MRI evaluation of ACL reconstruction

As the number of ACL reconstructions increased in the 1990s and 2000s, the MRI imaging of the postoperative knee also became more common (Recht and Kramer 2002). The indications for postoperative evaluation of ACL reconstruction with MRI include laxity, stiffness, pain, a new injury, preoperative planning of revision, and infection of the knee (Recht and Kramer 2002, Papakonstantinou et al. 2003).

Graft failure is a clinical term defined as pathologic knee laxity after ACL reconstruction (Bencardino et al. 2009). The reported conventional ACL reconstruction graft failure rate is 3-5% (Spindler et al. 2003, Ahlden et al. 2012). In total, 10-30% of patients have knee pain or laxity following conventional ACL reconstruction (Yunes et al. 2001). Knee pain is the most common but nonspecific symptom of complications of ACL reconstruction, thus making further diagnostic imaging necessary (Passler et al. 1999, Collins et al. 2008).

MRI is the best imaging modality for postoperative ACL reconstruction evaluation due to its superior soft tissue contrast, multiplanar capability, and lack of ionizing radiation (Farshad-Amacker and Potter 2013). The complications of ACL reconstruction that are diagnosed using MRI include complete and partial graft ruptures, graft impingement, arthrofibrosis, graft or tunnel cysts, hardware failure, and infection (Recht and Kramer 2002, Papakonstantinou et al. 2003, Bencardino et al. 2009). Some authors also consider tunnel widening a complication (Giaconi et al. 2009, Farshad-Amacker and Potter 2013). Because the clinical significance of many MRI findings is unclear at the moment, it is important to correlate the MRI findings with clinical examination (Casagranda et al. 2009).

4.9.1 MRI techniques

Most imaging protocol recommendations in the literature are for 1.5 tesla (T) scanners. However, the same diagnostic properties of the sequences are achieved with 3 T scanners that produce better images with shorter scanning times (Craig et al. 2005). Spin-echo (SE) sequences are recommended for ACL graft imaging because metallic debris and orthopedic hardware produce more artifacts in gradient-echo sequences (Papakonstantinou et al. 2003). The recommended slice thickness for 1.5 T is 3-4 mm for axial, sagittal, and coronal images, and 2.5 mm for oblique images (Casagranda et al. 2009, Araujo et al. 2013).

The integrity of the graft is assessed with sagittal PD-weighted and T2-weighted fat-saturated (FS) images and preferably with sagittal or coronal oblique PD-weighted or T2-PD-weighted images along the course of the graft (Papakonstantinou et al. 2003, Casagranda et al. 2009, Bining et al. 2009). Some authors recommend prescribing the routine sagittal plane along the lateral femoral condyle to minimize individual variability (Casagranda et al. 2009). In studies of graft signal intensity (SI), T2-weighted and PD-weighted sequences in sagittal, coronal, and oblique sagittal planes have been used (Sonoda et al. 2007, Saupe et al. 2008).

Coronal T1-weighted images are used to evaluate bone structure and coronal and sagittal T2-weighted FS images to assess the graft, bone bruise, and articular cartilage (Casagranda et al. 2009, Araujo et al. 2013). Axial T2-weighted FS images are recommended for the evaluation of the femoral attachment of the graft and also for graft integrity (Casagranda et al. 2009, Araujo et al. 2013). In studies of tunnel enlargement and tunnel communication, T1-weighted images in axial, coronal, and sagittal planes have been used (Siebold 2007, Järvelä et al. 2008, Siebold and Cafaltzis 2010).

4.9.2 Graft ligamentization period

The SI changes of the conventional ACL reconstruction graft during the first 12-24 months after operation are called the ligamentization period (Amiel et al. 1986b, White et al. 2005). Immediately after the operation, the graft has low SI in T1- and T2-weighted images. After 3-8 months, however, the graft SI increases to intermediate due to revascularization and re-synovialization of the graft (Howell et al. 1991b, Schatz et al. 1997, Rispoli et al. 2001, White et al. 2005). Then, the SI decreases to low SI resembling the native ACL 12-24 months after operation (Rispoli et al. 2001, Vogl et al. 2001, Hong et al. 2005, White et al. 2005). However,

recent studies have shown that the graft SI can remain increased at long-term follow-up after conventional ACL reconstruction with no clinical findings of graft failure (Saupe et al. 2008). Other authors have even stated that results on graft SI are contradictory and that knowledge about graft SI changes is currently insufficient (Miyawaki et al. 2014). The anatomical ACL reconstruction results in a more oblique position of the graft in the coronal plane, which has been proposed as an explanation of the prolonged increased SI of the anatomical grafts (Farshad-Amacker and Potter 2013). According to DB ACL reconstruction MRI studies, some increased graft SI can be seen after 12-24 months and should not alone be considered a sign of graft failure (Sonoda et al. 2007, Poellinger et al. 2009, Casagranda et al. 2009). In 4-strand hamstring grafts, the finding of some linear areas of increased signal between the graft strands is considered normal and represents the multistrand structure of the graft (Bencardino et al. 2009). BPTB grafts and quadriceps grafts constitute a single piece of tendon and should have more homogenous SI on MRI (Bencardino et al. 2009).

4.9.3 The magic angle effect in graft evaluation

The SI of tendons and ligaments depends on their orientation to the static magnetic field (B0) in short and moderate echo time (TE<37 ms) pulse sequences (Fullerton et al. 1985, Erickson et al. 1991, Peh and Chan 1998). In clinical imaging, this phenomenon is known as the magic angle (MA) artifact. It involves tissues that have a highly ordered collagen fibre structure: tendons, ligaments, labrums, menisci, intervertebral discs, articular cartilage, fibrocartilage, and also peripheral nerves (Bydder et al. 2007).

The highly ordered collagen fibers contain water that is bound to collagen molecules (Fullerton and Rahal 2007). The protons of water are subject to dipolar interactions whose strength depends on the orientation of the collagen fibers to B0

(Bydder et al. 2007). Usually, these interactions result in rapid dephasing of the MR signal that corresponds to the small amount of or lack of signal seen in tendons and ligaments (Bydder et al. 2007). The interactions of water bound to collagen molecules are gradually weaker in angles of collagen fibers from 20^o to 55^o to the B0 causing gradually increased SI up to sixfold in short TE pulse sequences (Fullerton and Rahal 2007). The MA is only detected if TE is shorter than 37 ms as in T1-weighted and PD-weighted pulse sequences (Peh and Chan 1998). In order to rule out the MA and to be able to reliably detect increased SI consistent with

tendon pathology, T2-weighted sequences with longer TE are recommended for MRI evaluation of tendons that course at an angle close to 55^o to B0, such as the supraspinatus and infraspinatus tendons, the patellar tendon and the peroneal tendon (Shalaby and Almekinders 1999, Wright et al. 2001, Wang et al. 2005). It is also possible to reduce the MA by patient and limb position with respect to B0

(Fullerton and Rahal 2007). In most clinical MR imagers, the B0 is oriented parallel to the table and the longitudinal axis of the patient (Fullerton and Rahal 2007). In theory, stretching a tendon while imaging may also cause SI alteration because stretching moves water out from tendon (Puxkandl et al. 2002, Fullerton and Rahal 2007).

In pathologic conditions, loss of organized collagen structure with increased free fluid, mucinous vacuoles, cellular debris, fibrin, and disorganization of collagen fiber orientation is seen in ligaments and tendons (Fullerton and Rahal 2007).

When the organized collagen structure is disturbed, the SI of tissue is increased on all MRI sequences in angles close to B0, and the MA effect of the normal collagen structure is not seen (Fullerton and Rahal 2007).

In ACL reconstruction, graft harvesting, preparation, and avascularization cause changes to the collagen structure (Amiel et al. 1986a, Goradia et al. 2000).

Specifically, the preparation of hamstring grafts by folding and suturing causes heterogenity of the collagen fibre orientation (Kondo et al. 2012). In animal models, the histologic structure of both patellar and hamstring grafts begins to resemble that of native ACL by 12 months postoperatively, but the highly organized collagen structure of the native ligament is not achieved (Amiel et al.

1986a, Goradia et al. 2000).

Because the ACL reconstruction graft angle to B0 is significantly less than 55^o in MRI, some authors state that the MA is an unlikely cause of increased graft SI in short TE sequences (Jansson et al. 2001). However, as the MA effect begins at fiber angles as low as 20^o to B0, some MA effect should be expected (Schick et al.

1995, Fullerton and Rahal 2007). In a well-designed clinical MRI study, both PD-weighted and T2-PD-weighted sequences were used in graft evaluation and increased graft SI was seen in a similar proportion of cases in both sequences (Saupe et al.

2008). In conclusion, PD-weighted sequences are not significantly affected and T2-weighted sequences are not affected by the MA effect in graft MRI evaluation (Saupe et al. 2008).

4.9.4 Graft disruption and partial rupture

A graft is considered disrupted when intact fibers are absent and fluid signal is replacing the graft (Recht and Kramer 2002, Papakonstantinou et al. 2003, Collins et al. 2008, Bencardino et al. 2009, Casagranda et al. 2009). A partial tear shows some preserved fibers and a focal thinning of the graft (Collins et al. 2008, Casagranda et al. 2009). As in MRI diagnosis of native ACL disruption, other previously described findings that suggest graft disruption such as graft laxity, increased graft SI, anterior tibial translation, uncovered horn of the lateral meniscus, and buckling of the PCL have low sensitivity but high specificity (Collins et al. 2008).

4.9.5 Graft elongation

An elongated graft has intact fibers but the knee is clinically unstable (Sanders 2002). The primary MRI finding is posterior bowing of the graft, and the secondary findings are anterior tibial translation and buckling of the PCL (Sanders 2002).

4.9.6 Graft impingement

Non-anatomical tunnel locations in ACL reconstruction may lead to graft impingement, in which the graft rubs against the roof of the femoral intercondylar notch resulting in extension loss (Recht and Kramer 2002, Karlsson et al. 2011).

The MRI findings of graft impingement are posterior bowing of the graft, contact of the graft with the notch roof, and increased SI in T2-weighted and PD-weighted images in the anterior part of the graft (Howell 1998, Recht and Kramer 2002, Papakonstantinou et al. 2003).

4.9.7 Arthrofibrosis and cyclops lesion

Another complication that causes limited range of motion is arthrofibrosis, an excessive production of fibrous tissue around the graft and capsular contraction (Papakonstantinou et al. 2003, Farshad-Amacker and Potter 2013). A cyclops lesion is localized arthrofibrosis forming a nodule anterior to the graft and results in extension loss of the knee (Recht and Kramer 2002, Papakonstantinou et al.

2003). The fibrous tissue is of low to intermediate SI in T1- and T2-weighted images (Farshad-Amacker and Potter 2013).

4.9.8 Cystic degeneration of the graft and tunnel cysts

Cystic degeneration of the graft is seen in MRI as a cystic lesion with low SI in T1-weighted images and high SI in T2-T1-weighted images between intact graft fibers (Papakonstantinou et al. 2003). If large, it may cause limitated range of motion, and if it is situated in the part of the graft inside the tunnel, cystic degeneration may be associated with tunnel enlargement (Schatz et al. 1997, Papakonstantinou et al.

2003).

Ganglion cysts may form inside the bone tunnels, mainly in the tibia (Bencardino et al. 2009, Farshad-Amacker and Potter 2013). A tunnel cyst is well defined and has high SI of fluid in T2-weighted images (Bining et al. 2009).

Usually, tunnel cysts are asymptomatic and confined inside the bone tunnel. On some rare occasions, a tibial tunnel cyst may extend distally out of the tunnel and bone and present as a palpable mass (Bencardino et al. 2009). Inside the tunnel, a tunnel cyst may cause tunnel enlargement (Frick et al. 2006, Farshad-Amacker and Potter 2013).

4.9.9 Tunnel enlargement

The enlargement of the bone tunnels occurs postoperatively after ACL reconstruction and may complicate revision surgery (Wilson et al. 2004, Samuelsson et al. 2009). The etiology of tunnel enlargement is unknown and probably multifactorial (Wilson et al. 2004, Giaconi et al. 2009). Some enlargement can be caused intraoperatively by drilling and graft fixation because interference screws extrude cancellous bone during screw insertion (Buelow et al. 2002, Jagodzinski et al. 2005, Foldager et al. 2010, Siebold and Cafaltzis 2010).

Postoperative tunnel enlargement occurs in the first 3 months after ACL reconstruction (Wilson et al. 2004). Tunnel enlargement occurs more commonly with hamstring grafts than with BPTB grafts (Samuelsson et al. 2009). More tunnel enlargement is associated with extracortical graft fixation than with aperture fixation (L'Insalata et al. 1997, Beynnon et al. 2005a). More tunnel enlargement is seen with bioabsorbable screws than with metal screws, but metal screws cause artifacts and make postoperative MRI evaluation more difficult (Brand et al. 2000,

Moisala et al. 2008). Most bioabsorbable screws are polylactide compounds and their resorption properties in clinical use are not well documentated (Konan and Haddad 2009). In a recent study, the remains of two newer-generation screws were present three years postoperatively (Cox et al. 2014). The resorption of bioabsorbable screws is hypothesized to induce production of cytokines that stimulate osteoclastic activity and lead to tunnel enlargement (Wilson et al. 2004).

In most studies, tunnel enlargement has not been associated with knee laxity or functional impairment (Samuelsson et al. 2009). Tunnel diameters can be measured using T1-weighted or PD-weighted MRI in which the tunnel margins are clearly visible (Järvelä et al. 2008, Siebold and Cafaltzis 2010).

4.9.10 Tunnel communication

Tunnel communication is seen only in DB ACL reconstruction where two graft tunnels are drilled close to each other in both the femur and in the tibia. Tunnel communication is defined as the absence of the bony bridge between the tunnels at 1 cm from the articular surface (Siebold and Cafaltzis 2010). In a prospective study with MRI done at 2 days and at 7 months postoperatively, tunnel communication was caused intraoperatively by drilling in 24% of patients in the tibia and later by tunnel enlargement in 19% of patients in both the femur and the tibia (Siebold and Cafaltzis 2010). Controversily, in other studies no postoperative tunnel communication was found (Järvelä et al. 2008, Hantes et al. 2010). The best modality to evaluate the bone tunnels is CT, which enables very good spatial resolution of bone and multiplanar reconstructions in the direction of the tunnels (Purnell et al. 2008, Hantes et al. 2010). If MRI is used to evaluate the postoperative knee, the bone tunnels are visible and it is possible to assess tunnel communication in axial, sagittal, and coronal T1-weighted images without the effective dose of CT (Siebold and Cafaltzis 2010).

4.9.11 Evaluation of native ACL insertion sites and graft tunnel location With the concept of anatomical ACL reconstruction, it has become necessary to measure the size of the ACL insertion of the individual patient in order to reproduce the anatomy with surgery (van Eck et al. 2010a, Yasuda et al. 2011). The size of the tibial insertion can be measured intraoperatively or in preoperative planning with 2D MRI, and the size of the femoral insertion can be measured with

3D MRI using either multiplanar reconstruction (MPR) or a special software (van Eck et al. 2010a, Han et al. 2012, Araujo et al. 2013). The AP tibial insertion site diameter of less than 14 mm favors considering anatomical SB reconstruction (van Eck et al. 2010a). Similarly, the width of the intercondylar notch less than 12 mm favors anatomical SB reconstruction (van Eck et al. 2010a).

In anatomical ACL reconstruction, the graft tunnels are aimed at anatomical locations (van Eck et al. 2010a). The easiest way to postoperatively evaluate the tunnel locations with MRI is to measure the inclination angle of the graft, which should resemble the native ACL inclination angle of 43^o to 57^o if the graft is in anatomical position (Illingworth et al. 2011, Araujo et al. 2013).

Experimentally, the anatomical ACL insertions have been described first with anatomical dissection and then with radio-opaque markers or drilling combined with radiography or CT (Bernard et al. 1997, Yamamoto et al. 2004, Zantop et al.

2008b, Lorenz et al. 2009, Forsythe et al. 2010). The commonly used quadrant method describes the locations of the anatomical insertions as percentages in a grid model projected on a sagittal image of the lateral femoral condyle and the sagittal and coronal images of the tibial plateau (Staubli and Rauschning 1994, Bernard et al. 1997, Lorenz et al. 2009). An alternative for describing the femoral insertions in a way similar to the arthroscopic view of the lateral femoral condyle is the anatomic coordinate axes method (Forsythe et al. 2010). The quadrant method was developed to make use of radiographs, but recently both the quadrant method and the anatomic coordinate axes method have been applied to 3D CT models in clinical studies (Hensler et al. 2013).

An interesting method to compare the patient’s individual native ACL insertion site location to tunnel location with 3D MRI models has been recently described (Abebe et al. 2009). In this method, both knees of patients are scanned with thin-slice 2D MRI and also with 3D MRI, and the image data is segmented to produce 3D models of both knees. The native ACL insertions of the intact knee are mirrored to the 3D model of the affected knee and the tunnel locations directly compared to overlaying native ligament insertions (Abebe et al. 2009). With this method, drilling of the femoral tunnel via accessory anteromedial portal has been shown to produce more anatomical tunnel locations when compared to transtibial femoral tunnel drilling (Abebe et al. 2009, Bowers et al. 2011).

In a recent publication, a simplistic method to measure femoral tunnel locations using a conventional sagittal 2D MRI sequence was illustrated (Noh et al. 2013). In that study, the distance from the most cranial point of the intercondylar notch to the tunnel aperture was measured. The researchers compared the location of

femoral tunnels drilled either transtibially or using the anteromedial portal.

Although comparison of these 2 groups was possible, it is impossible to tell if the tunnel locations were anatomical with their method. Currently, our knowledge of ACL insertion site anatomy is based on coordinate methods in the literature such as the quadrant method or the coordinate axes method (Lorenz et al. 2009, Forsythe et al. 2010).

4.9.12 Degenerative cartilage changes after ACL reconstruction

The studies on long-term osteoarthritis after ACL reconstruction have used radiographic evaluation of the knee, most often Kellgren-Lawrence or Altman scales (Kellgren and Lawrence 1957, Altman et al. 1995, Claes et al. 2013, Ajuied et al. 2013). Although recommended by the World Health Organisation as a reference standard for scientific studies, radiographic evaluation is an indirect measurement of only a small proportion of the articular cartilage (Brandt et al. 1991, Hayes et al.

2005).

With arthroscopic evaluation of the articular cartilage as a gold standard, MRI has evolved to be a powerful tool with high sensitivity and specificity for detecting chondral injuries (Potter et al. 1998, Brittberg and Winalski 2003). An MRI modification of the Outerbridge system is most often used with cartilage-sensitive PD-weighted and T2-weighted MR images (Brittberg and Winalski 2003, Potter et al. 2012). The bone-marrow edema associated with primary traumatic chodral injury most often seen in the lateral tibial plateau and the lateral femoral condyle in ACL injury is associated with increased cartilage loss in long-term follow-up (Potter et al. 2012).

4.9.13 Hardware failure

Complications to graft fixation devices such as bioabsorbable interference screws, metallic screws, buttons and pins include hardware fragmentation, loosening, and displacement (Casagranda et al. 2009). Metallic devices cause more artifacts to MRI than bioabsorbable devices, which in turn are easier to evaluate with MRI (Moisala et al. 2008).

4.9.14 Infection

The incidence of septic arthritis after ACL reconstruction is <0.5%

(Papakonstantinou et al. 2003). The clinical diagnosis in the early postoperative phase may be difficult. Symptoms of mild local pain, effusion, and increased C-reactive protein level are nonspecific and aspiration is necessary for the diagnosis (Schollin-Borg et al. 2003, Papakonstantinou et al. 2003). MRI findings of infection are joint effusion, synovitis, bone erosions, edema of adjacent soft tissues and bone marrow, soft tissue abscess, and sinus tract formation from bone to skin (Papakonstantinou et al. 2003). In the case of biobsorbable screws, an important differential diagnosis is foreign body reaction with local edema and tunnel cyst formation (Cox et al. 2014).