Degeneration in ACL Injured Knees with and without Reconstruction in Relation to Muscle Size and Fat Content—Data from the Osteoarthritis Initiative

Background Anterior cruciate ligaments (ACL) injuries represent a major risk factor for early osteoarthritis (OA). Purpose To evaluate the prevalence and 4-year progression of knee OA measured with 3T MR-imaging in individuals with ruptured, reconstructed or normal ACL and to assess the impact of thigh muscle characteristics. Methods A total of 54 knees (23/54 male, 31/54 female) were recruited from the Osteoarthritis Initiative (OAI). At baseline, 15/54 subjects had prevalent ACL ruptures and 15/54 subjects had prevalent ACL reconstruction (24/54 normal ACL). Western Ontario and McMasters Universities Arthritis Index (WOMAC) scores, Physical Activity Scores of the Elderly (PASE) and thigh muscle characteristics including strength, fat infiltration (Goutallier score) and thigh muscle cross-sectional area (CSA) MR measurements were obtained at baseline. Whole-organ MR-imaging Scores (WORMS) were obtained at baseline and at a 4-year follow-up time-point. Multivariate regression models, adjusting for covariates (age, gender, body mass index), were used for statistical analysis. Results At baseline, subjects with prevalent ACL ruptures had worse WORMS total scores (mean±SEM, 44.1±3.5) than subjects with ACL reconstruction (30.8±4.0; P = 0.015) and worse than subjects with normal ACL (21.3±3.0; P<0.001). Cartilage scores were worse in both femorotibial compartments in ACL injured knees than in knees with normal ACL (P<0.05). Knees with ACL reconstruction showed an increased degeneration of the medial meniscus (P = 0.036), cartilage degeneration at the medial femoral condyle (P = 0.011). In a multivariate regression model, including both ACL groups and total muscle characteristics as influence parameters, high thigh muscle CSA, high muscle/ fat ratio and low Goutallier scores were associated with less degenerative changes at the knee, independent of ACL status. Knees with ACL reconstruction showed an increased progression of cartilage degeneration at the medial tibia compared to the normal ACL group (P = 0.027). Conclusions High thigh muscle CSA is associated with less degenerative changes at the knee, independent of the ACL status and may potentially be advantageous in the prevention of early OA.


Introduction
Anterior cruciate ligament (ACL) ruptures are associated with an increased risk for post-traumatic osteoarthritis (OA) [1,2]. Conservatively treated patients often cannot return to the level of their previous activities [3]. To prevent early OA, in approximately two-thirds of patients with ACL ruptures, treatment consists of ACL reconstruction, aiming to restore mechanical stability [2,4,5]. Most long-term studies report good results ten or more years after surgery [6,7]. On the other hand, it has not been shown that surgical treatment reduces the risk for OA [8,9].
Quadriceps weakness, high muscle volume and other muscle parameters are known to be associated with knee joint degeneration [10,11]. Both, muscle characteristics of the thigh and knee joint degeneration may be determined quantitatively or semi-quantitatively using MR imaging. The muscle cross-sectional area (CSA) has been used to study muscle atrophy and hypertrophy on MR imaging. [12,13]. Using CSA, the vastus lateralis (VL)/ vastus medialis (VM) ratio has been shown to correlate with knee joint degeneration [14]. In comparison to conventional radiography, MR imaging allows for specific evaluation of the different joint structures and for detection of early OA [15]. There are several different MR imaging scores that aim to semi-quantitatively describe early degenerative changes of the knee joint, of which whole-organ magnetic resonance imaging score (WORMS) is the most frequently applied score [16][17][18]. Although many studies evaluate outcome after ACL reconstruction, no MR imaging study evaluates the impact of ACL status and thigh muscle characteristics on the longitudinal progression of degenerative changes at the knee [19].
The Osteoarthritis Initiative (OAI) is an ongoing longitudinal, NIH initiated multi-center cohort study with nearly 5000 participants. The aim is to provide a public research resource for investigation of the natural evolution of knee and hip OA and to identify potential prevention strategies. The OAI dataset includes clinical data, radiographs, bilateral 3T knee MRI studies and bilateral thigh MRI studies [20].
The purpose of the present study was to evaluate presence and 4-year progression of OA, measured with 3T MR imaging, in individuals with ruptured, reconstructed or normal ACL and the impact of thigh muscle area and other muscle characteristics. We hypothesized that was funded by NIH U01-AR055079 and P50-AR060752, by NIH K24-AR04884 and R01-AR052000 and by the OAI. The OAI is a publicprivate partnership comprised of five contracts (N01-AR-2-2258; N01-AR-2-2259; N01-AR-2-2260; N01-AR-2-2261; N01-AR-2-2262) funded by the National Institutes of Health, a branch of the Department of Health and Human Services, and conducted by the OAI Study Investigators. Private funding partners include Pfizer, Inc.; Novartis Pharmaceuticals Corporation; Merck Research Laboratories; and GlaxoSmithKline. The analyses in this study were funded through the NIH (National Institute of Arthritis and Musculoskeletal and Skin Diseases grants R01AR064771 and P50-AR060752). Private sector funding for the OAI is managed by the Foundation for the National Institutes of Health. This manuscript was prepared using an OAI public use data set and does not necessarily reflect the opinions or views of the OAI investigators, the NIH, or the private funding partners. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing Interests:
The funding does not alter our adherence to PLOS ONE policies on sharing data and materials. not only ACL reconstruction but also muscle characteristics have an impact on long-term knee OA prevalence and progression in individuals with ACL injuries.

Materials and Methods Subjects
Specific OAI datasets used in this study were medhist00 dataset, baseline datasets 0.2.2 and 0. E.1 and 4 year follow-up datasets 6.2.1 and 6.E.1 (http://www.oai.ucsf.edu/). The study protocol, amendments, and informed consent documentation were approved by the local institutional review boards. All clinical investigations have been conducted according to the principles expressed in the Declaration of Helsinki. Individuals from the OAI cohort with ACL reconstruction at baseline were identified by searching for the parameter"ever have ligament repair surgery"and by additional screening of the corresponding MR images for ACL reconstruction.
A subset of individuals from the OAI progression cohort (n = 304) was randomly selected. The progression cohort is characterized by the presence of symptomatic OA in at least one knee. Symptomatic OA is defined as the presence of pain, aching or stiffness in or around the knee on most days for at least 1 month during the past 12 months plus femorotibial osteophytes in the same knee on conventional radiographs. These knee MRIs were screened for the presence of complete ACL tear and for the presence of normal ACL. Individuals with mucoid degeneration of the ACL were excluded. Individuals were included in the study if they had had complete MRIs of the affected knee at baseline and at 4-year follow-up, had thigh MRIs at baseline and had no additional surgery in the meantime.

Imaging: Plain radiographs
Bilateral standing posterior-anterior knee radiographs were acquired for all subjects in the OAI. Additional details on radiograph acquisition in OAI are available at http://oai.epi-ucsf. org/datarelease/OperationsManuals.asp. Kellgren-Lawrence (KL) scores at baseline were obtained from the OAI database.

Muscle Fat infiltration
Thigh MRIs were reviewed on PACS workstations (Agfa, Ridgefield Park, NJ). Fatty infiltration was graded using a five-point semi-quantitative scale (Goutallier score) described by Goutallier et al and modified by Fuchs et al [27][28][29]: Grade 0, normal; grade 1, some fatty streaks; grade 2, < 50% fatty infiltration of the muscle; grade 3, as much fat as muscle; grade 4, >50% fatty infiltration of the muscle. All MRI studies were interpreted in consensus by two radiologists (P.M.J., L.N.). A third radiologist (T.M.L.) was consulted in case of disagreement.

Isometric strength
Isometric knee flexion and extension strength measurements were performed using a Good Strength apparatus (Metitur, Jyväskylä, Finland; www.oai.ucsf.edu/datarelease/ OperationsManuals.asp). Two submaximal practice trials were completed. Then strength (N) was measured three times for 3 seconds, each separated by 30 seconds; the highest value for a limb is used for maximal strength reported.

Statistical analysis
Statistical processing was performed with SPSS version 17.0 (SPSS Institute, Chicago, IL, USA) (P.M.J., T.B.). All tests were performed based on a 0.05 level of significance. Pearson correlations were used for correlation analyses of the different muscle characteristics. Partial Spearman correlations, adjusting for the covariates, were used for correlation analyses of different knee OA parameters with muscle characteristics. Multiple linear regression models, controlling for covariates, were used for comparisons between the different ACL groups. Additional multiple linear regression analyses were performed by including each muscle characteristic in the model. From the regression models we obtained adjusted means and standard errors (SEM) and 95% confidence intervals, if not stated otherwise. Covariates were OA risk factors including age (years), gender and body mass index (BMI; calculated as weight in kg/height m squared). Reproducibility Interreader-reproducibility for CSA measurements was calculated for each muscle separately and for the total thigh muscle. The coefficient of variation (CV) was determined using rootmean-square averages of standard deviations of repeated measurements as described previously [34]. Good reproducibility in our research group for WORMS assessment was reported previously [14,17,35,36].

Results
Subject characteristics are given in Table 1 and in S1 Appendix.  Table 2). The ACL rupture group had more severe scores for cartilage, meniscus and BME. Cartilage scores were worse in knees with ACL ruptures in all femorotibial compartments compared with subjects with knees normal ACL (all P<0.05). Knees with ACL reconstruction had worse cartilage scores at the medial femoral condyle as compared to subjects with normal ACL (MFC; 2.7±0.5 vs 1.0±0.3, P = 0.011). Knees with ACL reconstruction had worse progression at the medial tibia plateau (MT; 1.0 ±0.2 vs 0.3±0.2, P = 0.027). All other comparisons of WORMS progression scores did not show significant differences between the ACL groups.

WORMS and muscle characteristics
Associations between muscle characteristics (Pearson correlations) and differences of muscle characteristics between the ACL groups are given in Table 3 and Fig 2. For the entire study cohort, higher muscle CSA correlated significantly with lower WORMS total scores (Partial Spearman correlation, rho = -0.38, P = 0.007; Table 4) and with lower cartilage scores (rho = -0.40, P = 0.003), indicating less degenerative changes. More severe cartilage defects at the lateral tibia plateau (LT; rho = -0.35, P = 0.12), lateral femoral condyle (LFC; rho = -0.38, P = 0.006) and trochlea (rho = -0.35; P = 0.013) correlated significantly with smaller total muscle CSA. P-values were the smallest for the quadriceps CSA. High muscle/ fat ratio was associated with low WORMS scores (rho = -0.31; P = 0.026; cartilage score, rho = -0.35, rho = 0.013). High Goutallier scores correlated significantly with high total WORMS scores   additionally, P-values for the difference of MM scores between the "normal ACL" and the "ACL reconstruction" group and for the difference of the LT cartilage scores between the "normal ACL" group and the "ACL rupture" group lost significance (P>0.05). All other significant differences (P<0.05) remained significant after additional controlling for KL-scores.

Multiple regression models
Including each muscle characteristic in the multiple regression model revealed similar results as compared to the separate analyses. In the baseline analyses, ACL ruptures were associated with worse WORMS scores (P<0.001). ACL reconstruction was associated with increased cartilage degeneration at the MFC (P = 0.011). In the progression analysis, ACL reconstruction was associated with an increased progression of cartilage degeneration at the MT (P = 0.029). Also for the influence of muscle characteristics, the results from the multiple regression models were similar to the separate analyses. An additional significant correlation was found for higher "flexion strength" with less cartilage degeneration at baseline (P = 0.041; medial tibia, P = 0.049) and for more severe Goutallier scores with an increased progression of cartilage degeneration at the patella (P = 0.047) in the longitudinal analysis.

WOMAC
The mean WOMAC total score of the entire study cohort was 17.1±2.6 (mean±SEM). The differences for WOMAC scores between the ACL groups were not significant (all P>0.

Time since injury
The mean follow-up time after ACL reconstruction was 11.4±8.2 years (mean ± standard deviation (SD)). The mean time after the first injury of the analyzed knee was 21.9±12.9 years for the "ACL rupture" group (information available for 13/15 individuals) and 18.0±10.9 years for the "ACL reconstruction" group (information available for 14/15 individuals). This difference was not significant (P = 0.409). The "time after first injury" was significantly positively correlated with total WORMS scores (rho = 0.45, P = 0.028), MM scores (rho = 0.48, P = 0.017) and cartilage scores (rho = 0.42, P = 0.039). There were no significant correlations with muscle parameters or WOMAC scores (P>0.05). When comparing WORMS scores of the "ACL rupture" and the "ACL reconstruction" group and adjusting for the parameter "time since first injury" in a multiple regression model, the differences in total WORMS scores and BME scores remained significant (P = 0.013 and P = 0.019, respectively). When adjusting for all co-factors (age, gender, BMI) plus for "time since first injury" in a multiple regression model, significant differences were eliminated (P>0.05).

Reproducibility
The CV for total thigh muscle CSA measurements was 0.7% (171 mm 2 ) and varied between 0.2% and 4.0% (6 mm 2 and 46 mm 2 ) for the individual muscle CSA measurements.

Discussion
In this study morphological knee abnormalities and muscle characteristics in individuals with ruptured, reconstructed and normal ACL were evaluated in a baseline and a 4 year progression analysis of the OAI using 3T MRI. Worse baseline WORMS scores were found in subjects with ACL ruptures. Worse cartilage scores at the MFC and a statistical trend for increased WORMS total scores were found in subjects with ACL reconstruction, independent of muscle characteristics. In the progression analysis, ACL reconstruction was associated with an increased progression of cartilage degeneration at the MT. These results suggest, that (i) subjects with chronic ACL tears suffer from more advanced OA than subjects with ACL reconstruction, (ii) medial OA progresses despite ACL reconstruction and (iii) that high thigh muscle CSA, high muscle/ fat ratio and low muscle fat infiltration have a possible protective effect against the prevalence of knee OA.
Long-term studies report good results ten or more years after ACL surgery [6]. However, there is no definite evidence, that surgical treatment is superior to non-surgical treatment regarding the prevention of OA [37][38][39]. Cartilage degeneration and other degenerative changes at the knee progress despite functionally stable ACL reconstruction [8,9,15,40,41]. Degenerative cartilage changes appear seven years after surgery and about 10-17 years after the injury about 50% of patients develop OA [6,35,[42][43][44][45][46][47][48][49][50]. There are only few studies that compare surgical versus non-surgical treatment of ACL injuries in an over 10 year follow-up and most studies only performed conventional radiographs for radiological outcome evaluation [43, 45, 51]. Meunier et al found similar subjective outcome and similar rates of radiographic OA, but increased clinical instability in non-operatively treated subjects 15 years after injury [52]. In a 20 year follow-up, Louboutin et al found a lower percentage of radiographic OA in surgically treated patients [15], which is in line with our findings.
In our MR study, knees with ACL rupture showed significantly more degenerative changes. In case of ACL reconstruction, degeneration was found primarily in the medial femorotibial compartment for meniscus and cartilage scores, but not in the lateral compartment. This observation is in line with previous studies, that reported an increased degeneration in the medial femorotibial compartment after ACL reconstruction as opposed to the initial traumatic injury of the lateral femorotibial compartment [19,[53][54][55][56][57]. This may be due to not entirely restored kinematics, a persisting rotational instability and persisting anterior-posterior tibial translation after ACL reconstruction [2,55,58].
Previous studies reported, that thigh muscle volume and VL/ VM ratio are associated with knee OA [14]. We found that high thigh muscle CSA is associated with less knee OA changes in subjects with ACL injuries. The quadriceps muscle plays a particularly important role with respect to knee OA [10,11]. In contrast to the findings of Pan et al [14], VL/ VM ratio had no major influence on OA in our study cohort. Reasons may be, that we did not categorize the parameter, that most of our subjects were from the OAI progression cohort and that our study cohort was smaller. With respect to strength measurements, flexion strength showed a significant influence on cartilage scores in the multivariate regression model. In addition, strength and clinical WOMAC scores correlated significantly. Also other studies reported, that muscle weakness may be a risk factor for OA and that improving muscle strength improves the outcome after ACL reconstruction [10,11,59]. In our study, the influence of muscle characteristics on OA progression was not significant; only minor differences were depicted (Tables 2  and 4). This may be due to a small study cohort, a too short longitudinal follow-up or ceiling effects of WORMS parameters.
One major limitation of this study is the missing information on timing of the ACL rupture in relation to OAI baseline. However, most patients recalled a severe injury of the assessed knee joint and the difference in time between the first injury and the baseline of the study was not significantly different between the "ACL rupture" and the "ACL reconstruction" group. The importance of the parameter "time since first injury" was demonstrated, since it was associated with more severe knee joint degeneration. Although most patients recalled a severe knee injury, the exact origin of ACL ruptures in both, the "ACL rupture" and the "ACL reconstruction" group is not known. This missing information was one reason for performing a longitudinal progression analysis. The impact of ACL reconstruction versus chronic ACL rupture on long-term longitudinal progression of OA has not yet been evaluated [19]. However, most significant differences were found for baseline values. Particularly in subjects with prevalent OA, progression analysis is challenging [60]. Further, we were not able to differentiate between different surgical techniques for ACL reconstruction. It remains unclear whether muscle parameters had an influence on morphological knee OA or vice versa. Additionally, the groups we analyzed turned out to be relatively small and heterogeneous regarding age, gender and BMI after application of the inclusion and exclusion criteria, mainly availability of MR scans at baseline and at 4-year follow-up. Therefore we included these parameters as co-variates in the analyses. Power-analyses revealed that a total sample size of n = 30 was required for total WORMS scores and a total sample size of n = 54 was required for assessment of total muscle CSA. Still, the screening process and the analysis of knee and thigh MR images in this detailed fashion is extremely time-consuming and assessment of larger cohorts is aim of the ongoing osteoarthritis initiative.
In summary, in this 3T MRI study, ACL ruptures were associated with an increased prevalence of degenerative changes at the knee as compared to ACL reconstruction and as compared to knees with normal ACL. Knees with ACL reconstruction had more degenerative changes in the medial femorotibial compartment as compared to knees with normal ACL and showed an increased progression of cartilage degeneration at the MT. The results support the assumption of a lower risk for OA in knees with ACL reconstruction as compared to knees with chronic ACL ruptures, but patients still suffer from an increased risk for early OA, particularly of the medial femorotibial compartment, as compared to knees with normal ACL. Muscle parameters had an additional influence on the prevalence of OA. High muscle CSA had a moderate but significant association with reduced OA changes, independent of the ACL status. However, further progression of advanced OA may not be halted. In conclusion, the results might be helpful for patient education regarding the potential influence of physiotherapy and live-style interventions aiming BMI and body fat content reduction on their possible course of disease.