Introduction

Osteoarthritis (OA) of the knee has been detected by radiography in 15–85% of patients followed for 10–15 years after anterior cruciate ligament (ACL) reconstruction [1,2,3,4]. Although several studies have attempted to identify risk factors for OA after ACL reconstruction [5,6,7,8], whether ACL reconstruction can prevent OA in knees with ACL injury remains controversial. Furthermore, previous reports regarding the incidence of radiographic OA after ACL reconstruction were based on the long-term results of surgeries performed more than 15 years before the OA finding. Recent progress in surgical techniques, which focus on anatomic reconstruction, could improve the stability and kinematics of knees after ACL reconstruction. Therefore, the incidence of OA after modern ACL reconstruction might be lower than what has been reported in the literature. However, it would take another 10 or 15 years to demonstrate the incidence and characterise the risk factors for radiographic OA after modern ACL reconstruction.

To detect subclinical early degenerative changes before radiographic OA changes emerge, quantitative magnetic resonance imaging (MRI) techniques such as T1ρ relaxometry, T2 relaxometry, and delayed gadolinium-enhanced MRI of cartilage (dGEMRIC) have been used recently. These methods provide information on early tissue matrix degeneration substantially earlier than standard morphological assessments based on clinical MRI studies [9,10,11]. Previous studies using these methods reported changes in cartilage composition even in reconstructed knees from 6 months to 3 years after surgery [12,13,14,15]. However, some studies followed patients for only 6 months, while other studies investigated only one time point after surgery and did not compare preoperative and postoperative findings. In particular, the relationship between knee laxity after surgery and differences in T1ρ values has not previously investigated.

The purposes of this study were to evaluate subclinical early degenerative changes in patients who underwent ACL reconstruction by comparing T1ρ values before and 2 years after surgery prior to the emergence of OA changes and to investigate the relationship among surgical factors, clinical outcomes, and differences in T1ρ values.

It was hypothesised that ACL reconstruction can prevent early degenerative changes in articular cartilage and that a greater increase in T1ρ is correlated with a worse postoperative outcome.

Materials and methods

A total of 50 patients who underwent unilateral primary ACL reconstruction were recruited for this study. All study patients were evaluated using the MRI protocol mentioned below before and 2 years after surgery. The study group consisted of 25 men and 25 women. Patients who underwent ACL reconstruction more than 2 years after ACL injury were excluded from this study because Osaki et al. reported that the T1ρ values of articular cartilage were significantly higher before ACL reconstruction in patients who underwent surgery more than 2 years after their ACL injury [16]. None of the patients had radiographic OA changes before surgery. Individuals with other ligamentous injuries or surgeries were also excluded from the study. The mean age at surgery was 26.4 ± 10.5 years (range 15–53 years) and body mass index (BMI) was 23.2 ± 3.1 kg/m2 (range 18.7–34.3 kg/m2). The mean time from injury to reconstruction was 117 ± 126 days. Ethical approval was obtained from the institutional review board at our institution, and informed consent was obtained from all patients before their participation. The preoperative femorotibial angle (FTA) was measured using weight-bearing anteroposterior (AP) radiographs. The posterior tibial slope (PTS) was measured using lateral radiographs.

Surgical technique

All ACL reconstructions were performed arthroscopically. The surface of the cartilage was classified using the International Cartilage Repair Society (ICRS) grading system as follows: grade 0: normal cartilage, grade 1: superficial lesions, grade 2: defect less than 50% of the cartilage depth, grade 3: defect more than 50% of the cartilage depth, grade 4: defect down to the subchondral bone [17]. For the cartilage of the medial femur, there were 42 grade 0 knees, 5 grade 1 knees, 2 grade 2 knees, and 1 grade 3 knee. For the lateral femur, there were 48 grade 0 knees and 2 grade 1 knees. For the medial tibia, there were 49 grade 0 knees and 1 grade 1 knee. For the lateral tibia, there were 47 grade 0 knees and 3 grade 1 knees. Meniscal injury was treated appropriately with resection or meniscal sutures according to the type of tear. Patients were divided into two groups based on whether meniscectomy or meniscal suture was needed. Twenty-one patients needed a meniscectomy or meniscal sutures. Thirty-three patients underwent double-bundle ACL reconstruction with an autogenous semitendinosus tendon graft. With this method, two femoral tunnels were drilled within the native ACL femoral footprint through an anteromedial portal and two tibial tunnels were drilled within the tibial footprint in an outside-in fashion. Two double-folded semitendinosus tendon grafts were fixed on the femoral end with TightRope RT implant (Arthrex, Inc., Naples, FL, USA). On the tibial end, the grafts were fixed with a double-spike plate (DSP; MEIRA Corp., Nagoya, Aichi, Japan). Seventeen patients underwent single-bundle reconstruction with an autogenous bone-tendon-bone (BTB) graft. Femoral tunnels were drilled within the femoral footprint through the anteromedial portal and tibial tunnels were drilled in an outside-in fashion. A TightRope BTB was used to fix the femoral end, and an interference screw (Smith and Nephew Inc., Andover, MA, USA) was used for the tibial end. For both methods, range of motion exercises and partial weight-bearing with a functional knee brace (Breg, Inc., Carlsbad, CA, USA) were initiated 1 week after surgery. Full weight-bearing was allowed at 4 weeks after surgery.

MRI protocol and T1ρ mapping imaging assessment

MRI was performed on a 3-T system (Achieva 3.0T, Quasar Dual; Philips Healthcare, Best, The Netherlands) using an eight-channel phased-array knee coil. The MRI protocol was previously described [16]. Two-dimensional (2D) sagittal T1ρ mapping was generated from T1ρ-prepared images using the fast-field echo technique. T1ρ mapping was produced with Philips research integrated development environment (PRIDE) software written in the Interactive Data Language (IDL 6.3; ITT Inc., White Plains, NY, USA). T1ρ mapping was used for quantitative assessment. T1ρ values were calculated using MIPAV (medical image processing, analysis, and visualization) software (Biomedical Imaging Research Services Section, Centre for Information Technology, National Institutes of Health, Bethesda, MD, USA).

Four compartments of the knee were evaluated: medial femoral condyle (MFC), lateral femoral condyle (LFC), medial tibia (MT), and lateral tibia (LT). Femoral regions of interest (ROIs) in the cartilage were partitioned into three areas. From the centre of the circle marking the approximate circumference of each posterior femoral condyle, a line parallel to the distal femoral axis was drawn. Areas on either side of the line parallel to the femoral axis were defined as follows: the anterior area was 45° from the line (MFC1, LFC1), the middle area was 45° posterior of the line (MFC2, LFC2), and the posterior area was positioned 45° to 90° posterior of the line (MFC3, LFC3). ROIs on the tibial cartilage were divided into anterior (MT1, LT1) and posterior areas (MT2, LT2) (Fig. 1). This method demonstrated excellent agreement according to intraclass and interclass coefficients (intraobserver reliability: 0.95, interobserver reliability: 0.84). Three MR sagittal images, consisting of a centre slice of the medial or lateral compartment and both adjacent slices, were analysed within each ROI. The mean T1ρ value of the three slices within each ROI was calculated using the mean T1ρ value of each slice and the number of pixels in each slice. Segmentation was manually corrected to avoid artefacts caused by synovial fluid or other surrounding tissue. For each ROI, the mean T1ρ value before surgery was compared to the value 2 years after surgery. The difference in T1ρ values between these two time points (2 years after surgery minus before surgery) was defined as ΔT1ρ. The preoperative condition of the cartilage was also assessed with Whole-Organ MRI Scoring (WORMS) using the same compartmentalization as the ROIs for T1ρ mapping [18]. The correlation between ΔT1ρ and variables such as age, sex, BMI, meniscal injury, presence of a cartilaginous lesion, WORMS score for cartilage, and graft type was evaluated. All measurements were performed by one observer and were repeated in a blinded manner during the course of two sessions at least 1 month apart. Another observer independently made measurements of five randomly selected knees.

Fig. 1
figure 1

Each region of interest (ROI) was defined as follows: ROIs on the femoral articular cartilage were divided into an anterior area 45° from the line (MFC1, LFC1), a middle area 45° posterior from the line (MFC2, LFC2), and a posterior area 45° to 90° posterior from the line (MFC3, LFC3). ROIs on the tibial articular cartilage were divided into an anterior (MT1, LT1) and posterior area (MT2, LT2). Left: medial compartment. Right: lateral compartment. MFC medial femoral condyle, MT medial tibia, LFC lateral femoral condyle, LT lateral tibia

Full size image

Clinical assessment

At 1 year after surgery, clinical assessment of AP knee stability was based on the side-to-side difference relative to the normal contralateral knee on a stress radiograph taken with the knee in 30° of flexion with a 15-kg anterior stress applied by a Telos arthrometer. Anterolateral rotational instability (ALRI) [19,20,21] was measured with open MRI as the translation of the tibia relative to the femur at the centre of the lateral compartment during Slocum’s ALRI test performed during the open MRI examination. This value was compared with that of the normal contralateral knee. At 2 years after surgery, the Lysholm score, Tegner activity score [22], and the Knee injury and Osteoarthritis Outcome Score (KOOS) [23] were obtained using self-administered questionnaires. Correlations between ΔT1ρ values and assessments of knee stability and clinical outcomes were also evaluated. This study was approved by Institutional Review Board of Kyushu University (ID number of the approval: 23-75) and informed consent was obtained from all patients before their participation.

Statistical analysis

A sample size calculation was performed before conducting this study using JMP Pro software version 12.0 (SAS Institute, Cary, NC, USA). The results of a previous study that used the same measurement were used as pilot data [16]. In the present study, the minimal detectable value was set to 4.3 ms, assuming a standard deviation of 6.5 ms and a statistically significant value of P < 0.05. The sample size analysis revealed that a total of 20 knees were needed to obtain a power of 80%. All data are expressed as means ± SD, and analysis was performed using JMP Pro. The paired t test was used to compare differences in T1ρ before and 2 years after surgery in each ROI. To evaluate the effect of cartilaginous lesions, participants were divided into two groups (ICRS grade 1–3 in at least one compartment versus grade 0 in all compartments). To evaluate the effect of WORMS score for cartilage, participants were divided into two groups (WORMS grade > 1 versus grade 0 in each ROI). Student’s t test was used to detect the difference in mean ΔT1ρ between the two groups stratified by sex, presence of meniscal injury, presence of a cartilaginous lesion evaluated by arthroscopy, WORMS score for cartilage evaluated by MRI, and graft type (double-bundle or single-bundle). Pearson’s correlation coefficients were used to investigate the correlation between ΔT1ρ and age, BMI, time to surgery, and side-to-side differences on stress radiographs and ALRI for each ROI. Spearman’s correlation coefficients were used to evaluate the correlation between ΔT1ρ and the Lysholm score, KOOS subscale scores, and Tegner activity score for each ROI. Multiple regression analysis was also used to assess the contribution of these factors.

Results

Mean preoperative FTA and PTS were 174.6 ± 1.5° and 9.3 ± 2.5°, respectively. Mean T1ρ values in each area of cartilage on the femoral condyles and tibial plateau are summarised in Table 1. The paired t test revealed a significant increase in T1ρ before versus 2 years after surgery in the MT1, MT2, LFC1, and LT1 areas, and a decrease in the LFC3 and LT2 areas. Table 2 summarises WORMS score for cartilage. Correlations between ΔT1ρ in each ROI and surgical factors were also demonstrated in Table 3.

Table 1 Mean T1ρ values in each area of cartilage on the femoral condyles and tibial plateau
Full size table
Table 2 Preoperative WORMS for cartilage
Full size table
Table 3 Correlations between ΔT1ρ in each ROI and surgical factors
Full size table

The mean anterior side-to-side difference was 1.6 ± 2.0 mm on stress radiographs. The mean difference in ALRI was 1.5 ± 4.1 mm. Pearson’s correlation coefficient analysis showed a significant correlation between side-to-side difference on stress radiographs and ΔT1ρ in the MFC3 (r = 0.37, P = 0.013) and MT2 (r = 0.35, P = 0.021) areas (Table 3). Multiple regression analysis of these surgical factors showed a significant correlation only between side-to-side difference on stress radiographs and ΔT1ρ in the MFC3 (β = 0.36, P = 0.020) and MT2 (β = 0.38, P = 0.013) areas.

The results of the Lysholm score, Tegner score, and KOOS are shown in Table 4 and Fig. 2. ΔT1ρ for the MFC2 area was negatively correlated with KOOS symptoms (ρ = -0.349, P = 0.027) and quality of life (QOL) (ρ = − 0.374, P = 0.017) subscale scores.

Table 4 Lysholm score and Tegner score
Full size table
Fig. 2
figure 2

KOOS profile. KOOS The Knee Injury and Osteoarthritis Outcome Score, ADL activity of daily living, QOL quality of life

Full size image

Discussion

The most important finding of this study was that an increase in T1ρ in the medial compartment is likely to occur even with ACL reconstruction. Significant increases in T1ρ were observed in the anterior femur and tibia (MT1, MT2, LFC1, and LT1). In contrast, T1ρ in the posterolateral femur and tibia (LFC3 and LT2) decreased. These trends are consistent with previous studies of T1ρ and T2 mapping for knees after ACL reconstruction [14, 15, 24]. On the medial side, altered kinematics of the knee after ACL reconstruction may cause articular cartilage damage [25], leading to higher T1ρ values [26]. The reason for the decrease on the lateral side is that the posterolateral area had experienced cartilage damage at the time of injury [27], which led to decreased T1ρ values during the postoperative period. Whereas previous studies had short follow-ups (e.g., 6 months or 1 year) or MRI examination at only one time point, this study followed patients for 2 years postoperatively and MRI was performed at preoperative and postoperative visits. Longitudinal analysis with the paired t test enabled a more robust evaluation. Moreover, our method for determining ROI was possibly more reliable for longitudinal studies than past studies [9, 12, 14]. It enabled the determination of ROIs independent of the position of the meniscus, which depends on the angle of knee flexion. Instead, our method was based on bone morphology and enabled us to have minimal impact on the area of interest while identifying the same location during both time points.

In this study, the difference between preoperative and postoperative T1ρ values (ΔT1ρ) was calculated and a relationship between ΔT1ρ and surgical factors was detected. There was a significant correlation between sex and ΔT1ρ in the MFC1 area. According to a systematic review, many studies assessed female sex as a potential risk factor for OA [28]. Higher age was correlated with smaller ΔT1ρ in MFC3 and MT2. It is important that the difference in T1ρ values before and after surgery was evaluated using MRI to overcome the concerns due to the baseline variations among each individual. For example, if the T1ρ value were already elevated preoperatively, the difference would potentially be small. Moreover, there was a significant negative correlation between age and sports activity as reflected by the Tegner score at 2 years after surgery (Spearman’s correlation; ρ = − 0.341, P = 0.034), which might have affected the results. There was no significant correlation between ΔT1ρ and duration to surgery. It is well known that a longer duration to surgery is associated with greater degenerative changes in cartilage, resulting in clinical OA changes of the knee [29,30,31]. Because this study recruited patients who underwent surgery within 2 years of injury, ΔT1ρ might not be correlated with duration to surgery. The presence of meniscal injury did not affect ΔT1ρ of any ROI. Past studies noted that meniscal injury is a risk factor for degenerative changes in knee articular cartilage [5, 15, 16]. In contrast, some studies have demonstrated no significant difference between patients with or without meniscal tears [13] and between patients who underwent meniscectomy versus meniscal repair [32]. In this study, we treated meniscus injury with repair as much as possible in accordance with current trends to save the meniscus [33,34,35], which might result in no significant increase in T1ρ. However, meniscus injuries involve many factors, such as type, grade, and location of the injury, and treatment type. Therefore, it is difficult to assess the influence of meniscus injury precisely, which might have affected the results. Regarding the presence of a cartilaginous lesion observed with arthroscopy during ACL reconstruction and WORMS score for cartilage evaluated by preoperative MRI, there was no significant difference whether a cartilaginous lesion was present, although Hirose et al. showed that cartilaginous lesions are related to progressive degenerative changes in cartilage [13]. Few patients with cartilaginous lesions and WORMS grade > 1 were included in our study, so a significant difference might not have emerged. There was no significant difference in ΔT1ρ between the two graft types in this study. Clinical results for both graft types have been inconsistent [36, 37]. Further study with more patients that have varying degrees of cartilaginous lesions with reconstruction using different graft types could clarify the influence of such factors on T1ρ values.

AP laxity demonstrated on stress radiographs affected ΔT1ρ values in the MFC3, MT2, and LFC2 areas. Multiple regression in a model with other surgical factors showed a significant relationship between AP laxity and ΔT1ρ in the MFC3 and MT2 areas. Past studies with long-term follow-up showed significantly more severe degenerative changes and OA progression in patients with increased AP laxity [38]. When the reconstructed ACL is not functional and has laxity, the loading pattern and amount of load might differ. Otherwise, there was no significant relationship between ΔT1ρ in each ROI and rotational laxity expressed by the ALRI test, which was inferred to be due to rotational laxity being less detectable than AP laxity [19, 20]. No other studies have demonstrated a relationship between knee laxity and quantitative MRI values. Thus, AP laxity is possibly a risk factor for early cartilage degeneration, especially in the posteromedial area. Elevated T1ρ values on the lateral side on the preoperative MRI might offset the increase in T1ρ due to laxity, which might explain why there was no correlation between AP laxity and ΔT1ρ of the lateral area and why the ALRI results did not affect ΔT1ρ in any ROI.

ΔT1ρ for the MFC2 area was negatively associated with KOOS symptoms and QOL subscale scores. There have a few reports about the relationship between MRI relaxation time and clinical outcomes [12, 15]. However, these reports were based on short follow-up periods and did not consider differences before versus after surgery. In this study, the difference in T1ρ before versus 2 years after surgery was calculated and a relationship between this difference and clinical outcomes was detected.

This study has several limitations. First, our study included a relatively small number of patients even though it was large enough from a statistical perspective to detect a difference in T1ρ values. Some multiple regression results might have been affected by the study’s sample size. In addition, various surgical factors potentially acted as confounders, so there was a possibility that the factors that lead to subtle but important changes in T1ρ had been overlooked. Second, we evaluated cartilage degeneration only with T1ρ values. Although T2 and dGEMRIC are also established methods for detecting early cartilage degeneration, a systematic review and meta-analysis showed T1ρ relaxometry is superior to T2 relaxometry and dGEMRIC for discriminative validity [39]. Therefore, we believe we could conduct a reliable evaluation with T1ρ relaxometry. Third, uninjured contralateral knees in patients with ACL tears were not assessed. Previous studies have shown increases in T1ρ values in the uninjured contralateral knee in a defined period of time [12, 40]. However, a longitudinal assessment was conducted and differences between preoperative and 2-year postoperative T1ρ values were evaluated, which enabled the identification of factors affecting increases in T1ρ. Fourth, some intensity alternation in the condyles because of the impact during injury might affect T1ρ values. Patients who underwent surgery less than 2 years after their ACL injury were recruited for this study based on a previous study [16] to minimise the influence of differences in T1ρ values among patients. However, there were some differences even among patients who received surgery within 2 years of injury, especially on the lateral side. Finally, this cohort of patients was only followed for 2 years whereas past studies had longer follow-up periods. Although it is unclear whether the increase in T1ρ results in clinical OA, we believe we were able to detect subclinical degenerative changes in cartilage in this study. Longer follow-up is needed to determine the threshold T1ρ value for clinical OA.

Conclusion

An increase in quantitative MRI T1ρ values in the medial compartment likely occurs even with ACL reconstruction. AP laxity made T1ρ values increase significantly in the posteromedial femur and tibia. Increased T1ρ in the central medial femoral condyle is associated with knee symptoms.