MR Arthrography of the Glenohumeral Joint: Two Concentrations of ...

14 downloads 0 Views 2MB Size Report
SLAP superior labrum anteroposterior. 1 From the Departments of Radiology. (C.A.B., M.Z., J.H.) and Orthopedic Sur- gery (C.G.), University Hospital Balgrist,.
Christoph A. Binkert, MD Marco Zanetti, MD Christian Gerber, MD Juerg Hodler, MD

Index terms: Arthrography, contrast media, 414.122 Contrast media, comparative studies Gadolinium Magnetic resonance (MR), arthrography, 414.121411, 414.121415, 414.12143 Shoulder, arthrography, 414.122 Tendons, injuries, 414.41, 414.481 Radiology 2001; 220:219 –224 Abbreviation: SLAP ⫽ superior labrum anteroposterior 1

From the Departments of Radiology (C.A.B., M.Z., J.H.) and Orthopedic Surgery (C.G.), University Hospital Balgrist, Forchstrasse 340, CH-8008 Zurich, Switzerland. Received July 6, 2000; revision requested August 19; final revision received January 8, 2001; accepted January 16. Address correspondence to C.A.B., Dotter Interventional Institute, Oregon Health Sciences University, L605, 3181 SW Sam Jackson Park Rd, Portland, OR 97201-3098 (e-mail: [email protected]).

©

RSNA, 2001

MR Arthrography of the Glenohumeral Joint: Two Concentrations of Gadoteridol versus Ringer Solution as the Intraarticular Contrast Material1 PURPOSE: To compare two concentrations of gadoteridol with Ringer solution as the contrast material for magnetic resonance (MR) arthrography of the glenohumeral joint. MATERIALS AND METHODS: One hundred fifty-six consecutive MR arthrograms were randomly obtained with either 2 mmol/L gadoteridol (n ⫽ 52), 4 mmol/L gadoteridol (n ⫽ 52), or Ringer solution (n ⫽ 52). MR arthrograms were assessed quantitatively (for contrast-to-noise ratio [CNR]) and qualitatively (for overall image quality, image contrast, degree of joint distention, and motion artifacts). MR diagnoses were compared with arthroscopic or surgical reports in 88 patients. RESULTS: The mean CNR at imaging was 40.4 with 2 mmol/L gadoteridol, 45.6 with 4 mmol/L gadoteridol, and 48.7 with Ringer solution. The CNR with 2 mmol/L gadoteridol was significantly lower than that with 4 mmol/L gadoteridol (P ⫽ .025) and Ringer solution (P ⫽ .012). Qualitative differences between the two gadoteridol concentrations were not significant. Ringer solution was significantly worse with regard to overall quality, motion artifacts, image contrast, and joint distention compared with both gadoteridol concentrations. Ringer solution was slightly more sensitive and less specific than the gadoteridol solutions in the detection of supraspinatus tears and less sensitive and more specific in enabling diagnosis of superior labrum anteroposterior lesions. CONCLUSION: MR arthrograms of the shoulder obtained with gadoteridol and those obtained with Ringer solution provided equivalent diagnostic accuracy. The authors, however, preferred the image quality of the gadoteridol-enhanced arthrograms.

Author contributions: Guarantor of integrity of entire study, C.A.B.; study concepts, C.A.B., J.H.; study design, M.Z., C.A.B., J.H.; literature research, C.A.B.; clinical studies, all authors; data acquisition, C.A.B., C.G.; data analysis/interpretation, M.Z., J.H.; statistical analysis, C.A.B., J.H.; manuscript preparation, C.A.B.; manuscript definition of intellectual content, J.H.; manuscript editing and revision/ review, M.Z., J.H.; manuscript final version approval, all authors.

Magnetic resonance (MR) arthrography performed with gadolinium-based contrast material appears to be more accurate than standard MR imaging for assessing the rotator cuff, especially those accompanying partial tears (1,2). Fat-suppressed sequences further improve the diagnostic performance of MR arthrography (3). Controversies related to the cost, invasiveness, and, in many countries, the nondefined legal status of intraarticular gadolinium-based solutions have spurred interest in alternative contrast materials. One such alternative solution is saline, which is inert, inexpensive, and legally unproblematic. In two investigations (4,5) involving the use of saline for MR arthrography, there were good results in detecting abnormalities associated with glenohumeral instability. However, no direct comparison with intraarticular gadolinium-based contrast material was performed in those studies. In another investigation (6), MR arthrograms obtained with gadolinium-based contrast material and with saline were compared both qualitatively and quantitatively. However, the influence of the different contrast materials on interobserver agreement and surgical confirmation of the diagnosis was not evaluated in that study. The purpose of the present investigation was to compare two concentrations of a 219

TABLE 1 Quantitative and Qualitative Evaluation Contrast Material

Mean CNR ⫾ SD*

Overall Quality†

Image Contrast†

Joint Distension†

Motion Artifacts†

2 mmol/l gadoteridol 4 mmol/l gadoteridol Ringer solution P value‡ 2 mmol/L vs 4 mmol/L gadoteridol 2 mmol/L gadoteridol vs Ringer solution 4 mmol/L gadoteridol vs Ringer solution

40.4 ⫾ 11.7 45.6 ⫾ 11.4 48.7 ⫾ 19.9

4.0, 3.8 (0.24) 4.0, 4.0 (0.51) 3.6, 3.5 (0.24)

4.3, 4.4 (0.29) 4.5, 4.5 (0.37) 4.0, 4.0 (0.21)

4.0, 4.2 (0.23) 4.4, 4.3 (0.44) 3.6, 3.7 (0.45)

3.5, 3.7 (0.48) 3.7, 3.7 (0.41) 3.3, 3.5 (0.31)

.03 .01 .30

.7 .003 .003

.50 .005 ⬍.001

.70 .002 .002

.50 .01 .006

* CNR ⫽ contrast-to-noise ratio. † Data are grades based on a scale of 1–5. For each set of data, the first number is the grade assigned by reader 1, and the second number is the grade assigned by reader 2. The numbers in parentheses are ␬ values. ‡ P values for differences in grades obtained by performing the Wilcoxon rank sum test. P values for differences in CNRs obtained by performing the unpaired Student t test.

gadolinium-based contrast material with Ringer solution for MR arthrography of the shoulder by assessing quantitative and qualitative image parameters and diagnostic efficacy, with surgery or arthroscopy as the standard of reference.

MATERIALS AND METHODS Patients Between March 1999 and January 2000, 156 consecutive patients (64 female, 92 male; age range, 14 – 83 years; mean age, 51 years) referred for MR arthrography of the glenohumeral joint participated in this study. The only exclusion criterion was prior shoulder surgery or arthroscopy. Our hospital’s ethics committee approved the study, and informed consent was obtained from all patients.

MR Arthrography MR arthrography was performed with a 20-gauge needle and fluoroscopic guidance. Local anesthesia was induced. The intraarticular position of the needle tip was confirmed by injecting 1 mL of iopamidol (Iopamiro 200; Bracco Diagnostics, Milan, Italy). One milliliter of mepivacain (Scandicain 2%; AstraZeneca, So ¨ derta¨lje, Sweden) was then injected, followed by 10 mL of contrast material. The total injected volume was 12 mL. On the basis of our experience, this standardized volume enables delineation of the relevant structures without leakage in the majority of patients. The contrast material used was either 2 mmol/L gadoteridol (ProHance; Bracco Diagnostics), 4 mmol/L gadoteridol, or Ringer solution (sodium chloride 0.9%, potassium chloride 0.03%, calcium chloride 0.015%, sodium hydrocarbonate 0.017%; Braun Medical, Melsungen, Ger220



Radiology



July 2001

many). The two gadoteridol solutions were prepackaged by the manufacturer and did not require additional dilution. The three types of contrast material were administered in 52 patients each. They were assigned to the patients randomly by using the random number function of a spreadsheet (EXCEL 97; Microsoft, Redmond, Wash). The orthopedic surgeons at our hospital prefer Ringer solution over saline for irrigation during arthroscopy. The results of an in vitro study (7) involving the use of rat articular cartilage demonstrated that Ringer solution is the best agent for joint irrigation during arthroscopy or other procedures. MR imaging was performed by using a 1.0-T unit (Siemens Expert; Siemens Medical Systems, Erlangen, Germany) equipped with a dedicated receive-only shoulder coil. For the gadoteridol-enhanced arthrograms, T1-weighted transverse (580/20 [repetition time msec/echo time msec], 4-mm section thickness, 140 ⫻ 160-mm field of view, 224 ⫻ 512 matrix) and parasagittal (700/12; 5-mm section thickness, 160 ⫻ 160-mm field of view, 192 ⫻ 256 matrix) images, as well as angled coronal dual-spin-echo (3,500/14 and 85, 4-mm section thickness, 184 ⫻ 210-mm field of view, 230 ⫻ 512 matrix) and T1-weighted fat-suppressed (800/20, 4-mm section thickness, 160 ⫻ 160-mm field of view, 192 ⫻ 256 matrix) images were obtained. For the arthrograms obtained with Ringer solution, the dual-spin-echo sequence was performed in all three planes. An angled coronal T2-weighted fat-suppressed spinecho sequence (4,500/96, 4-mm section thickness, 160 ⫻ 160-mm field of view, 196 ⫻ 256 matrix) was added. The total imaging time was 16 minutes 1 second for the gadoteridol-enhanced arthrograms compared with 20 minutes 48 seconds

for the Ringer solution– enhanced arthrograms.

Technical Efficacy For quantitative assessment, the angled coronal image closest to the center of the humeral head was chosen. The fat-suppressed T1-weighted sequence was selected for evaluation of the gadoteridol-enhanced arthrograms, and the fatsuppressed T2-weighted sequence was selected for evaluation of the Ringer solution– enhanced arthrograms. These sequences were chosen because we believe that they are the most useful for assessment of intraarticular abnormalities on MR arthrograms. Contrast-tonoise ratios were calculated as follows: Signal intensities were measured by the first author (C.A.B.) by using the standard MR console and the MR unit manufacturer’s software (Siemens Medical Systems). Circular regions of interest were drawn in the areas of the axillary recess, the most normal part of the supraspinatus tendon, and the air lateral to the body. The region-of-interest sizes in the axillary recess and supraspinatus tendon were made as large as possible without including adjacent structures. The region of interest in the air was 10 cm2. Contrast-to-noise ratios were calculated by dividing the difference in mean signal intensity of fluid between the axillary recess and the supraspinatus tendon by the SD of the background signal intensity (8). Two staff radiologists (M.Z., J.H.) specializing in musculoskeletal radiology independently performed a qualitative evaluation. They had not been involved in the quantitative assessment. The readers were blinded to the gadoteridol concentration used; however, blinding with regard to the Ringer solution protocol Binkert et al

TABLE 2 Assessment of Supraspinatus Tendon Tears and SLAP Lesions Contrast Material

Prevalence

Sensitivity

Specificity

Categorical Agreement

Interobserver Agreement

Supraspinatus tendon full-thickness tears 2 mmol/L gadoteridol

40 (12/30)

92 (11/12) 100 (12/12) 91 (10/11) 100 (11/11) 100 (8/8) 100 (8/8)

94 (17/18) 100 (18/18) 87 (13/15) 93 (14/15) 79 (19/24) 88 (21/24)

63 (19/30) 80 (24/30) 69 (18/26) 73 (19/26) 66 (21/32) 72 (23/32)

0.59

67 (4/6) 50 (3/6) 71 (5/7) 71 (5/7) 50 (3/6) 50 (3/6)

79 (11/14) 93 (13/14) 60 (6/10) 80 (8/10) 100 (19/19) 89 (17/19)

50 (10/20) 50 (10/20) 53 (9/17) 47 (8/17) 56 (14/25) 46 (11/24)

4 mmol/L gadoteridol

42 (11/26)

Ringer solution

25 (8/32)

SLAP lesions 2 mmol/L gadoteridol

30 (6/20)

4 mmol/L gadoteridol

35 (6/17)

Ringer solution

24 (6/25)

0.64 0.51 0.52 0.56 0.48

Note.—All data, except those for interobserver agreement, which are ␬ values, are percentages. The numbers in parentheses are numbers of tears or lesions. For sensitivity, specificity, and categorical agreement, the data on the first line are results obtained by reader 1, and those on the second line are results obtained by reader 2.

was not possible because of differences in the sequences used. Several image aspects were graded on a scale of 1–5, with 1 defined as worst quality and 5 defined as perfect quality: (a) For overall image quality, a score of 5 meant excellent—that is, perfect contrast between the relevant structures, crisp images, and no artifacts; 3, moderate—that is, contrast not perfect for some structures, some blurring, and some artifacts, but diagnosis still possible; and 1, inferior—that is, contrast insufficient for all structures, severe blurring, and/or severe artifacts, and diagnosis may not be possible. (b) For subjective contrast between joint fluid and adjacent structures, a score of 5 meant excellent—that is, best signal intensity difference; 3, moderate— that is, visible signal intensity difference and still sufficient for diagnostic purposes; and 1, inferior—that is, no visible signal intensity difference. (c) For degree of glenohumeral joint distention, a score of 5 meant full distension—that is, complete delineation of the undersurface of the supraspinatus tendon and separation of the joint capsule from the labrum; 3, moderate distension—that is, either supraspinatus tendon or labrum not delineated, or both structures partially delineated; and 1, poor distension—that is, neither the supraspinatus tendon nor the labrum delineated. (d) For extent of motion artifacts, 5 meant no motion artifacts; 3, moderate artifacts—that is, diagnosis still possible; and 1, severe artifacts—that is, diagnosis may not be possible. ␬ statistics were used to calculate interobserver agreement (9): a ␬ value less than 0.20 indicated poor agreeVolume 220



Number 1

ment; 0.21– 0.40, fair agreement; 0.41– 0.60, moderate agreement; 0.61– 0.80, good agreement; and 0.81–1.00, very good agreement.

performed to compare gradings; the mean grade of the two observers was used. A P value of less than .05 indicated a statistically significant difference.

Diagnostic Efficacy

RESULTS

In addition to the quantitative and qualitative evaluations of image quality, the diagnostic efficacy of the different contrast material concentrations was assessed. The same two readers who performed the qualitative grading made a diagnosis regarding the supraspinatus, infraspinatus, and subscapularis tendons; the superior and anteroinferior labra; and the intraarticular part of the long biceps tendon. The two readers performed this evaluation independently and were blinded with regard to clinical and intraoperative findings. For the rotator cuff tendons, the following diagnoses were made: normal, partial tear, or full thickness tear. For the biceps tendon, a diagnosis of normal, tendinopathy, or tear was made. For the anteroinferior labrum, a diagnosis of normal, degeneration, or tear was made. For the superior labrum, a diagnosis of normal, degeneration, or superior labrum anteroposterior (SLAP) lesion was made. Established criteria from the literature (1–3,10 –14) were used to differentiate the various abnormalities. MR findings were compared with those of arthroscopy (n ⫽ 62) or open surgery (n ⫽ 26). Sensitivity, specificity, categorical agreement, and interobserver agreement were calculated. The unpaired Student t test was performed to compare contrast-to-noise ratios. The Wilcoxon rank sum test was

Technical Efficacy The results are summarized in Table 1. The contrast material was injected correctly in all 156 patients. The mean contrast-to-noise ratio for arthrograms obtained with 4 mmol/L gadoteridol was 45.6. This value was significantly higher than the mean contrast-to-noise ratio for the 2 mmol/L gadoteridol-enhanced arthrograms (40.4, P ⫽ .03) and lower than that for the Ringer solution– enhanced arthrograms (48.7, P ⫽ .30). Ringer solution– enhanced arthrograms were more variable than gadoteridol-enhanced arthrograms, as indicated by the SDs: 19.9 for Ringer solution vs 11.7 for 2 mmol/L gadoteridol and 11.4 for 4 mmol/L gadoteridol. In the qualitative evaluation, the gadoteridol-enhanced arthrograms were better than the Ringer arthrograms with regard to overall image quality, image contrast, degree of joint distention, and motion artifacts (Table 1). The differences between the 4 mmol/L and 2 mmol/L gadoteridol-enhanced arthrograms were not significant.

Diagnostic Efficacy Contrary to the results found with regard to technical efficacy, no major differences in diagnostic performance were found among the different intraarticular

Gadoteridol versus Ringer Solution for Glenohumeral Joint MR Arthrography



221

Figure 1. Coronal T1-weighted fat-suppressed MR arthrogram (800/20) obtained with 2 mmol/L gadoteridol in a 64-year-old woman with a surgically proved full-thickness supraspinatus tendon tear (arrow) shows contrast material in the subdeltoid bursa (arrowheads).

Figure 2. Coronal T1-weighted fat-suppressed MR arthrogram (800/20) obtained with 4 mmol/L gadoteridol in a 55-year-old man clearly delineates a full-thickness supraspinatus tendon tear (arrow), which was confirmed at surgery.

contrast materials. On all three types of arthrograms, the prevalences of abnormalities were low for several anatomic structures, which, therefore, were not evaluated in depth. The prevalence of a subscapularis tendon tear was two (7%) of 30 at 2 mmol/L gadoteridol arthrography, one (4%) of 26 at 4 mmol/L gadoteridol arthrography, and zero of 32 at Ringer solution arthrography. Infraspinatus tendon tear was seen during surgery in four (13%) of 30 cases with 2 mmol/L gadoteridol, in two (8%) of 26 with 4 mmol/L gadoteridol, and in one (3%) of 32 with Ringer solution. A long biceps tendon tear was surgically confirmed in two (7%) of 30 surgical cases with 2 mmol/L gadoteridol, in two (8%) of 26 with 4 mmol/L gadoteridol, and in zero of 32 with Ringer solution. During arthroscopy, an anteroinferior labrum tear was found in two (10%) of 20 patients by using 2 mmol/L gadoteridol, in one (6%) of 17 patients by using 4 mmol/L gadoteridol, and in two (8%) of 25 patients by using Ringer solution. Higher prevalences were found for supraspinatus tendon tears and SLAP lesions, so these were further analyzed (Table 2). For detection of supraspinatus tears (Figs 1–3), sensitivity was excellent for all the contrast materials and for both readers (91%–100%). The specificity of the gadoteridol-enhanced arthrograms (87%–100%) was higher than that of the Ringer solution– enhanced arthrograms (reader 1, 79%; reader 2, 88%). On the basis of categorical agreement data, the Ringer solution– enhanced arthrograms (reader 1, 66%; reader 2, 72%) did not

differ consistently from the gadoteridolenhanced arthrograms (63%– 80%). Interobserver agreement was better for the gadoteridol-enhanced arthrograms (␬ ⫽ 0.59 for 2 mmol/L and ␬ ⫽ 0.64 for 4 mmol/L) than for the Ringer arthrograms (␬ ⫽ 0.51). Sensitivity for SLAP lesion detection (Figs 4, 5) was highest with 4 mmol/L gadoteridol (71% for both readers) and lowest with Ringer solution (50% for both readers). On the other hand, specificity for SLAP lesion detection was lowest with 4 mmol/L gadoteridol (reader 1, 60%; reader 2, 80%) (Fig 6) and highest with Ringer solution (reader 1, 100%; reader 2, 89%) (Fig 7). Diagnostic efficacy, as measured by using categorical agreement, was similar in the different contrast material groups (46%–56%) (Table 2). Interobserver agreement was best with 4 mmol/L gadoteridol-enhanced arthrograms (␬ ⫽ 0.56), followed by 2 mmol/L gadoteridol (␬ ⫽ 0.52) and Ringer solution (␬ ⫽ 0.48) arthrograms, although the differences were not large.

222



Radiology



July 2001

DISCUSSION MR arthrograms obtained with gadolinium-based contrast material are being used increasingly to detect shoulder abnormalities (1–3,10 –14). In laboratory settings, gadopentetate dimeglumine diluted with variable portions of saline has been shown to be superior to human serum albumin or saline for MR arthrography of the knee (15). In that cadaver study (15), optimal signal intensity combined with the lowest possible contrast

Figure 3. Coronal T2-weighted fat-suppressed spin-echo MR arthrogram (4,500/96) obtained with Ringer solution in a 74-year-old man with an arthroscopically proved partial supraspinatus tendon tear shows increased signal intensity (arrow) at the insertion of the supraspinatus tendon and fluid (arrowheads) in the subdeltoid bursa. Both readers misinterpreted these findings as a full-thickness supraspinatus tendon tear.

material concentration was achieved at T1-weighted MR arthrography with 0.4 – 0.6 mmol/L gadopentetate dimeglumine. The gadolinium-based contrast material concentrations in cadaver or laboratory studies may not be optimal in clinical situations, in which various aspects, such as the filling factor, distances or technical variations of the MR coil, field inhomogeneity in the required eccentric position, motion artifacts, preexisting fluid, and not always clearly defined mixtures of gadolinium-based contrast material combined with local anesthetics and iodine-containing contrast material used for needle positioning, may influence signal intensity and other factors relating to image quality. Most clinical studies focusing on evaluation of MR arthrograms of the glenohumeral joint have involved an intraarticular gadolinium-based contrast material concentration of 2 mmol/L (1–3, 16). A few authors (14,18) have used a higher concentration (4 mmol/L). Our data indicate that despite objectively better contrast (measured as the contrast-tonoise ratio) with 4 mmol/L versus 2 mmol/L gadoteridol, the differences are not relevant according to qualitative criteria. The focus of our investigation, however, was not to determine the appropriate contrast material concentration but rather to compare gadoteridol with Ringer solution, a potentially inexpensive and safe agent, similar to pure saline. The diagnostic performance of saBinkert et al

Figure 4. Coronal T1-weighted fat-suppressed MR arthrogram (800/20) obtained with 4 mmol/L gadoteridol in a 41-year-old woman with an arthroscopically proved SLAP lesion clearly demonstrates a SLAP lesion (arrow), with contrast material at the base of the superior labrum.

Figure 5. Coronal T2-weighted fat-suppressed spin-echo MR arthrogram (4,500/96) obtained with Ringer solution in a 26-year-old man with an arthroscopically proved SLAP lesion shows a subtle signal intensity increase (arrow) at the base of the superior labrum. The correct diagnosis was made by only one reader.

line-enhanced arthrograms for detection of anterior glenohumeral instability has proved to be good (4,5). Such arthrograms may be less useful for distinguishing partial- from full-thickness tears of the rotator cuff, because it is not possible to differentiate between preexisting fluid in the subacromial-subdeltoid bursa and contrast material leaking from the joint. To our knowledge, only one comparative investigation (6) based on differences in signal intensity and the qualitative assessment of image quality has been performed. It is not easy to perform such an investigation, however. Because of ethical considerations, it is not feasible to perform three arthrographic examinations Volume 220



Number 1

Figure 6. Coronal T1-weighted fat-suppressed spin-echo MR arthrogram (800/20 msec) obtained with 4 mmol/L gadoteridol in a 21-yearold man with an arthroscopically normal superior labrum shows extensive, blurred hyperintensity (arrow), which was misinterpreted as a SLAP lesion by both readers.

Figure 7. Coronal T2-weighted fat-suppressed spin-echo MR arthrogram (4,500/96) obtained with Ringer solution in a 65-year-old man with an arthroscopically proved SLAP lesion demonstrates well-delineated hyperintensity (arrow) at the base of the superior labrum. A partial supraspinatus tendon tear (arrowhead) also is shown. Both readers made the correct diagnosis.

with three different contrast materials in the same patient. Therefore, we used a randomized study setup and a relatively large number of patients to reduce the effects of interindividual differences. Another problem of investigations to assess intraarticular contrast materials is that the inclusion of patients on the basis of available surgical reports creates a bias toward severe, and therefore easy to diagnose, disease. We believe that our study, with 88 of 156 shoulders, included a relatively large proportion of patients with independent standards of reference,

without excluding patients who had not undergone surgery and thus had presumably less pronounced disease. With the standardized sequences used for this investigation, the Ringer solution– enhanced arthrograms performed best on the basis of mean contrast-tonoise ratios measured for intraarticular fluid versus supraspinatus tendon. Considering the experience reported by Zanetti and Hodler (6), this was not anticipated. In their study, however, a T2*weighted gradient-echo sequence was used to obtain the saline solution arthrograms and a T1-weighted gradient-echo sequence was used to obtain the gadolinium-enhanced arthrograms. Their results, therefore, cannot be directly compared with those of this study. In the current study, the mean contrast-to-noise ratio for the Ringer solution– enhanced arthrograms was higher than that for the gadoteridol-enhanced arthrograms. This has to be considered together with the SDs, however, which indicated higher variability on the Ringer solution– enhanced arthrograms. Uniform image quality is important for a consistent diagnosis to not only the radiologist but also the clinicians. The qualitative evaluations did not reveal significant differences between the two concentrations of gadoteridol. Ringer solution– enhanced MR arthrograms, however, had significantly lower scores for all four graded features—that is, overall image quality, image contrast, joint distention, and motion artifacts. There may have been relevant observer bias, especially because these results in part contradicted the quantitative evaluation results. This argument is strengthened by the fact that interobserver variability for image quality was high. However, the lower scores for joint distention suggest that the iso-osmolar Ringer solution is absorbed more rapidly than is gadolinium-based contrast materials. At our institution, MR imaging usually is performed within 5–15 minutes after arthrography. At institutions where there are longer delays between contrast material administration and MR imaging, this aspect may become even more relevant. Adding epinephrin or using higher volumes may solve this problem. Willemsen et al (5) injected up to 40 mL of saline in patients with an expanded articular capsule after recurrent anterior dislocation of the shoulder. Repetitive shoulder dislocation leads to an enlarged shoulder capsule; this partly explains the volume difference— 40 mL versus the 12 mL used in our study (with few patients

Gadoteridol versus Ringer Solution for Glenohumeral Joint MR Arthrography



223

with glenohumeral instability)—and the similar contrast material doses recommended by other investigators (5,16,18). Motion artifacts are not directly dependent on the type of intraarticular contrast material, but rather they are related to differences in the sequences used. The slightly lower grade for motion artifacts on the Ringer solution– enhanced arthrograms was probably due to a higher sensitivity to motion associated with the high fluid-to-background signal intensity difference and to the longer acquisition time associated with the fat-suppressed T2-weighted spin-echo sequence. In those study patients who underwent either open surgery or arthroscopy, the prevalence of infraspinatus, subscapularis, biceps, and anteroinferior labral abnormalities was too low for a more detailed analysis. For this reason, the following discussion is focused on supraspinatus and superior labral lesions. This is reflective of the referral habits of the shoulder surgeons at our institution, who order MR arthrography to help distinguish partial from small full-thickness tears of the rotator cuff, SLAP lesions, and biceps tendon abnormalities but not to identify shoulder instability (which is assessed with computed tomographic arthrography owing to the better demonstration of small fragments at the anteroinferior glenoid rim). All three types of arthrograms (Figs 1–3) were excellent in depicting supraspinatus tears, as indicated by the high sensitivities. However, the gadoteridol-enhanced MR arthrograms were slightly more specific (Figs 1, 2). This is probably explained by the fact that on Ringer solution– enhanced arthrograms, leaked intraarticular contrast material cannot be distinguished from preexisting fluid relating to bursitis (Fig 3). It is also interesting that the lower concentration of gadoteridol was slightly more specific than the higher concentration. Occasionally, continuous but degenerated rotator cuff tendons may be imbibed by contrast material, and this may lead to false-positive diagnoses. Such effects may be stronger with higher contrast material concentrations. The results for SLAP lesion detection were not very good, probably because of the previously discussed problem in differentiating between a normal recess and

224



Radiology



July 2001

a detachment (Figs 4 –7) (18). For SLAP lesions, the findings for sensitivity and specificity were the opposite of those for supraspinatus tendon lesions: The gadoteridol-enhanced arthrograms (Figs 4, 6) were more sensitive but less specific than the Ringer solution– enhanced arthrograms (Figs 5, 7). There are potentially confounding factors at the base of the superior labrum, including undercutting by hyaline articular cartilage, basal labral degeneration, and loose connective tissue connecting the labrum to the glenoid bone, all of which may cause increased signal intensity on MR images. Such aspects may be more pronounced on T1weighted images than on T2-weighted images. Therefore, the higher specificity of Ringer solution may not be related specifically to the contrast material but rather to the image sequence used with it. ␬ values for diagnostic accuracy were slightly but consistently higher for the gadoteridol-enhanced arthrograms. This is another indication—in addition to the contrast-to-noise ratio variability–that gadoteridol-enhanced arthrograms may lead to more consistent image characteristics. There may have been bias in this respect, however, because the readers were used to reading MR arthrograms obtained with intraarticular gadoliniumbased contrast material. In conclusion, MR arthrograms of the shoulder obtained with gadoteridol and those obtained with Ringer solution provide equivalent diagnostic accuracy. However, the image quality of the gadoteridol-enhanced arthrograms was preferred by the authors. References 1. Flannigan B, Kursunoglu-Brahme S, Snyder S, Karzel R, Del Pizzo W, Resnick D. MR arthrography of the shoulder: comparison with conventional MR imaging. AJR Am J Roentgenol 1990; 155:829 – 832. 2. Hodler J, Kursunoglu-Brahme S, Snyder SJ, et al. Rotator cuff disease: assessment with MR arthrography versus standard MR imaging in 36 patients with arthroscopic confirmation. Radiology 1992; 182:431– 436. 3. Palmer WE, Brown JH, Rosenthal DI. Rotator cuff: evaluation with fat-suppressed MR arthrography. Radiology 1993; 188: 683– 687. 4. Tirman PF, Stauffer AE, Crues JV III, et al. Saline magnetic resonance arthrography in the evaluation of glenohumeral instability. Arthroscopy 1993; 9:550 –559.

5.

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

17.

18.

Willemsen UF, Wiedemann E, Brunner U, et al. Prospective evaluation of MR arthrography performed with high-volume intraarticular saline enhancement in patients with recurrent anterior dislocations of the shoulder. AJR Am J Roentgenol 1998; 170:79 – 84. Zanetti M, Hodler J. Contrast media in MR arthrography of the glenohumeral joint: intra-articular gadopentetate vs saline: preliminary results. Eur Radiol 1997; 7:498 –502. Bulstra SK, Kuijer R, Eerdmans P, van der Linden AJ. The effect in vitro of irrigating solutions on intact rat articular cartilage. J Bone Joint Surg Br 1994; 76:468 – 470. Maubon AJ, Ferru JM, Berger V, et al. Effect of field strength on MR images: comparison of the same subject at 0.5, 1.0, and 1.5 T. RadioGraphics 1999; 19:1057– 1067. Altmann DG. Some common problems in medical research: practical statistics for medical research. London, England: Chapman & Hall, 1991; 396 – 435. Bencardino JT, Beltran J, Rosenberg ZS, et al. Superior labrum anterior-posterior lesions: diagnosis with MR arthrography of the shoulder. Radiology 2000; 214:267– 271. Shankman S, Bencardino J, Beltran J. Glenohumeral instability: evaluation using MR arthrography of the shoulder. Skeletal Radiol 1999; 28:365–382. Torstensen ET, Hollinshead RM. Comparison of magnetic resonance imaging and arthroscopy in the evaluation of shoulder pathology. J Shoulder Elbow Surg 1999; 8:42– 45. Wnorowski DC, Levinsohn EM, Chamberlain BC, McAndrew DL. Magnetic resonance imaging assessment of the rotator cuff: is it really accurate? Arthroscopy 1997; 13:710 –719. Zanetti M, Weishaupt D, Gerber C, Hodler J. Tendinopathy and rupture of the tendon of the long head of the biceps brachii muscle: evaluation with MR arthrography. AJR Am J Roentgenol 1998; 170:1557–1561. Hajek PC, Sartoris DJ, Neumann CH, Resnick D. Potential contrast agents for MR arthrography: in vitro evaluation and practical observations. AJR Am J Roentgenol 1987; 149:97–104. Chandnani VP, Gagliardi JA, Murnane TG, et al. Glenohumeral ligaments and shoulder capsular mechanism: evaluation with MR arthrography. Radiology 1995; 196:27–32. Pfirrmann CW, Zanetti M, Weishaupt D, Gerber C, Hodler J. Subscapularis tendon tears: detection and grading at MR arthrography. Radiology 1999; 213:709–714. Smith DK, Chopp TM, Aufdemorte TB, Witkowski EG, Jones RC. Sublabral recess of the superior glenoid labrum: study of cadavers with conventional nonenhanced MR imaging, MR arthrography, anatomic dissection, and limited histologic examination. Radiology 1996; 201: 251–256.

Binkert et al