Abstract
Background:
Magnetic resonance imaging (MRI) showing an “intact” anterior cruciate ligament (ACL) graft may not correlate well with examination findings. Reasons for an ACL graft dysfunction may be from malpositioned tunnels, deficiency of secondary stabilizers, repeat injuries, or a combination of factors.
Purpose:
To evaluate the concordance/discordance of an ACL graft assessment between an arthroscopic evaluation, physical examination, and MRI and secondarily to evaluate the contributing variables to discordance.
Study Design:
Case series; Level of evidence, 4.
Methods:
A total of 50 ACL revisions in 48 patients were retrospectively reviewed. The ACL graft status was recorded separately based on Lachman and pivot-shift test data, arthroscopic findings from operative reports, and MRI evaluation and was categorized into 3 groups: intact, partial tear, or complete tear. Two independent evaluators reviewed all of the preoperative radiographs and MRI scans, and interrater and intrarater reliability were evaluated. Concordance and discordance between a physical examination, arthroscopic evaluation, and MRI evaluation of the ACL graft were calculated. Graft position and type, mechanical axis, collateral ligament injuries, chondral and meniscal injuries, and mechanism of injury were evaluated as possible contributing factors using univariate and multivariate analyses. Sensitivity and specificity of MRI to detect a torn ACL graft and meniscal and chondral injuries on arthroscopic evaluation were calculated.
Results:
The interobserver and intraobserver reliability for the MRI evaluation of the ACL graft were moderate, with combined κ values of .41 and .49, respectively. The femoral tunnel position was vertical in 88% and anterior in 46%. On MRI, the ACL graft was read as intact in 24%; however, no graft was intact on arthroscopic evaluation or physical examination. The greatest discordance was between the physical examination and MRI, with a rate of 52%. An insidious-onset mechanism of injury was significantly associated with discordance between MRI and arthroscopic evaluation of the ACL (P = .0003) and specifically with an intact ACL graft on MRI (P = .0014). The sensitivity and specificity of MRI to detect an ACL graft tear were 60% and 87%, respectively.
Conclusion:
Caution should be used when evaluating a failed ACL graft with MRI, especially in the absence of an acute mechanism of injury, as it may be unreliable and inconsistent.
Anterior cruciate ligament (ACL) surgery is a frequent orthopaedic procedure with over 100,000 ACL reconstructions performed annually, which with the ever increasing active population continues to grow.6,7,19 Although the procedure is highly successful, the rate of revision ACL reconstruction remains at approximately 10%.7,14,19,20 Magnetic resonance imaging (MRI) is commonly used as a tool to assess the ACL graft in a revision setting and secondarily to evaluate the knee for other intra-articular injuries. In primary ACL injuries, MRI has been shown to be similar in diagnostic efficacy to a physical examination; however, in the setting of a reinjury or a poorly functioning graft, the reconstructed ACL is difficult to assess on MRI. 6 Multiple studies have demonstrated abnormal MRI findings in poorly functioning ACL grafts to include increased signal within the graft, concomitant chondral and meniscal injuries, and abnormal tunnel positions.6,8,23-25,28 An intact ACL graft on MRI may not correlate well with the actual function of the graft, especially in a patient with the chief complaint of instability.8,25,28 The radiographic evaluation of tunnel placement in failed ACL reconstruction has been well documented7,8,13,16,17,26; however, tunnel malposition has not been correlated with the integrity of the ACL graft assessed on MRI or with physical examination and arthroscopic findings in a revision ACL reconstruction setting.
The purpose of this study was to review all revision ACL reconstruction procedures conducted at a single institution to evaluate the concordance or discordance of the ACL graft assessment between an arthroscopic evaluation, physical examination, and MRI. Secondarily, it was to evaluate the possible contributing variables to discordance including graft type, mechanism of injury, tunnel positions, and concomitant intra-articular lesions.
Materials and Methods
The operative records of all surgical procedures performed at our institution between January 1, 2005, and August 16, 2011, were reviewed, revealing a total of 813 ACL reconstruction procedures. The search was narrowed to the 5 attending surgeons who perform revision ACL reconstruction, revealing 460 ACL reconstruction procedures. Individual surgeon case logs, operative reports, and medical records were then reviewed to determine which ACL reconstructions were revision procedures, in which 64 revisions in 60 patients were identified.
The inclusion criteria were patients with a minimum age of 18 years who underwent revision ACL reconstruction at our institution with preoperative MRI scans and plain radiographs, an operative report describing the arthroscopic findings, and a preoperative history and physical examination. Patients who underwent more than 1 revision were also included if they had a second set of complete records and new MRI scans before their subsequent revision. Exclusion criteria were incomplete records, lack of prerevision MRI scans, revision due to a deep infection requiring graft debridement, and primary extra-articular procedures. Once applied, there were a total of 10 patients excluded because of the lack of an MRI study or incomplete medical records. Three additional revisions were excluded because of infections necessitating graft debridement as the primary cause of revision, and 1 was excluded because the initial ACL procedure was an extra-articular procedure. This resulted in a study group of 50 revisions in 48 patients (Figure 1). Two patients had undergone 2 re-revision procedures at our institution during the study time period and were included.
Flow diagram of the study.
All patients underwent MRI at the same institution with a 1.5-T superconducting magnet using a standard knee evaluation protocol without contrast on a dedicated knee coil. Imaging was as follows: coronal T1-weighted and T2-weighted fast spin echo (FSE) with fat-suppressed (FS) sequences; sagittal T1-weighted or proton density and T2-weighted FSE FS sequences; and axial T2-weighted FSE FS sequences.
Demographic data were collected to include age at the time of revision, time between primary reconstruction and revision, time between MRI and revision, sex, extremity, height, weight, body mass index (calculated), and active duty status. Electronic medical records and inpatient medical records to include operative reports were reviewed to identify the prior ACL graft type (patellar tendon, hamstring, or allograft), femoral fixation (interference, cortical button, or cross pin), and tibial fixation (interference, post/suspension, or interference and post). The chief complaint in the form of instability, pain, or stiffness was recorded along with the mechanism of injury classified as traumatic, twisting/hyperextension, or insidious. In addition, any superficial infections not necessitating graft debridement were recorded along with whether the revision procedure was staged.
Physical examination data were recorded based on the preoperative history and physical examination. If these data were not available, a documented examination under anesthesia was used from the operative reports. Range of motion; Lachman test for anterior translation and endpoint (1 = 0-5 mm, 2 = 6-10 mm, 3 = >10 mm; a = solid endpoint, b = soft endpoint); pivot-shift test for rotational stability (0 = normal, 1 = glide, 2 = shift); and medial collateral ligament, lateral collateral ligament, and posterior cruciate ligament status were recorded. KT-1000 arthrometer data were recorded when available. The ACL graft was classified into the categories of intact, lax, or torn based on the physical examination data. A Lachman grade 1a or 2a with a grade 0 pivot shift was classified as intact. A Lachman grade 2b or 3b with a grade 1 or 2 pivot shift was classified as torn. Any other examination combination was classified as lax.
The operative reports were reviewed to evaluate the actual arthroscopic ACL status as either intact, partial tear, or torn. Meniscal tears and cartilage injuries were classified by location only (patellofemoral joint, medial or lateral compartments) and not by severity because of the high variation of details described in the operative reports. The arthroscopic findings served as the gold standard for comparison of all intra-articular findings.
The pre-revision/post–primary reconstruction MRI scans and plain radiographs of all patients were reviewed and recorded by 2 independent evaluators twice for a total of 4 evaluations using OsiriX digital imaging software (Geneva, Switzerland). The status of the ACL was recorded as intact, partial tear, or torn. Meniscal tears, collateral ligament injuries, and cartilage injuries were also assessed. Intraobserver and interobserver reliability were calculated using the 4 evaluations. The independent MRI evaluations were then combined, and a consensus on the ACL status was obtained based on the most frequently assessed status of the 4 observations. Any discrepancies were resolved using a third evaluator (senior author, M.T.P.). The evaluators were free to independently scroll through coronal, sagittal, and axial planes of all sequences using the same version of OsiriX software and were blinded to the official radiologist reports, clinical and physical examination data, and arthroscopic findings. Two of the evaluators were fellowship-trained (sports) orthopaedic surgeons with 8 and 10 years of postfellowship experience, and the other was a fourth year orthopaedic resident at the time of the study.
Measurements of the ACL graft on MRI were taken to include the femoral tunnel position, tibial tunnel position, and graft angle. The femoral graft position was measured based on the quadrant method previously described.5,11,26 On sagittal images, the distance of the center of the graft from the posterior cortex measured along the roof of the intercondylar notch (Blumensaat line) was calculated by measuring the posterior and anterior margins of the graft and dividing by 2 (Figure 2). This was then converted to a percentage by dividing the graft center position by the total width of the notch. A graft positioned in the posterior 25% quadrant is considered ideal. 26 The tibial graft position was measured using the method described by Frank et al 12 and was similar to previously published methods. 2 With use of the anterior cortex of the tibia as a reference, the anterior and posterior margins of the graft were measured and divided by 2 to determine the center of the graft position (Figure 3). The total width of the tibia was measured from the anterior cortex to the posterior cruciate ligament insertion. The graft center position was then converted to a percentage of the total tibial width. The anatomic position is between 41% and 50% of the tibial width. 12 The femoral graft angle was measured on coronal images by the angle subtended by a line along the inferior cortex of the medial and lateral femoral condyles to a line from the center of the graft when it first appears in the femoral notch to the center of the graft in the tibia (Figure 4).16,17 All 4 measurements (2 for each evaluator) were then averaged and converted to categorical variables for analysis.
Femoral graft position measurement on sagittal T1-weighted magnetic resonance imaging sequence. The posterior cortex is used as the zero reference, and measurements are taken along the roof of the intercondylar notch (Blumensaat line). The total width of the roof is measured from the posterior to anterior cortex. The posterior and anterior graft margins are measured, summed, and divided by 2 to obtain the center of the graft and then converted to a percentage of the posterior to anterior notch roof distance.
Tibial graft position measurement on sagittal T1-weighted magnetic resonance imaging sequence. The anterior cortex is marked as a reference, and the distance to the anterior and posterior graft margins is measured as well as the distance from the anterior cortex to the posterior aspect of the posterior cruciate ligament insertion. The center of the graft is calculated and then converted to a percentage from anterior to posterior on the tibia.
Femoral graft angle on coronal T1-weighted magnetic resonance imaging sequence. A line is drawn along the inferior aspect of the femoral condyles at the cortical surface as a reference. The femoral graft angle is measured from the center of the graft exiting the tibia to the center of the femoral tunnel.
Plain radiographs were used to determine any degenerative changes using the Kellgren-Lawrence scale. 18 The mechanical axis of both the normal and pathological sides was measured on hip-to-ankle standing radiographs when available using a line drawn from the center of the femoral head to the center of the tibial plafond, with the medial edge of the tibial plateau serving as the 0% point and the lateral edge as the 100% point.4,22
Analysis was conducted to determine the concordance and discordance between a physical examination, arthroscopic evaluation, and MRI evaluation of the ACL graft at the time of revision surgery. For analytical purposes, a lax graft on examination was equivalent to a partial tear on MRI or arthroscopic evaluation. We secondarily reviewed the mechanism of injury (insidious, traumatic, or twisting), tunnel positions and angles, collateral ligament injuries, chondral injuries, meniscal tears, and incidence of radiographic osteoarthritis as possible factors for discordance. The Fisher exact test was used to determine significance when comparing categorical variables, and the Student t test was used when comparing categorical to continuous variables. The P value was set at <.05, and means ± standard deviations were calculated for continuous variables. Sensitivity and specificity were calculated comparing variables assessed on MRI to the arthroscopic evaluation, with the arthroscopic evaluation used as the gold standard. This study was approved by our institutional review board.
Results
A total of 48 patients who underwent 50 revision ACL reconstructions made up the study group (45 male, 3 female). The mean age at the time of revision reconstruction was 29.1 ± 5.9 years, and the mean time from primary reconstruction to revision was 4.6 ± 3.6 years (range, 6.7 months to 13.8 years). The mean time interval between MRI and arthroscopic examination was 6.4 ± 5.9 months (range, 0.53-38.1 months). The mechanism of injury was insidious without a discernible event in 48% of revisions, and the chief complaint was instability in 92%. No patients had flexion contractures or loss of range of motion. The initial ACL graft was a patellar tendon in 26%, hamstring tendon in 42%, and allograft in 32%. Superficial infections were present in 3 patients who did not require graft debridement. Staged revision reconstruction was performed in 26% of patients. See Table 1 for complete demographic data.
Summary of Demographic Data a
ACL, anterior cruciate ligament; BMI, body mass index; MRI, magnetic resonance imaging.
The ACL status on physical examination was determined to be torn in 82% and lax in 18%, with 78% of patients demonstrating a Lachman grade 2b and grade 2 pivot shift (Table 2). Four patients demonstrated a grade III posterolateral corner (PLC) injury. KT-1000 arthrometer data were only available on 4 patients and were therefore not used for analysis.
Summary of Physical Examination Findings (Categorical Data) a
ACL, anterior cruciate ligament; MCL, medial collateral ligament; PLC, posterolateral corner.
Determined combining Lachman and pivot-shift tests.
On arthroscopic evaluation, the ACL graft was torn in 70% and partially torn in 30%. No patient had an intact ACL. Meniscal tears were present in 66% of patients, with 50% medial tears, 32% lateral tears, and 16% with both medial and lateral tears. The rate of chondral injuries was 70% overall. Cartilage injuries of the medial compartment were present in 52% of patients. The interobserver and intraobserver reliability for MRI evaluation of the ACL graft were moderate, with combined κ values of .41 and .49, respectively.
When the MRI data were combined, the ACL graft was read as intact in 24%, partially torn in 30%, and torn in 46% of patients (Figure 5). The graft was in the posterior quadrant of the femur in 54% and in the anatomic central (41%-50%) position of the tibia in 50%. The femoral tunnel position was vertical (between 11 and 12 o’;clock or 12 and 1 o’clock) in 88% (Table 3).
This chart compares anterior cruciate ligament graft integrity on the 3 assessment modalities. No grafts were found to be intact on physical examination or arthroscopic evaluation; however, 24% of grafts were read as intact on magnetic resonance imaging.
Combined MRI Data a
ACL, anterior cruciate ligament; MRI, magnetic resonance imaging.
The ACL status was determined by the majority of readings by the authors.
Graft positions were determined by the mean measurements between authors and then categorized for analysis.
On radiographic evaluation, 74% of patients had some form of arthritic changes, with 66% demonstrating Kellgren-Lawrence grade 1 or 2 osteoarthritis (Table 4). Mechanical axis data were available on 30 patients; the mean values were 42.3% ± 15.1% on the pathological side and 38.3% ± 14.1% on the unaffected side.
Degenerative Changes on Radiographs
0 = no osteoarthritis changes; 1 = doubtful joint space narrowing and possible osteophytes; 2 = definite osteophytes and definite joint space narrowing; 3 = moderate multiple osteophytes, definite joint space narrowing, sclerosis, and possible bone contour deformity; 4 = large osteophytes, marked joint space narrowing, severe sclerosis, and definite bone contour deformity.
When the status of the ACL on MRI was compared with physical examination and arthroscopic findings, the greatest discordance was between the physical examination and MRI with a rate of 52%, followed closely by the arthroscopic evaluation and MRI at 44%. The greatest concordance was between the physical examination and arthroscopic evaluation at 72% (Figure 6). The ACL graft was read as intact in 6 patients on MRI when it was found to be torn on arthroscopic evaluation. For comparison, on the radiologist-read MRI reports in the medical records, the ACL graft was read as intact in 9 patients who had torn grafts on arthroscopic examination.
Discordance rates of physical examination (PE), magnetic resonance imaging (MRI), and arthroscopic findings of the anterior cruciate ligament graft status with specific discordance readings between the 3 assessment modalities. The greatest discordance was between the physical examination and MRI; however, there was a 44% discordance rate between the actual graft status at the time of revision as assessed by the arthroscopic evaluation compared with the MRI assessment.
With the arthroscopic evaluation as the diagnostic standard, the sensitivity of MRI to diagnose an ACL graft tear was 60%, and specificity was 87%. For detecting medial meniscal tears, the sensitivity and specificity of MRI were 64% and 56%, respectively, and for lateral meniscal tears were 31% and 82%, respectively. Cartilage lesions overall were detected with a sensitivity of 41% and specificity of 91% (Table 5).
MRI Versus Arthroscopic Evaluation a
Arthroscopic evaluation was used as the gold standard for analysis. The magnetic resonance imaging (MRI) readings were based on combined findings of study reviewers. ACL, anterior cruciate ligament.
Bivariate analysis revealed that the mechanism of injury was significantly associated with concordance/discordance between MRI and arthroscopic evaluation of the ACL (P = .0003). No other variables were significant to account for the concordance/discordance discrepancy with the exception of femoral graft position, which trended toward significance between concordance/discordance of the physical examination and MRI evaluation (P = .057) (Table 6).
P Values for Comparison of Categorical Variables to Concordance/Discordance of ACL Graft Assessment Modalities a
Data expressed as P values (Fisher exact test) with significance at <.05. ACL, anterior cruciate ligament; MRI, magnetic resonance imaging; PLC, posterolateral corner.
When multivariate analysis was conducted, an insidious-onset mechanism of injury was associated with an intact ACL graft on MRI (P = .0014); however, no other variables were significantly related to this (Table 7). A PLC injury was associated with a torn ACL graft on arthroscopic evaluation at the time of revision (P = .0014). Unrelated to the ACL status, PLC incompetency on physical examination was significantly associated with an abnormal PLC on MRI evaluation (P = .0029).
P Values for Comparison of Variables to Mechanism of Injury a
Data expressed as P values with significance at <.05. The Fisher exact test was used for categorical variables, and the Student t test was used for continuous variables. ACL, anterior cruciate ligament; PLC, posterolateral corner.
When comparing the arthroscopic results of meniscal tears and cartilage injuries to the presence of degenerative changes on radiographs, the presence of both a meniscal tear and a chondral injury was significantly associated with the presence of, at minimum, grade 1 degenerative changes (P = .0419).
Discussion
The key finding in this study is a high discordance rate between the MRI evaluation of an ACL graft and the actual ACL graft status as determined by the arthroscopic evaluation at the time of revision ACL reconstruction. One of the most concerning findings was that the ACL graft was read as intact on MRI but was torn on arthroscopic evaluation in 6 cases based on the authors’ combined assessments of the ACL grafts and in 9 cases based on the radiology reports (Figure 7). This represents a potential diagnostic error when evaluating a patient for recurrent instability after ACL reconstruction if MRI is the solitary diagnostic tool and might be a factor in the delay of treatment in the revision ACL reconstruction setting. 19 In contrast, although the physical examination still had a discordance rate of 28% with the actual arthroscopic findings, the discordance was only between partial and complete tears. Every patient examined was determined to have an incompetent (lax or torn) ACL on physical examination before revision surgery.
Magnetic resonance imaging scans of (A) sagittal T1-weighted proton density and (B) T2-weighted fast spin echo fat-suppressed sequences demonstrating an intact-appearing anterior cruciate ligament (ACL) graft in a patient who had a torn ACL graft at revision reconstruction. He had insidious-onset instability and a Lachman grade 2b with a pivot glide. Technical error was likely the cause of failure because of a vertical femoral tunnel and posterior tibial graft placement.
Two previous retrospective studies used an arthroscopic evaluation to assess various MRI characteristics of primary ACL tears and ACL graft tears.8,28 The study by Van Dyck et al 28 classified ACL tears into the categories of stable and unstable (unstable = completely torn at the time of surgery) in patients who underwent arthroscopic examination and compared the MRI findings to a single category of combined clinical and arthroscopic findings; the sensitivity and specificity of MRI to detect an unstable tear were 59% and 81%, respectively. They misread 14 of 17 unstable tears in the chronic tear setting, citing scar tissue as a major limitation, and concluded that the extent of a primary ACL injury and subsequent treatment should be determined based on the clinical history, symptoms, and physical examination and not on the basis of imaging abnormalities. In a similar study, Collins et al 8 evaluated various primary and secondary MRI signs of ACL graft disruption and compared them with arthroscopic findings as the gold standard with a maximum 6-month interval between MRI and surgical evaluation. The sensitivity and specificity for MRI to detect a complete graft tear were 72% and 100%, respectively, with the authors citing complete discontinuity of graft fibers to be the most specific and predictive MRI characteristic. In our study, the sensitivity and specificity for MRI to detect a complete ACL graft tear were 60% and 87%, respectively, which were slightly lower than those in the Collins et al 8 study. In contrast, our study did not focus on specific MRI characteristics of the ACL graft for statistical analysis but on the analysis of possible subjective clinical factors and technical factors of the graft placement to attempt to explain the diagnostic error, of which we only found significance with an insidious-onset mechanism of injury. We also used concordance/discordance as our primary analysis tool rather than sensitivity, specificity, and accuracy to allow for comparison of 3 independent categories (MRI, physical examination, and arthroscopic evaluation) and subsequently univariate and multivariate analyses. We also were able to calculate interobserver and intraobserver reliability of the ACL graft assessment with MRI because we had 2 independent reviewers of the MRI scans who were blinded to the clinical history, physical examination, and arthroscopic data. To our knowledge, this is the first study to evaluate the reliability of an MRI assessment of an ACL graft. Our moderate κ values of .41 for interobserver reliability and .49 for intraobserver reliability demonstrate the difficulty and potential inaccuracies of an MRI assessment of an ACL graft.
Possible explanations for the difficulty of an ACL graft assessment have been previously highlighted and include scar tissue formation, graft elongation, collagen structure alteration from mechanical stress or instrumentation, degenerative changes, partial tears, neoligamentization, and increased vascularity within the first 18 to 24 months from reconstruction.8,25,28 Saupe et al 25 evaluated 47 patients at a mean of 80 months from ACL reconstruction with MRI of the surgically treated knee. They found increased T1 and T2 signals in 70% and 64% of grafts, respectively; however, they found no correlation with KT-1000 arthrometer laxity (compared with the contralateral side) or International Knee Documentation Committee function scores. This increased signal present in the majority of ACL grafts several years from primary reconstruction could be a potential cause for the difficulty in interpreting the ACL graft on MRI when assessing a patient for possible revision surgery. This may also account for the higher rate of partial tears read on MRI, which typically have heterogeneous signals throughout the graft, indicating incomplete revascularization.10,25,27,29
A second key finding of our study is that the insidious onset of knee instability after ACL reconstruction without a specific traumatic event or injury was significantly associated with an intact ACL graft on MRI as well as a high discordance rate of MRI compared with the gold standard of arthroscopic evaluation when assessing the ACL graft. The possible causes of atraumatic graft failure are technical error including graft malpositioning, unrecognized ligamentous injuries, and graft biological failure, of which biological failure is much less common and more poorly understood.3,4,6,7,9,14,20 The most common technical errors are anterior femoral and tibial tunnel positions and vertical femoral tunnel placement in the coronal plane. ∥ Nonanatomic graft placement has been shown to result in longer and more vertical grafts and, in turn, altered knee kinematics, possibly predisposing to graft failure. 1 Another possible explanation for an insidious onset of ACL graft failure and an intact-appearing ACL graft on MRI is the typical elongation pattern of ACL graft failure as opposed to more proximal ruptures in native ACLs. 10 Although the femoral graft was anterior in 46% and vertical (between 11 and 12 o’clock) in 88% of patients, and the tibial tunnel outside of the anatomic window (41%-50% of the tibial sagittal width) in 50%, we were unable to find a significance to explain the inaccurate MRI findings. We were also unable to find further variables associated specifically with the mechanism of injury.
Our study cohort demonstrates a high rate of chondral and meniscal injuries at 70% and 66%, respectively. These are similar to rates previously reported in the literature of revision ACL reconstruction, with the rate of chondral injuries ranging from 51% to 73% and meniscal tears from 45% to 77%.4,6,9,15,16,19,29 The Multicenter ACL Revision Study group reported combined meniscal and chondral injuries in 57%, which is similar to our rate of 48%. 29 We were also able to show that these combined injuries are significantly related to degenerative changes on radiographs. This is intuitive but has never been formally correlated in a revision ACL setting. The diagnosis of these injuries also may be difficult in the revision setting, as the sensitivity and specificity of MRI for the detection of medial and lateral meniscal tears were moderate at 64% and 56% for medial meniscal tears and 31% and 82% for lateral meniscal tears, respectively. Chondral lesions are often present on arthroscopic evaluation more than initially anticipated based on MRI, with a sensitivity of 41%; however, if seen on MRI, they are likely to be significant at the time of surgery, with a specificity of 91%.
There are a few limitations of this study. The first is that it is a retrospective study with no control group. A true assessment of the accuracy of MRI for evaluating ACL grafts would be to blindly compare patients with successful primary ACL reconstructions with this group that underwent revision reconstruction. This would allow for more powerful statistical analysis and, similar to prior studies, may identify subtle imaging characteristics to better differentiate a poorly functioning graft from a well-functioning graft. A second limitation is the MRI technique. Coronal oblique MRI sequences have been shown to significantly increase the diagnostic accuracy of ACL assessments in primary ACL injuries when compared with standard sagittal MRI sequences. 21 Future prospective studies evaluating ACL grafts should consider this technique. The third limitation is a moderately sized study group. A larger cohort would potentially highlight more statistically significant variables contributing to the discordance of MRI and the actual status of an ACL graft. A final limitation is the lack of KT-1000 arthrometer data to augment the physical examination data. Most ACL reconstruction studies include this; however, our records contained insufficient data to use the KT-1000 arthrometer readings, again highlighting the retrospective limitations of this study.
In conclusion, caution should be used when evaluating a failed ACL graft with MRI alone, especially in the absence of an acute mechanism of injury, as it may be unreliable and inconsistent. Magnetic resonance imaging should be used as an adjunctive assessment tool with more emphasis placed on the subjective complaint, mechanism of injury, and physical examination when assessing a patient for possible revision ACL reconstruction. Careful attention should also be directed toward possible associated injuries such as meniscal tears and cartilage injuries, as they may not be fully detected on preoperative imaging.
Footnotes
Acknowledgements
The authors thank Dr Maryam Navaie, Dr Susan Eskridge, and Emily Chavez from Advanced Health Solutions (La Jolla, California, USA) for their assistance with study organization and statistical plans. They also thank Dr Robert Riffenburgh at Naval Medical Center San Diego for his assistance with statistical analysis.
The views expressed in this article are those of the authors and do not necessarily reflect the official policy or position of the Department of the Navy, the Department of Defense, or the US Government
One or more of the authors has declared the following potential conflict of interest or source of funding: D.J.S. has received fees for speaking from Arthrex and Pacific Medical. M.T.P. has received royalties or fees as a consultant from Arthrex and Joint Restoration Foundation; and has received funds for research from the AOSSM Young Investigators Grant (2005), Orthopedic Research and Education Foundation (2002, 2004, 2011), and AANA (2006, 2008). Indirect support was provided, in kind, by Advance Health Solutions LLC, including the donation of research services, computer equipment, and facilities.
