点击显示 收起
1 From the Dept of Medical Imaging, Mount Sinai Hosp and Univ Health Network (L.M.W.), Dept of Public Health Sciences, Clinical Epidemiology and Health Services Research (A.D.), and Dept of Orthopedic Surgery, Orthopedic and Arthritic Hosp (P.H.M.), Univ of Toronto, 600 University Ave, Toronto, Ontario, Canada M5G 1X5; Dept of Radiology, Thomas Jefferson Univ Hosp, Philadelphia, Pa (M.E.S.); Institute of Diagnostic Imaging, Univ Hosp Zurich, Switzerland (D.W.); and Institute for CT and MR Imaging at Schiller Park, Linz, Austria (J.K.). From the 2000 RSNA scientific assembly. Received Jan 31, 2001; revision requested Mar 1; revision received Jun 13; accepted Jul 5. L.M.W. supported by RSNA Research and Education Foundation. A.D. supported by Health Career Award, Canadian Institutes of Health Research, SSHCR/NHRDP.
| ABSTRACT |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
MATERIALS AND METHODS: Three hundred sixty-four patients who had previously undergone meniscal preservation surgery were prospectively examined with conventional MR imaging, indirect MR arthrography, and direct MR arthrography. Ninety-four patients (104 postoperative menisci) underwent subsequent second-look arthroscopic surgery. Each case was evaluated for (a) surfacing intrameniscal intermediate- or T1-weighted signal intensity, (b) surfacing intrameniscal T2-weighted signal intensity, (c) morphologic changes beyond those expected postoperatively, (d) joint effusion on conventional MR or indirect MR arthrographic studies, and (e) overall presence or absence of recurrent meniscal tear.
RESULTS: Seventy-one arthroscopically proved recurrent meniscal tears were found. In the diagnosis of recurrent meniscal tears, sensitivity, specificity, positive predictive value, negative predictive value, and accuracy were 86%, 67%, 83%, 71%, and 80%, respectively, for conventional MR imaging; 83%, 78%, 90%, 64%, and 81%, respectively, for indirect MR arthrography; and 90%, 78%, 90%, 78%, and 85%, respectively, for direct MR arthrography. No significant difference in the diagnostic accuracy of one method relative to another was observed (P > .54). Surfacing intrameniscal T2-weighted signal intensity was the most specific sign, with the highest positive predictive value of a recurrent tear.
CONCLUSION: Although a small incremental increase in accuracy is associated with the use of direct MR arthrography over conventional MR imaging and indirect MR arthrography, no significant difference in diagnostic accuracy among the three techniques was demonstrated for detection of recurrent or residual meniscal tear.
Index terms: Knee, arthrography, 452.121411, 452.121412, 452.121415, 452.121416, 452.12143 • Knee, ligaments, menisci and cartilage, 452.4852 • Knee, MR, 452.121411, 452.121412, 452.121415, 452.121416, 452.12143
| INTRODUCTION |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
Recurrent or residual symptoms following meniscal preservation surgery may be related to residual or new tears of the meniscus or other causes of intra- or extraarticular abnormalities. Clinical identification of the meniscal origin of recurrent or residual symptoms may be confounded by the postoperative nature of the joint and by the nonmeniscal surgical procedures possibly performed.
Classic conventional magnetic resonance (MR) imaging signs of a tear of a meniscus on which surgery was never performed include intrameniscal signal intensity abnormality unequivocally extending to the meniscal surface at short-echo-time sequences and abnormal meniscal morphologic features (6,7). These signs have been shown to have limited diagnostic utility in the assessment of possible postoperative tears of the meniscus (8–11). This has led to a prevailing notion that MR imaging has a limited role in the assessment of the postoperative meniscus.
Results of subsequent investigations have advocated the use of different MR imaging criteria, such as visualization of surfacing linear intrameniscal signal intensity with associated intrameniscal fluid at conventional MR imaging (12–14), or the use of direct or indirect MR arthrography (15,16) to increase the accuracy and specificity of MR imaging in the evaluation of possible recurrent or residual tears of the postoperative meniscus. To our knowledge, there currently is no consensus as to the best MR imaging technique for the evaluation of possible postoperative meniscal tears.
The purpose of this study was to prospectively investigate the accuracy of conventional MR imaging, direct MR arthrography, and indirect MR arthrography in the assessment of possible recurrent or residual meniscal tears in patients with symptoms of possible joint derangement following prior meniscal preservation surgery.
| MATERIALS AND METHODS |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
All MR imaging examinations were performed with a 1.5-T unit (Signa; GE Medical Systems, Milwaukee, Wis) equipped with a dedicated quadrature extremity coil. Conventional MR imaging consisted of sagittal spin-echo (SE) intermediate-weighted sequences (1,200/20 [repetition time msec/echo time msec], 4-mm section thickness with 1-mm spacing, 256 x 256 matrix, 14 x 14-cm field of view, one signal acquired), sagittal and transverse T2-weighted fast SE sequences with fat saturation (4,000/60–75, echo train length of eight, 4-mm section thickness with 1-mm spacing, 256 x 256 matrix, 14 x 14-cm field of view, two signals acquired), and coronal intermediate-weighted fast SE sequences (3,800/45, echo train length of eight, 4-mm section thickness with 1-mm spacing, 256 x 256 matrix, 14 x 14-cm field of view, two signals acquired).
Indirect MR arthrography was performed with the intravenous administration of gadopentetate dimeglumine (0.1 mmol/kg) (Magnevist; Berlex Laboratories, Wayne, NJ) before MR imaging. No directed patient exercise preceded imaging. Direct MR arthrography was performed immediately following the intraarticular injection of 30–50 mL of a 2-mmol/L gadopentetate dimeglumine–saline solution. Intraarticular positioning of the injection needle was confirmed in all cases with an injection of less than 1 mL of the iodinated contrast material iopamidol (Isovue 300; Bracco Diagnostics, Princeton, NJ) with fluoroscopic observation. The intraarticular injection of the gadopentetate dimeglumine–saline solution was continued until resistance to injection was felt.
The imaging parameters for direct and indirect MR arthrography consisted of sagittal three-dimensional spoiled gradient-recalled-echo sequences with fat suppression (33.5/9.8, 45° flip angle, 1.5-mm section thickness, 256 x 160 matrix, 14 x 14-cm field of view, one signal acquired), sagittal T2-weighted fast SE sequences (4,000/60–75, echo train length of eight, 4-mm section thickness with 1-mm spacing, 256 x 256 matrix, 14 x 14-cm field of view, two signals acquired), and transverse and coronal T1-weighted SE sequences with fat saturation (450/16, 4-mm section thickness with 1-mm spacing, 256 x 160 matrix, 14 x 14-cm field of view, 1.5 signals acquired) followed by sagittal T1-weighted SE sequences with fat suppression (500/11, 4-mm section thickness with 1-mm spacing, 256 x 256 matrix, two signals acquired). Sagittal T1-weighted SE imaging was performed as the last sequence of the MR examination; this resulted in a 25–30-minute delay in indirect MR arthrogram acquisition between sagittal fat-suppressed T1-weighted SE imaging and the intravenous injection of contrast material.
The MR images were reviewed independently by two experienced musculoskeletal radiologists (L.M.W., D.W.) who were blinded to the clinical and surgical data, including the location, type, and extent of prior meniscal surgery; the date of prior meniscal surgery; the results of subsequent second-look arthroscopic surgery; and the overall frequency of recurrent meniscal tears in the study group at second-look arthroscopic surgery. Discrepancies were evaluated by both readers at a separate sitting, with agreement reached by means of consensus. The readers evaluated each study for (a) increased intrameniscal signal intensity extending to the meniscal articular surface at short-echo-time intermediate- or T1-weighted sequences, (b) increased intrameniscal signal intensity equal to that of joint fluid extending to the meniscal articular surface at T2-weighted sequences, (c) meniscal morphologic changes beyond those expected postoperatively, and (d) assessment of the presence or absence of recurrent or residual meniscal tear. The overall determination of the presence or absence of recurrent or residual meniscal tear was not made by using either an absolute combination or a lack of individual findings, but rather it was made as an overall assessment of the presence or absence of a tear by the interpreting reviewer. The conventional MR imaging and indirect MR arthrographic studies were evaluated also for the presence or absence of joint effusion, which was defined as a larger than 5-mm anteroposterior distension of the suprapatellar recess measured in the midline of the joint or a larger than 10-mm anteroposterior distension of the lateral joint recess measured on the last lateral sagittal image of the lateral recess of the joint (17).
Ninety-four patients with 104 postoperative menisci (71 medial and 33 lateral) underwent subsequent second-look arthroscopic surgery within 5 months after their MR examinations. Forty-two of these 94 patients underwent conventional MR imaging; 28, indirect MR arthrography; and 24, direct MR arthrography. There were 62 male and 32 female patients; their mean age was 40 years (age range, 17–74 years). The mean time between MR imaging and second-look arthroscopic surgery was 8.6 weeks (range, 1–20 weeks; median, 6 weeks). The mean time between prior meniscal surgery and second-look arthroscopic surgery was 88.2 months (range, 2–240 months; median, 14 months). In all cases, arthroscopy was performed by one of five experienced orthopedic surgeons (one of whom was P.H.M.), who were aware of the MR imaging findings.
Detailed surgical information regarding the type of prior meniscal surgery (ie, meniscal repair or meniscectomy) and the extent of meniscal resection was available for 57 of the 94 patients: 23 who underwent conventional MR imaging, 16 who underwent indirect MR arthrography, and 18 who underwent direct MR arthrography. Prior meniscal repair (ie, meniscal arrows or meniscal suturing) was documented for 11 patients. The degree of prior partial meniscal resection (ie, partial meniscectomy) was documented as less than 25% of the meniscus resected in 17 patients and as 25%–75% of the meniscus in 29 patients.
Results of second-look arthroscopic surgery were used as the standards of reference for determining the presence or absence of repeat or residual meniscal tear. The arthroscopic criteria for a possible recurrent or residual meniscal tear included the identification of an unstable meniscal fragment, flap tear, or intrameniscal cleavage plane judged to be causing mechanical effect in the joint. Such arthroscopic findings were also assessed in conjunction with each patient’s clinical complaints and possible findings at physical examination.
Statistical analyses of the collected data included evaluation of the sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy of each MR imaging technique and evaluation for each potential diagnostic sign assessed in the evaluation of a recurrent or residual meniscal tear. Differences in the accuracy of the different MR techniques evaluated in the diagnosis of recurrent or residual meniscal tears were analyzed by using logistic regression modelling. Similarly, logistic regression analysis was performed to assess the potential differences in diagnostic accuracy of conventional MR imaging and indirect MR arthrography in the assessment of a recurrent or residual meniscal tear in either the presence or absence of a joint effusion.
For cases with a detailed description of prior meniscal surgery, the diagnostic accuracy of the MR imaging modalities for assessment of recurrent or residual meniscal tears was analyzed separately for menisci on which prior meniscectomy of less than 25% of the meniscus was performed, on which prior meniscectomy of 25%–75% of the meniscus was performed, and with prior repairs. Finally, interobserver variability in the assessment of recurrent or residual meniscal tears was evaluated by using value analysis (18).
| RESULTS |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
In the diagnosis of recurrent meniscal tear, the finding of surfacing intrameniscal T1- or intermediate-weighted signal intensity at conventional MR imaging, indirect MR arthrography, and direct arthrography had sensitivities of 76%, 83%, and 90%, respectively; specificities of 53%, 56%, and 78%, respectively; and accuracies of 68%, 75%, and 86%, respectively. The finding of surfacing intrameniscal T2-weighted signal intensity at conventional MR imaging, indirect MR arthrography, and direct MR arthrography had sensitivities of 72%, 65%, and 84%, respectively; specificities of 73%, 100%, and 89%, respectively; accuracies of 73%, 75%, and 86%, respectively; and high PPVs: 84%, 100%, and 94%, respectively. At conventional MR imaging, indirect MR arthrography, and direct MR arthrography, morphologic changes of the meniscus beyond those attributable to prior surgery (Fig 7) had sensitivities of 60%, 61%, and 63%, respectively; specificities of 86%, 89%, and 67%, respectively; and accuracies of 68%, 69%, and 64%, respectively (Table 3).
|
|
| DISCUSSION |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|
At initial investigations of the value of MR imaging in the assessment of menisci following meniscal preservation surgery, findings illustrated the spectrum of MR imaging appearances after meniscal surgery, and the pitfalls of applying the standard MR imaging criteria of contour irregularity and increased surfacing (grade-3) signal intensity in the diagnosis of recurrent meniscal tears were described. Smith and Totty (11) examined 40 patients with a history of partial meniscectomy. They described variable MR imaging appearances of meniscal remnants and difficulties in using the classic MR criteria of tears in the assessment of the postsurgical meniscus. Deutsch et al (9) reviewed the MR studies obtained in 17 patients and performed before and after conservative therapy or repair of meniscal tears. They noted persistent increased surfacing intrameniscal (grade-3) signal intensity, a standard MR imaging criterion of meniscal tear, in all patients despite arthroscopic or clinical evidence of meniscal healing.
Farley et al (12), in an investigation involving 30 patients who previously underwent arthroscopic meniscal repairs, similarly found that the presence of grade 3 signal intensity within the postoperative meniscus at short-echo-time MR sequences was not an accurate indicator of recurrent meniscal tear. However, similar to us, they did find that the corresponding increased surfacing signal intensity extending through a full-thickness defect of the meniscus at T2-weighted MR imaging was specific for recurrent meniscal tear. Such signal intensity on T2-weighted MR images was presumed to represent an arthrographic effect of free joint fluid tracking through the tear site.
The findings of subsequent investigations of the utility of conventional MR imaging have indicated that increased intrameniscal signal intensity extending to the meniscal surface at T2-weighted MR imaging is indicative of possible reparative fibrovascular scar tissue at the site of incompletely healed tears or of fluid at the site of recurrent residual tear (13,14). Lim et al (13), by using the sign of intrameniscal fluid within a line extending to the meniscal surface as evidence of a recurrent meniscal tear, found conventional MR imaging to have high specificity (88%) and diagnostic accuracy (82%) in the assessment of the postoperative meniscus; these findings are similar to our results.
Indirect MR arthrography has been proposed as a potential means of increasing the conspicuity of meniscal tears (19,20). The technique obviates fluoroscopic guidance of needle placement and the joint puncture needed with direct MR arthrography. Although joint distension is not achieved with this technique, joint fluid enhancement following the intravenous administration of gadolinium-based contrast material has been suggested as a means of increasing the conspicuity of joint fluid imbibition at the site of recurrent or residual meniscal tear. In addition, indirect MR arthrography has the potential to cause enhancement of vascular and cellular proliferation within the fibrovascular scar tissue that is seen histologically at the margins of torn or healing menisci (21,22).
Investigators have studied the enhancement patterns of meniscal repairs as possible indicators of the status of meniscal healing. In a study of indirect MR arthrography performed in 22 patients within 12 months after meniscal repair, Tanaka et al (23) observed intrameniscal enhancement at the site of prior surgery, in the absence of recurrent meniscal tear. Such assessment of enhancement patterns in early postoperative menisci was not performed in the current study. Our study results did not show a significant difference between indirect MR arthrography and either conventional MR imaging or direct MR arthrography in the diagnosis of meniscal tear.
Direct MR arthrography has been advocated as a means of assessing recurrent postoperative tears of the meniscus, with the imbibition of contrast material into the meniscus following iatrogenic distension of the joint indicating recurrent meniscal tear. Applegate et al (15), in a study involving 37 patients who previously had undergone meniscal surgery, observed a higher diagnostic accuracy of direct MR arthrography (88%) relative to conventional MR imaging (66%) in the diagnosis of recurrent meniscal tears. The results of a more recent investigation (16), in which direct MR arthrography, conventional arthrography, and conventional MR imaging were performed in a small number of patients (n = 12) with prior meniscal surgery, showed direct MR arthrography with gadolinium-based contrast material to be the most accurate of the techniques evaluated. However, the study results did not show a significant difference between direct MR arthrography and conventional MR imaging in the evaluation of the postoperative meniscus. Similarly, we observed a small incremental but not significant increase in diagnostic accuracy for direct MR arthrography compared with conventional MR imaging and indirect MR arthrography in the diagnosis of recurrent or residual meniscal tears.
Our study results indicate high overall accuracy for conventional MR imaging, indirect MR arthrography, and direct MR arthrography in the diagnosis of recurrent or residual meniscal tears. As in previous studies of potential MR imaging signs of recurrent meniscal tears, in the current study, we observed increased intrameniscal signal intensity extending to the meniscal surface on T2-weighted images to be the most specific sign assessed and to have the highest PPV for recurrent or residual meniscal tear. As expected, the finding of increased signal intensity extending to the meniscal surface at short-echo-time intermediate- or T1-weighted imaging was the most sensitive of the signs assessed; however, it had low specificity in the diagnosis of recurrent tear.
The presence of joint effusion at conventional MR imaging or indirect MR arthrography increased the overall diagnostic accuracy of the modalities; however, this trend was not statistically significant. This finding was consistent with the trend of a small incremental—yet not significant—increase in the diagnostic accuracy of direct MR arthrography compared with that of conventional MR imaging that we observed. This trend was related presumably to the iatrogenic induction of joint effusion and the distension of the joint.
Similar to findings in prior studies (9,15), our results corroborate the findings of high diagnostic accuracy of MR imaging in the diagnosis of recurrent meniscal tear in the setting of prior resection of less than 25% of the meniscus, with lower accuracy in the diagnosis of recurrent tear following prior resection of 25%–75% of meniscal tissue. In our study, recurrent meniscal tears were found at second-look arthroscopy in all cases with a history of prior meniscal repair. All recurrent tears in this group of patients who had previously undergone meniscal repair were correctly diagnosed by using MR imaging, regardless of the technique used.
Each patient examined in our investigation was randomly assigned to one of three MR imaging protocols. As a result, different groups were studied with each imaging technique. In addition, for ethics board approval of the study protocol, patients had the right to decline any one imaging protocol in favor of another, and this resulted in an unequal distribution of patients in the three arms of the investigation. A study design in which each patient underwent all three imaging protocols before second-look surgery would have provided a more direct comparison of each of the techniques; however, due to practical limitations, this was not feasible.
An acknowledged limitation of our study is that, because of ethical reasons, it was not possible to blind the referring surgeons to the findings of preoperative MR imaging. Knowledge of such findings may have influenced the clinical decision to proceed or not to proceed to arthroscopy and thus created a selection bias in favor of a positive meniscal pathologic entity in patients proceeding to second-look arthroscopic surgery and possibly a bias in the arthroscopic evaluation of the meniscus. In addition, in the majority of cases, only those patients with preoperative clinical and imaging findings that were amenable to possible arthroscopic intervention underwent subsequent arthroscopic surgery. As a result, a selection bias toward clinically important meniscal lesions that were associated with mechanical instability and clinically important complaints and symptoms was possibly created. It is possible that the clinically occult residual or recurrent meniscal tears that did not result in imaging and second-look arthroscopy may not have been defined as accurately at preoperative MR imaging examination. Despite this potential limitation, it is this clinical subgroup of symptomatic patients after meniscal preservation surgery in whom second-look arthroscopy is contemplated and MR imaging has the greatest potential clinical utility and value.
An additional limitation in determining the diagnostic accuracy of MR imaging in the assessment of menisci involves the use of arthroscopy as the standard of reference for evaluation of meniscal tear. The potential limitations of arthroscopy are well described (24), and although experienced orthopedists performed arthroscopy in all cases in the current study, there may be some subjective variability among individual arthroscopists in the diagnostic assessment of a possible residual or recurrent meniscal tear.
In the evaluation of MR arthrographic signs of possible recurrent meniscal tear, we did not subcategorize increased signal intensity extending to the meniscal articular surface on T1-weighted images as either equivalent to or lower than the signal intensity of gadolinium-based contrast material on indirect or direct MR arthrograms. As a result, the specificity of the finding of surfacing signal intensity on T1-weighted MR arthrographic studies in the diagnosis of recurrent meniscal tear in our investigation may be lower than the specificity that may have been observed if the finding were limited to surfacing signal intensity equivalent to that of intraarticular gadolinium-based contrast material on T1-weighted studies.
Finally, a limitation of our imaging protocol design may have been the omission of a coronal T2-weighted sequence. This sequence may have been of diagnostic value by aiding in the visualization of surfacing intrameniscal signal intensity that was not seen on sagittal views alone at T2-weighted imaging.
In conclusion, we observed high overall diagnostic accuracy of conventional MR imaging, indirect MR arthrography, and direct MR arthrography in the evaluation of possible recurrent or residual tears of the postoperative meniscus. Although we observed a small incremental increase in diagnostic accuracy with direct MR arthrography, as compared with the accuracies of conventional MR imaging and indirect MR arthrography, no significant difference in diagnostic accuracy among the three evaluated techniques was demonstrated.
| ACKNOWLEDGMENTS |
|---|
| REFERENCES |
|---|
| Top ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES |
|---|