Introduction

The development of degenerative changes within the knee following anterior cruciate ligament (ACL) injury is well recognized. However, defining the exact prevalence of osteoarthritis (OA) after ACL injury or reconstruction of the ACL is a challenge for several reasons. Long-term clinical follow-up studies are difficult to perform, since most patients presenting with an ACL injury are young and many will change geographic location in the years following surgery. Outcome studies with long-term follow-up periods tend to have a large number of patients lost to follow-up for this reason. These factors underlie the widespread variation in the estimates of OA prevalence, with values varying from 10% to 90%.

Furthermore, the group of patients with these injuries is heterogeneous, with widely varying ages, preinjury levels of activity, and different expectations following treatment. ACL tears may occur in isolation, but a significant proportion is associated with collateral ligament injuries and concomitant or subsequent meniscal tears, which may also influence the development of degenerative change. In addition to these considerations, some patients may have chondral damage or degenerative changes within the knee at the time the ACL injury is sustained. Notwithstanding these considerations, in the past number of years there have been numerous contributions to the literature based on longer-term follow-up of prospective cohorts and data analysis from large population databases, which has been helpful in providing more evidence about the incidence of osteoarthritic change following ACL injury and reconstruction.

Animal Models

Animal models have been used to study the effects of ACL injury on the knee, and some studies have looked at the development of arthritic change. Clearly the mechanics of animal knee joints will differ from human joints, and some models such as that of the mouse will show a faster progression of arthritic change in response to an experimental intervention. In a mouse model evaluating OA progression, greater degrees of ligament instability were associated with more rapid progression of arthritic change. Kamekura et al. engaged in histological evaluation of articular cartilage in a mouse model with an injury incorporating ACL tear and evaluated the animals with histological assessment of articular cartilage, demonstrating degenerative change by 56 days following injury with minor fissuring of articular cartilage, frequent loss of the surface zone, and the flattened elongated chondrocytes of the upper zone, cell death in the articular cartilage superficial zone, and atrophy of articular chondrocytes.

Other animal studies tend to reflect to some extent the rather variable findings in clinical studies in humans. Tochigi et al. found a correlation between instability and histological OA scores in a rabbit model that studied partial and complete ACL transection. Models evaluating ACL reconstruction (ACLR) show contrasting results. O’Brien et al. did not show a chondroprotective effect of ACLR in an ovine model, with more osteoarthritic changes being observed in an ovine model of ACL tears, followed by reconstruction in comparison with controls. Murray and Fleming showed in a porcine model that knees reconstructed with a biologically enhanced ACL graft developed less degenerative articular lesions 1 year following surgery. These findings illustrate the variability that tends to characterize animal studies, evaluating the relationship of degenerative change and ACL injury and reconstructive surgery in animal models, and the difficulty in extrapolating findings to humans.

Biomarkers and Pathophysiology of Osteoarthritis Following Anterior Cruciate Ligament Injury

Studies have shown that restoring knee stability through ACLR does not have a clear-cut relationship with the development of posttraumatic OA. It therefore follows that there may be other mechanisms, rather than the mechanical disturbance of stability, that are responsible for the development of OA, both in the chronic ACL-deficient (ACL-D) knee and in the reconstructed knee. There has been increased interest in biochemical markers associated with the pathophysiology of OA following ACL tears. It is also possible that some biomarkers may be useful in the early identification of patients at risk of developing degenerative change. However, a wide range of biomarkers have been studied, and their precise role in diagnosis and management or prevention in OA remains to be established.

A recent systematic review of the relevant literature identified 20 studies evaluating biomarkers after ACL injury. These studies included those evaluating ACL-D knees (12) and studies of ACL-reconstructed knees (ACL-R, 8). In these studies biomarkers were assessed in synovial fluid, serum, plasma, and urine. The biomarkers studied can be classified either as collagen, proteoglycan, or bone biomarkers, or as inflammatory cytokines.

Perhaps not surprisingly, inflammatory cytokines such as IL-1ra14, IL-614, IL-814, and TNF-a14 have been reported to be elevated in the early weeks after injury in synovial fluid. Following ACL injury, synovial fluid also had elevated levels of proteoglycan fragments, matrix metalloproteinase (MMP), and tissue inhibitor of metalloproteinase TIMP-1 in the first 2 months after injury and, beyond this, greater synovial fluid concentrations of procollagen II C-propeptide (CPII), MMP-3, and TIMP-1. However, 2 years following injury, synovial fluid concentrations of proteoglycan fragments and TIMP-1 did not differ from controls.

Changes in biomarkers have also been noted after ACLR. Inflammatory markers such as interleukins are elevated, particularly in the early stages after surgery. Within weeks of surgery, ACL-R patients have increased synovial fluid concentration of TIMP-1 and greater urinary concentrations of C-terminal cross-linked telopeptide of type II collagen (CTX-II) when compared with control participants. Even after the first year following surgery, ACL-R patients had greater concentrations of urinary CTX-II, and a larger ratio of urinary CTX-II to serum CPII procollagen II C-propeptide (CPII) when compared with control participants.

At the present time it is not certain which biomarkers have the greatest prognostic significance in relation to development of OA after ACL injury and reconstruction. Based on the current literature it seems likely that markers associated with increased collagen and proteoglycan breakdown will be most closely associated with an increased risk of OA. These biomarkers may have a role in identifying patients most at risk of OA, and for defining a high-risk group who would benefit either from other interventions or perhaps modification of sporting activity levels.

Biomechanics

The important role of the ACL in knee kinematics is well described, and loss of this function is known to be associated with an increased risk of OA. The role of biochemical disturbance with chondral breakdown has been referenced, but perhaps the most obvious explanation is that the instability that occurs as a consequence of ACL injury results in a hostile biomechanical environment within the knee, and this in turn is responsible for the development of knee OA. The loss of ACL function is known to result in a number of alterations to normal knee biomechanics. This includes deterioration of the physiologic roll-glide mechanism, resulting in increased anterior tibial translation, increased internal tibial rotation, and an increased mean contact stress in the medial and lateral compartment posterior sectors under anterior and rotational loads, respectively. It has also been shown that the tibia is significantly more internally rotated in an ACL-D knee, possibly interfering with the screw-home mechanism of tibiofemoral kinematics. During stair ascent and descent, as well as during the entire gait cycle, ACL-D knees display a more varus and internally rotated tibial position when compared with ACL-intact knees. Significant reductions in extension were observed during the midstance in ACL-D knees, but with significantly higher anterior tibial translation and higher flexion angles than the intact contralateral side. In high-demand activities such as side-cutting motions, the ACL-D knee increases offset toward less valgus and more external tibial rotation potentially as an adaptation to avoid pivot shift dynamically.

Under normal circumstances, synovial joints can tolerate physiological forces without developing OA. The changes in joint kinematics as noted previously will result in nonphysiological loading that may result in articular cartilage damage. It has been shown that excessive contact stresses applied to articular cartilage can result in damage to the articular cartilage and the underlying subchondral bone with associated alteration of normal chondrocyte function. Failure of the ACL also results in excessive shear loading of the menisci, which is known to be associated with an increased risk of meniscal tears and in turn contributes to acceleration of the degenerative process.

All of this evidence would logically support the rationale for restoration of normal kinematics by ACLR, but studies on knees following the reconstruction show that although knees have improved kinematics following ACLR, abnormalities remain. This failure to restore completely normal biomechanical behavior may be one of the factors contributing to a lack of convincing evidence that ACLR prevents osteoarthritic change.

Meniscus Tears

In general most studies reporting on the risk of OA after ACL injury have identified concomitant or subsequent meniscectomy as associated with an increased risk of development of OA. The need for meniscal resection at the time of surgery has been identified as being associated with a poorer outcome and higher rates of degenerative change at the time of follow-up. The meniscal protective effect of ACLR has been documented in outcome studies. Perhaps because of the complex multifactorial causation of OA following ACLR, it has been difficult to prove that the ACL-R protective effect on the meniscus is associated with a reduction in the risk of OA. In the most recent review and meta-analysis of the literature, the numbers of patients with long-term radiological follow-up were considered inadequate to allow statistical analysis of the effect on meniscectomy and ACL injury and reconstruction on the risk of OA development.

Chalmers et al. published a review of studies with long-term outcomes at 10 years and evaluated among other outcomes the requirement for further surgery and radiographic changes. They identified 27 cohorts containing 1585 patients who had undergone reconstruction, and 13 containing 685 patients who had been treated nonoperatively. Analysis of the data in these studies confirmed that ACL-R knees had a lower requirement for further surgical intervention and, in particular, a reduced rate of meniscal surgery. One would expect that this might translate into a lower rate of osteoarthritic change, but this was not the case. They noted that radiographically evident degenerative joint disease was observed in 35.3% of reconstructed knees and in 32.8% of those treated nonoperatively, a difference that was not significant. They concluded that ligamentous instability and meniscal tears were not the main factors in the development of osteoarthritic change after ACL injury. The authors pointed out that these studies included many patients treated with older techniques of reconstruction, which may not be as successful as modern methods, and the mean time from injury to reconstruction was 20 months, which would be considered a long interval by modern standards.

Chondral Lesions

Another well-recognized feature of ACL injury is the association with chondral lesions. Several large recent studies based on registries of ACL injury and reconstruction have yielded more information about the relationship of these lesions to the injury and possibly a link to development of subsequent arthritic change within the knee. Along with meniscal tears, chondral lesions are linked to an increased risk of degenerative change developing in the knee. The pathogenesis of these lesions may have more than one cause. They may be a consequence of direct chondral damage at the time of injury. It has been shown that ligamentous knee injury is strongly associated with bone bruising. Magnetic resonance imaging scans performed acutely following ACL rupture have shown occult subchondral lesions in 85% of patients, mainly involving the lateral femoral condyle and lateral tibial plateau. Histologic analysis of these bone bruises has shown associated areas of chondrocyte degeneration and necrotic osteocytes, which suggests that significant damage to the articular cartilage is sustained at the time of injury.

Scandinavian databases on ACL injury and surgery have provided good quality data regarding chondral lesions in ACL-injured knees undergoing reconstruction. Røtterud et al. reported on the prevalence of chondral lesions at the time of ACLR in 15,783 primary ACLRs. They recorded that 4196 (26.6%) of the patients had 1 or more articular cartilage lesions (ICRS grades 1–4), and of these, 1012 patients (6.4%) had one or more full-thickness cartilage lesions. The most common location was the medial femoral condyle (51% of the full-thickness lesions were located here). Interestingly, they noted that the risk of a full-thickness defect was independent of time from injury to surgery within the first 12 months, but that the risk of chondral defects increased after that time point. The same authors in separate studies noted that patients with full-thickness articular defects had inferior outcomes after ACLRs, compared with patients with no cartilage defects. Cox et al. reported very similar outcomes in 1512 ACLR patients followed for 6 years. Chondral defects were noted on the medial femoral condyle and lateral femoral condyle in 25% and 20% of patients, respectively, at the time of surgery and were associated with significantly reduced International Knee Documentation Committee (IKDC) and Knee injury and Osteoarthritis Outcome Score at 6 years.

Natural History of the Anterior Cruciate Ligament-Deficient Knee

The development of arthrosis following ACL rupture has been recognized for a long time. Early studies reported very high rates of OA associated with ACL tears. In a review by Gillquist and Messner in 1999, it was concluded that in the long term (10–20 years) as many as 70% of all ACL-D knees had radiological signs of arthrosis, although clinical symptoms of knee arthritis were infrequent. Older studies were compromised by smaller numbers, variable degrees of follow-up, and less stringent clinical definitions of OA. More recent reviews have had the advantage of a larger evidence base with more published literature and longer term follow-up, with more exactly defined outcomes. Øiestad et al. published a comprehensive review in 2009 that includes a systematic review of 7 prospective and 24 retrospective studies involving 3069 patients. This review only included studies with a minimum of 10-year follow-up. The prevalence of OA after 10 years in the isolated ACL-D knee without meniscal injury ranged from 0% to 13%. In studies reporting the outcome of ACL tears combined with meniscal injury, the prevalence of OA was between 21% and 48%. No differences were reported for the prevalence of knee OA between surgically treated (29%–51%) and nonsurgically treated subjects with ACL injury (31%–48%). As the authors of the review point out, there is some difficulty in interpretation of the data, as several methods of defining the presence of OA are in common use. Classification systems used in studies have included those described by Ahlbäck , Fairbank , Kellgren and Lawrence , and the IKDC. This variation in osteoarthritic grading systems makes the comparison of outcomes in studies difficult.

In order to overcome this drawback in literature interpretation, Ajuied et al. carried out a meta-analysis of studies, reporting the prevalence of OA which used the Kellgren and Lawrence classification exclusively with a minimum of 10-year follow-up. They identified nine studies suitable for inclusion, with a total of 615 patients. The majority (85%) of patients underwent ACLR. At the time of final follow-up, 20.3% of ACL-injured knees had moderate to severe osteoarthritic changes. Using the contralateral normal knee as a control, the same extent of osteoarthritic change was found in 4.9%. The authors calculated the relative risk of developing moderate to severe OA following ACL injury was 3.84. ACLR did not influence the development of OA, although it has to be taken into account that the majority of patients in this meta-analysis underwent reconstruction so the comparison group of nonoperatively treated knees was rather small.

Arthrosis Following Anterior Cruciate Ligament Reconstruction

With the knowledge that ACL tears are associated with an increased risk of OA, it would seem a reasonable assumption that ACLR would play a useful role in prevention of arthrosis in the long-term. However a meta-analysis by Lohmander and Roos in 1994 found no evidence that ACLR slowed the progression of arthrosis following ACL rupture. There are factors which may have contributed to poor outcomes with higher rates of OA in earlier studies. Early surgical reconstruction was often not routinely offered to patients, and many patients will have had a relatively long period of symptomatic instability before reconstruction. The prevalence of meniscectomy and chondral lesions would therefore have been higher. Finally, techniques of reconstruction were more variable and technical errors in surgical reconstruction were probably more common, which would also have contributed to the development of OA.

Despite these considerations, more recent reviews have yielded similar conclusions in relation to the influence of ACLR on the development of OA. Øiestad et al. reported conflicting evidence of the role of ACLR. In their review they reported no differences for the prevalence of knee OA between surgically treated (29%–51%) and nonsurgically treated subjects (24%–48%). In the most recent meta-analysis, Ajuied et al. reported on six studies with a mean 10-year follow-up using the Kellgren-Lawrence grading for OA. They reported that after ACLR in knees with or without meniscal injury, the relative risk (RR) of developing OA was 3.62 when compared with the uninjured control group at follow-up, whereas nonoperatively managed knees had an RR of 4.98, a modest difference. They concluded ACLR has a role in the prevention of OA after ACL injury. However, they also observed that ACL-R knees tended toward a higher proportion of moderate and severe osteoarthritic changes (grades III and IV; RR = 4.71), when compared with ACL-injured knees managed nonoperatively (RR = 2.41). These findings require careful interpretation. One contributing feature was the heterogeneity of the included studies for nonoperative treatment. An alternative explanation is that patients being managed nonoperatively are often advised to modify sporting activity or voluntarily choose to avoid sports associated with a high risk of knee instability, which may diminish the risk of OA developing. Patients after successful reconstruction are more likely to return to a full range of unrestricted sporting activity, which may be associated with factors likely to influence development of OA, such as reinjury and meniscal tears.