8.1 Developing surgical techniques and the role of MRI
The ACL and treatment of its injuries are under extensive scientific research. A search from PubMed with “anterior cruciate ligament” yields over 13000 articles and with “anterior cruciate ligament reconstruction” over 6000 articles with the number of new articles increasing every year. The field of ACL surgery is developing fast with interest towards restoring the native ACL anatomy (Arnold et al. 2013). First, the DB ACL reconstruction challenged the conventional ACL reconstruction techniques (Hussein et al. 2012b, Suomalainen et al. 2012).
Recently, the concept of individualized anatomical ACL reconstruction using either SB or DB grafts, depending on size of the ligament insertions, has been introduced with good short-term results (van Eck et al. 2010a, Hussein et al. 2012a). It has been shown that ACL reconstruction restores knee stability in mid-term enabling return to sports, but the long-term results of reducing the development of osteoarthritis are currently insufficient (Ardern et al. 2011, Claes et al. 2013, Ajuied et al. 2013, Frobell et al. 2013). The need for more accurate and objective outcome measurements to show the potential benefits of new surgical approaches has been recognized (Karlsson et al. 2011).
MRI is the best imaging method for evaluating ACL reconstruction and has been used to investigate complications in operated patients with symptoms or new trauma (Collins et al. 2008, Bencardino et al. 2009). The majority of radiologic literature on ACL reconstruction MRI has been published before DB ACL reconstruction became widely noticed or is confined to MRI findings of conventional ACL reconstruction (Recht and Kramer 2002, Papakonstantinou et al. 2003, Saupe et al. 2008, Collins et al. 2008, Bining et al. 2009, Bencardino et al.
2009, Abebe et al. 2009, Bowers et al. 2011, Potter et al. 2012).
8.2 Graft SI
The first reports on DB ACL reconstruction MRI noted that graft ligamentization behavior and findings of complications are similar to those found in conventional SB ACL reconstruction, on which our knowledge has been based (Casagranda et al.
2009, Poellinger et al. 2009). The concept of graft SI changes representing some process that weakens the graft is derived from early conventional ACL reconstruction graft impingement research (Howell et al. 1991b). Already in 2007, Sonoda et al. reported that increased SI of the AM graft was associated with anterior laxity and increased SI of the PL graft with rotational laxity in DB ACL reconstruction MRI (Sonoda et al. 2007). Their findings were in concordance with the understanding of the biomechanical roles of the bundles of the native ACL at that time (Chhabra et al. 2006). However, in their study the MRI and clinical examination were done 1 year after operation when the grafts are still in the ligamentization period when increased SI is very frequently seen (Recht and Kramer 2002). Moreover, it has been shown that graft SI may remain increased for up to 12 years after conventional SB reconstruction with no association with knee laxity (Saupe et al. 2008). In our DB ACL reconstruction material in Study IV, the increased graft SI was not associated with knee laxity in 2-year follow-up. This is in contrast with the previous short-term results (Sonoda et al. 2007). A new finding in study IV was that increased AM graft SI was associated with reduced knee flexion range. A possible explanation for this is that increased graft SI may represent some process that causes the stiffening of the AM graft. However, because there was no difference in other clinical evaluations, the significance of reduced knee flexion remains uncertain.
In a recent article, it was noted that for a currently unknown reason there is more increased graft SI for a longer period of time postoperatively in anatomical than in conventional ACL reconstruction, a similar finding to the one in our Study I (Farshad-Amacker and Potter 2013). Our hypothesis is that the oblique course of the PL graft may affect its signal properties in orthogonal sagittal and coronal images resulting in 4 times more frequent finding of increased SI than in AM graft (I). This cannot be contributed to the MA effect because in the supplementary study the PL grafts were more vertically oriented than the AM grafts, and because the increased SI was also analyzed using T2-weighted images unaffected by the MA effect (Fullerton and Rahal 2007). Interestingly, stretching of a tendon alters the MRI signal because water moves out of the tendon (Puxkandl et al. 2002, Fullerton and Rahal 2007) and the PL bundle of the ACL is stretched in the imaging
position, the knee unloaded and extended (Amis and Dawkins 1991). Accordingly, if the PL graft is located anatomically it may show increased MR signal because of its biomechanics-induced reduced hydration.
It is noteworthy that although increased SI was associated with partial graft tears that are prone to volume-averaging artifacts, the majority of cases with increased SI were seen in intact grafts (I). Accordingly, increased graft SI can be thought of as a common postoperative DB ACL MRI finding with no clear clinical relevance at this moment.
8.3 Graft visibility
In study III, the grafts were graded visible when intact graft fibers were seen, partially visible when few intact fibers were seen, and invisible when no graft fibers were seen. We chose this grading because it has been used in previous surgical publications (Suomalainen et al. 2011). When graft analysis is based on the presence or absence of visible, intact graft fibers, the grading is similar to the graft disruption analysis in study I and in the radiologic literature (Recht and Kramer 2002, Papakonstantinou et al. 2003). Graft visibility should not be confused with the finding of increased graft SI in which the graft fibers (intact graft or a thin bundle in partial rupture) are always visible but have more signal, i.e. they are more bright compared to the PCL (Howell et al. 1991b). Naturally, it is impossible to evaluate the SI of an invisible graft that cannot be seen.
We found in study III that more anterior location of the both tibial tunnels was associated with more partially visible and invisible grafts, which is the same as more partial ruptures and graft disruptions. A possible reason for this finding may be subtle impingement of the ruptured AM grafts to the roof of the intercondylar notch. However, this is not the case with the PL grafts because their course from the femoral to the tibial tunnels runs far more posteriorly from the notch roof. It is known that nonanatomical tunnel locations are a major cause of graft failure, which is a clinical term describing an unstable knee after ACL reconstruction and not an MRI finding (Giffin and Harner 2001, MARS Group et al. 2010). In our study III all patients with abnormal AP or rotational laxity had visible grafts and the findings of partial visibility or invisibility were not associated with knee laxity.
Furthermore, the measured tunnel locations were close to anatomical in studies II and III. The reason why more PL grafts were partially visible remains unknown. It is important to correlate the MRI findings of graft invisibility with the clinical
evaluation of knee stability to prevent overdiagnostics in postoperative evaluation of DB ACL reconstruction.
8.4 Graft evaluation in clinical work
After ACL reconstruction, 10-30% of patients have nonspecific symptoms of knee pain or laxity and may need further diagnostic workup with MRI (Yunes et al.
2001). In clinical radiology, the MRI finding of increased graft SI may be confusing in graft evaluation because it has previously been ascribed as a pathologic finding indicative of possible graft impingement, degeneration, and partial tearing (Howell et al. 1991a, Hodler et al. 1992).
The results of the supplementary study show that the MA effect is responsible for some of the increased graft SI seen in T1-weighted and PD-weighted images (AM graft angle to B0 approximately 38o (SD 6.3) and the PL graft angle 33o (SD 10.0) in cases with finding of increased SI). The MA effect starts from 20o to B0
with the maximum in 55o (Fullerton and Rahal 2007). Nevertheless, the results of increased graft SI in studies I and IV are reliable because the MA effect does not affect the T2-weighted images with TE 78 ms (>37 ms limit of MA) (Peh and Chan 1998, Fullerton and Rahal 2007). The current knowledge on the MA effect is focused on native tendons and the effects of graft preparation and healing on the MA effect are not known, but presumably the MA effect should be weaker if the highly organized collagen structure is damaged as in ACL reconstruction grafts (Fullerton and Rahal 2007).
Based on the current literature and the results of studies I and IV, the graft SI changes are often seen and probably not clinically significant. While PD-weighted images can be used in graft evaluation, some MA effect is expected and even more so in T1-weighted images. T2-weighted images are not affected by the MA effect and no information is lost if graft evaluation would be based only on them. This is important in clinical practice with time restrictions. Moreover, graft evaluation should primarly be aimed to diagnose graft ruptures by analyzing graft morphology (Saupe et al. 2008, Casagranda et al. 2009).
8.5 Tunnel enlargement and communication
Tunnel communication is a possible complication seen only in DB ACL reconstruction. Its etiology includes errors in drilling and postoperative tunnel enlargement (Siebold and Cafaltzis 2010). In surgery, the tunnels are drilled close to each other at the native ligament footprint leaving only a 1-2 mm bony wall between the tunnels (Järvelä 2007). Outward from the intra-articular surface, the tunnels diverge making the area close to the joint more susceptible to tunnel communication. The magnitude of postoperative tunnel enlargement in DB ACL reconstruction studies (tunnel enlargement range from 20% to 48% in the literature and 53% to 59% in study II) makes tunnel communication a likely phenomenon.
Tunnel communication has, however, been reported surprisingly seldom in the literature (Siebold 2007, Siebold and Cafaltzis 2010).
In study II, tunnel communication was seen in 7 patients (11%) in the femur and in 19 patients (29%) in the tibia. Tunnel communication was associated with tunnel enlargement. It is easy to understand that the 1-2 mm bony wall between the tunnels disappears with the 3-4 mm enlargement of both tunnels. Another possible etiology for tunnel communication might be compromized blood supply to the thin wall of cancellous bone and subsequent avascular necrosis (Kristensen et al. 2013).
In most studies, tunnel enlargement has not been associated with knee laxity (Samuelsson et al. 2009, Kawaguchi et al. 2011). On the contrary, tunnel communication has been supposed to spoil knee stability and graft function after DB ACL reconstruction, although there is no supporting data in the literature (Christel et al. 2008, Pombo et al. 2008, Hantes et al. 2010). In study IV, we demonstrated for the first time that tunnel communication is not associated with knee laxity in mid-term follow-up. We also described an MRI finding of an intermediate-signal intensity substance resembling fibrous tissue that was visible between the grafts and the bony walls of the enlarged tunnels. We suspect that this tissue may stabilize the grafts in cases of tunnel communication, but further study with histologic correlation is warranted.
Tunnel communication in the tibia was associated with increased flexion range of motion in study IV. This finding is at odds to our theory of stabilizing fibrous tissue inside the tunnels. However, as there was no statistical difference in other clinical evaluation results, the significance of this finding remains unclear.
Correct diagnosis of tunnel communication is important since it may cause difficulties in revision ACL surgery. If there is not enough bone left for drilling at
the anatomical tunnel insertions, a staged operation with bone grafting may be necessary (Wilson et al. 2004).
8.6 Tunnel location measurement in clinical work
The most important reasons for graft failure after operation are new trauma and a misplaced graft leading to non-bearable biomechanics (Giffin and Harner 2001, MARS Group et al. 2010). The main principle of individualized anatomical ACL reconstruction is to surgically reproduce the native ACL anatomy as closely as possible, including the size and the location of anatomical insertions (van Eck et al.
2010a, Hofbauer et al. 2013). This outcome of surgery can be most effectively measured with MRI. Previous measurement methods have been radiography and CT. These modalities have ionizing radiation issues and have considerably lower soft-tissue contrast compared to MRI. Other scientific MRI graft location evaluation methods make use of computerized 3D models of the insertions with mirrored overlay of the anatomical insertions of the contralateral knee (Abebe et al.
2009, Bowers et al. 2011). Although providing detailed information, these procedures are time-consuming and require special software. Thus, they are not suitable for clinical imaging workflow as such.
We found that the modified quadrant method is applicable to routine 2D sagittal and coronal MR images with reasonably good reproducibility, as estimated in inter-observer agreements of measurements (II). The measured tunnel locations (II) were very close to the range of anatomical insertion site locations in the literature (Yamamoto et al. 2004, Colombet et al. 2006, Edwards et al. 2008, Tsukada et al. 2008, Zantop et al. 2008a, Lorenz et al. 2009).
As the femoral tunnel locations proved more difficult to measure, we propose a simple method for clinical MR imaging work to determine if the grafts and the tunnels are in the anatomical locations. First, the AP location of the tibial tunnels is measured from the anterior margin of the tibial plateau and divided by the AP depth of the plateau using sagittal images (Lorenz et al. 2009). The anatomical location of the AM tunnel is at 36-41% (weighted mean 36%) of the tibial depth and the anatomical location of the PL tunnel is at 47-52% (weighted mean 51%), respectively (Colombet et al. 2006, Tsukada et al. 2008, Zantop et al. 2008a, Lorenz et al. 2009). Then, the inclination angle of the AM graft is measured. It is the angle between the graft and the tibial horizontal line, which runs perpendicular to the long axis of the tibia (Illingworth et al. 2011). The anatomical inclination of the
ACL ranges from 43o to 57o and this can be used as the anatomical range of the graft angle, too (Illingworth et al. 2011, Araujo et al. 2013). Finally, the AM tunnel aperture should be located proximally and the PL tunnel aperture distally at the lateral femoral condyle in the extended knee similarly to the anatomical attachments of the native ACL (Chhabra et al. 2006, van Eck et al. 2010a). This proposed simple method makes use of 3 diameter measurements and 1 angle measurement from a single sagittal imaging sequence on a standard radiologic workstation (Figure 21).
Figure 21. The proposed simple method for measuring the anatomical location of the grafts and the tunnels. (a) The location of the AM tunnel is at 37% and the PL tunnel at 57% of the AP depth of the tibial plateau (anatomical ranges 36-41% and 47-52%, respectively). The AM graft (black arrow) inclination angle is 43o (anatomical range 43o-57o) The PL graft is marked with a white arrow. (b) The femoral tunnel apertures are located at the anatomical configuration: the AM tunnel (black arrow) proximally and the PL tunnel (white arrow) distally in the extended knee (sag PD MRI).
8.7 Limitations of the study and future considerations
Some important issues should be considered when assessing the results of this study. First, the MRI was done from 2006 to 2009 with a 1.5 T scanner using the routine clinical sequences of that time. This is also an advantage because similar scanners are still in use today and are the most commonly used worldwide.
Therefore, our results are easily applied to clinical radiologic work. Our oblique sagittal and coronal images were only parallel to the AM graft and the PL graft evaluation was done using 4 mm orthogonal slices. The PL graft courses obliquely in orthogonal sequences making the graft SI evaluation more difficult and possibly resulting in volume-averaging artifacts between the graft and the intermediate signal around it (I, IV). Secondly, re-arthroscopies, which would have allowed surgical confirmation of the MRI findings, were not indicated in stable knees (I-IV). Thirdly, femoral tunnel locations and tunnel communication would have been more reliably measured using sub-millimeter 3D MPR sequences with better spatial resolution (II, III, IV). Fourthly, we cannot rule out drilling errors causing iatrogenic tunnel communication without immediate postoperative imaging (II, III, IV). Finally, there was more than a 3-month time interval between the imaging (mean 22 months postoperatively, range 16-29 months) and the clinical evaluations (mean 25 months postoperatively, range 24-33 months) making it possible that the clinical findings were better because the knees were allowed more time to heal (III, IV). In addition, the resulted wide time frames of both the MRI and the clinical evaluation make the data more heterogenous.
The 2-year follow-up time of the study is too short to predict the long-term outcome of DB ACL reconstruction. We will continue the clinical and the imaging follow-up. The future prospects for imaging would include the use of 3T MRI with better contrast, spatial resolution and acquisition time, and also obtaining oblique sequences separately along both grafts. A thin-slice 3D MPR sequence would make tunnel evaluation more reliable. Most importantly, the development of osteoarthritic changes should be investigated with MRI, possibly in conjunction with advanced quantitative MRI sequences of articular cartilage, to predict the long-term outcome of the patients (Li and Majumdar 2013).