mo - 6mo)	−1.1	3.5
	IKDC	6mo	85.7	11.4
		24mo	89.4	13.6
		Δ(24mo - 6mo)	4.6	12.5

$$F_{max}={\beta }_0+{\beta }_{1}SI+{\beta }_{2}CSA$$ Eq. 1

Outcome Measures

Three traditional outcome measures were assessed at 6- and 24-months post-surgery: 1) single-leg hop test distance expressed as the ratio of the surgical to the contralateral limb (functional outcome),[37] 2) the side-to-side difference (surgical – contralateral) in anteroposterior knee laxity using an arthrometer (KT-1000 Knee Arthrometer)[42] (clinical outcome), and 3) the International Knee Documentation Committee (IKDC) subjective score (a validated patient reported outcome).[13, 33] During the single-leg hop test, the patients wore a brace on the surgically treated knee for added protection. The arthrometer measurements were performed by an experienced independent examiner, who was blinded to surgical laterality and treatment using knee sleeves. The anteroposterior laxity values of both knees were measured at 130N of applied anterior shear load to the tibia relative to the femur. The IKDC Subjective Score was calculated based on patients’ responses to the IKDC Questionnaire. Traditional outcome measure values are summarized in Table 1.

Statistical Analyses

Initially, the bivariate relationship between qMRI measures at 6-months and traditional outcomes measured at 6 and 24 months were evaluated using Pearson’s r. Pearson’s r was also computed between MRI outcomes and the change in traditional outcomes between 6 and 24 months to determine whether MRI measures were associated with the subsequent trajectory of traditional outcomes.

To further quantify the relationship between MRI measures and traditional outcomes, multivariable regression analyses were performed for variables in which both correlation the correlations were significant. Additional demographic variables (i.e., age, sex, and body mass index (BMI)) were included as candidate variables for the models. The modeling utilized a backwards elimination procedure based on a significance-to-stay of p<0.05. Statistical analyses were performed using SAS (SAS Institute Inc, Cary, NC, USA) and the Python “statsmodels” package.[38] While the sample size of the current study was the maximum available given the previously described constraints (Figure 1), a power analysis determined that the available sample size was sufficient to detect at least a 10% difference in traditional outcomes with 95% power assuming alpha equal to 0.05.

RESULTS

The MRI outcome measures of ACL mean CSA (r=−0.37, p=0.04) and volume (r=−0.37, p=0.04 measured at 6-months were negatively correlated with the single-leg hop ratio at the same timepoint. Volume (r=−0.30, p=0.02) and estimated failure load (r=−0.30, p=0.03) were also negatively correlated with IKDC score at 6 months post-surgery. No significant correlations were observed between 6-month MRI outcomes and the traditional outcome measured at 24 months. MRI measures of ACL CSA (r=0.44, p=0.01), volume (r=0.44, p=0.01), and estimated failure load (r=0.48, p=0.01) were positively correlated to the change in single-leg hop ratio from 6 to 24 months (Table 2). Normalized signal intensity was not significantly correlated with traditional outcome measures at any timepoint.

Table 2.

Pearson (r) correlation results (p<.05 in bold).

MRI Measure	Traditional Outcome	Timepoint	r	P-Value
Cross-Sectional Area	Single-Leg Hop Ratio	6mo	−0.37	0.04
		24mo	0.05	n.s.
		Δ(24mo - 6mo)	0.44	0.01
	Arthrometer SSD	6mo	−0.17	n.s.
		24mo	−0.08	n.s.
		Δ(24mo - 6mo)	0.07	n.s.
	IKDC	6mo	−0.27	n.s.
		24mo	−0.04	n.s.
		Δ(24mo - 6mo)	0.20	n.s.
Signal Intensity	Single-Leg Hop Ratio	6mo	−0.08	n.s.
		24mo	−0.28	n.s.
		Δ(24mo - 6mo)	−0.17	n.s.
	Arthrometer SSD	6mo	−0.03	n.s.
		24mo	−0.07	n.s.
		Δ(24mo - 6mo)	−0.04	n.s.
	IKDC	6mo	0.13	n.s.
		24mo	0.17	n.s.
		Δ(24mo - 6mo)	0.06	n.s.
Volume	Single-Leg Hop Ratio	6mo	−0.37	0.04
		24mo	0.06	n.s.
		Δ(24mo - 6mo)	0.44	0.01
	Arthrometer SSD	6mo	−0.14	n.s.
		24mo	−0.07	n.s.
		Δ(24mo - 6mo)	0.06	n.s.
	IKDC	6mo	−0.30	0.02
		24mo	−0.06	n.s.
		Δ(24mo - 6mo)	0.21	n.s.
Estimated Failure Load	Single-Leg Hop Ratio	6mo	−0.27	n.s.
		24mo	0.22	n.s.
		Δ(24mo - 6mo)	0.48	0.01
	Arthrometer SSD	6mo	−0.12	n.s.
		24mo	−0.03	n.s.
		Δ(24mo - 6mo)	0.08	n.s.
	IKDC	6mo	−0.30	0.03
		24mo	−0.13	n.s.
		Δ(24mo - 6mo)	0.13	n.s.

Based on significant correlation results, multivariable regression analyses controlling for additional demographic variables were performed. MRI variables that were significantly correlated with 6-month traditional outcome variables did not meet the criteria for inclusion in the final models (p>0.05) after accounting for demographic variables. MRI variables that were significantly correlated with the change in single-leg hop ratio from 6 to 24 months remained significant even when accounting for demographic variables (Table 3, Figure 2). CSA explained 19.3% of the variance (p=0.01), volume explained 19.7% of the variance (p=0.01), and estimated failure load explained 23.4% of the variance (p=0.01).

Table 3.

Backwards stepwise regression results (p<0.05 in bold).

Dependent Variable	Independent Variable	Standardized Coefficient	P-Value
Single-Leg Hop Ratio	Intercept	−0.08	n.s.
	CSA	0.42	0.01
Single-Leg Hop Ratio	Intercept	−0.07	n.s.
	Volume	0.42	0.01
Single-Leg Hop Ratio	Intercept	−0.06	n.s.
	Estimated Failure Load	0.48	0.01

Figure 2.

Final stepwise regression plots (note: all values are standardized).

DISCUSSION

The most important finding of this analysis was that CSA, ligament volume, and estimated failure load measured via MRI at 6 months post-surgery were predictive of a functional outcome (single-leg hop ratio) trajectory from 6 to 24 months post-surgery, even when controlling for demographics. Correlation analysis revealed a significant positive correlation between mean CSA, volume, and estimated failure load to single-leg hop ratio. Overall, this finding suggests that a larger, stronger ligament, as determined using qMRI at 6 months, was indicative of greater recovery of functional performance over the following 18 months. Furthermore, given that there were no significant associations between these variables at either 6 or 24 months after controlling for demographics, the significant association between these variables and the change in single-leg hop ratio from 6 to 24 months may be a new biomarker to consider apart from measures at either timepoint. A potential explanation for this is that the measurement of patients’ longitudinal change during this period controls for other demographic or unmeasured variables that contribute to the variability of their cross-sectional measurements at either 6 or 24 months.

Interestingly, two of these same MRI variables (volume and estimated failure load), were negatively correlated with IKDC at 6 months. This is possibly due to subjective perception of rehabilitation progress relative to baseline activity level. However, it’s worth noting that BEAR patients have been found to have higher IKDC scores at both 6- and 24-months post-surgery relative to comparable ACL reconstruction patients.[31] Furthermore, in the present analysis, these correlations were no longer significant after accounting for demographic variables.

The regression analysis sought to control for additional potentially confounding variables, including age, sex, and BMI when examining the quantitative relationship between CSA, volume, and estimated failure load and the change in single-leg hop ratio from 6 to 24 months. Although these additional variables have been identified as potentially having an effect on ACL surgical outcomes,[12, 20, 23, 25, 26, 39] they were not significant explanatory variables when examining the relationship between these MRI variables and the change in single-leg hop ratio from 6 to 24 months.

Importantly, CSA, volume, and estimated failure load remained significant predictors and demonstrated relatively large effect sizes with standardized coefficients of 0.42, 0.42, and 0.48, respectively. This finding indicates each standard deviation (11.9mm²; 21.2% of mean CSA) increase in CSA resulted in a 0.42 standard deviation (0.07; 83.1% of mean single-leg hop ratio change) increase in single-leg hop ratio change from 6 to 24 months. Likewise, for volume each standard deviation (581.2mm³; 24.9% of mean volume) increase resulted in a 0.42 standard deviation (0.07; 83.1% of mean single-leg hop ratio change) increase in single-leg hop ratio change from 6 to 24 months. For the estimated failure load, each standard deviation (156.9N; 24.8% of mean estimated failure load) increase in estimated failure load resulted in a 0.48 standard deviation (0.08; 95.0% of mean single-leg hop ratio change) increase in single-leg hop ratio change from 6 to 24 months. It is likely that these effect sizes are clinically significant, given that the changes in single-leg hop ratio are greater than the previously reported differences between patients with normal versus below average knee function,[28] and between patients who did or did not return to their preinjury level of performance.[43]

To our knowledge, relating non-contemporaneous qMRI measures to traditional outcome measures has been performed post-ACL surgery for cartilage,[14, 27] but not for the ACL or ACL graft. In a recent systematic review, Van Dyck et al.[41] evaluated studies using qMRI to measure ACL graft healing. Of the studies reviewed relating qMRI measures to clinical, functional, and/or patient-reported outcomes, all outcome comparisons were performed at the time of the qMR imaging.[41]

In previous research, Biercevicz et al. found that ACL graft volume and median signal intensity acquired from a T₁-weighted sequence were associated with single-leg hop test performance at 3- and 5-year follow-ups, as well as Knee Injury and Osteoarthritis Outcome Score (KOOS) quality of life, sport/function, pain, and symptom sub-scores at 5-year follow up, when the MR images were acquired at the same timepoints.[4] Marchiori et al. found that T₂ relaxation time of the graft was correlated with arthrometric knee laxity at 4- and 18- month follow up.[29] Once again, the T₂ relaxation time and arthrometric knee laxity were measured at the same timepoints, rather than between timepoints. Given that qMRI measures of healing grafts have been associated with traditional outcome measures acquired at the same timepoint,[4, 45] the current study yields new insight by suggesting that the healing state of the ligament, as measured by qMRI at an early timepoint, is also predictive of the future healing trajectory of the ACL for 2 years following surgical intervention. This could potentially shorten the feedback loop between modifications to rehabilitation routine and their effect on future outcomes. Additionally, it would help with personalizing rehabilitation based on individual qMRI outcomes.

Pertaining to the use of qMRI of cartilage to predict non-contemporaneous outcomes following ACL surgery, Ithurburn et al. found that lower patient-reported outcomes at 2 years post-return-to-sport were associated with elevated cartilage T_1ρ and T₂ relaxation times at 5 years post-return-to-sport.[14] For use of qMRI as a noninvasive, prospective biomarker, investigating the reverse relationship would be ideal. Li et al. also found that elevated cartilage T_1ρ relaxation time at 6 months post-surgery was associated with postural stability asymmetries, as measured by the Y-Balance test, at 2 years post-surgery.[27] Such cartilage-specific biomarkers could be combined with ACL-specific biomarkers to develop a fuller perspective of a patient’s healing trajectory.

There are a few study limitations to consider. First, the R² fit of the final models were relatively low, explaining between 19.3% (CSA) and 23.4% (estimated failure load) of variance. This finding likely reflects several factors, including the coarseness of traditional outcome measures, psychological readiness, and the number of potential confounding variables both measured and unmeasured. However, the goal of these stepwise regressions was to evaluate the significance of the MRI variables when clinically relevant demographic variables were controlled for, not to fit an optimal prediction model between MRI and traditional outcome measures. In addition, subjects who underwent revision surgery to the ipsilateral ACL (n=9; Figure 1) in the 24 months post-surgery were not included in the analysis. This excluded subset potentially represents an important, albeit small, group of at-risk patients that were not accounted for in the present analysis. Future work will specifically focus on identifying MRI risk factors for revision surgery, and on developing machine learning models that directly predict future traditional outcome measures using qMRI measures, specifically targeting the traditional outcome measures that show significant associations with qMRI in the present study. Direct prediction with machine learning would have the added benefit of avoiding time consuming image segmentation, which is currently an impediment to the use of qMRI in clinical settings.

CONCLUSIONS

The qMRI-based measures of CSA, volume, and estimated failure load were predictive of a functional outcome trajectory from 6-24 months post-ACL restoration surgery. These variables measured using qMRI at 6 months post-surgery may be a prospective marker of the functional outcome trajectory from 6-24 months post-surgery. Biomarkers that are indicative of future outcomes could enable interventions before adverse outcomes occur, for example by informing return-to-sport decisions or modifications of post-operative rehabilitation. Earlier interventions could potentially improve surgical outcomes and reduce the risk of reinjury.

LIST OF ABBREVIATIONS

ACL

anterior cruciate ligamen

ACLR

anterior cruciate ligament reconstruction

BEAR

bridge-enhanced anterior cruciate ligament restoration

BMI

body mass index

CISS

Constructive Interference in Steady State

CSA

cross-sectional area

IKDC

International Knee Documentation Committee

KOOS

Knee Injury and Osteoarthritis Outcome Score

SI

signal intensity

qMRI

quantitative magnetic resonance imaging