Radiology
HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
QUICK SEARCH:   [advanced]


     

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Medina, L. S.
Right arrow Articles by Barnes, P. D.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Medina, L. S.
Right arrow Articles by Barnes, P. D.
(Radiology. 1999;211:767-771.)
© RSNA, 1999

Pediatric Imaging

Occult Lumbosacral Dysraphism in Children and Young Adults: Diagnostic Performance of Fast Screening and Conventional MR Imaging1

L. Santiago Medina, MD, MPH, May Al-Orfali, MD, David Zurakowski, PhD, Tina Young Poussaint, MD, James DiCanzio, MS and Patrick D. Barnes, MD

1 From the Department of Radiology, Health Services and Policy Section, Children's Hospital Medical Center, 3333 Burnet Ave, Cincinnati, OH 45236 (L.S.M.), and the Department of Radiology, Harvard Medical School, Children's Hospital, Boston, Mass (M.A.O., D.Z., T.Y.P., J.D., P.D.B.). Received January 27, 1998; revision requested April 7; revision received September 11; accepted November 9. Address reprint requests to L.S.M.


   Abstract
TOP
 Abstract
Introduction
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 
PURPOSE: To compare fast screening and conventional magnetic resonance (MR) imaging for the detection of occult dysraphic myelodysplasias in children and young adults.

MATERIALS AND METHODS: A retrospective case-control study included 101 patients (mean age, 4.9 years; range, 1 day to 26 years) suspected of having occult lumbosacral dysraphism. Sixty case patients had myelodysplastic lesions (19 filar lipoma, 14 syringomyelia, 10 intradural lipoma, eight dermal sinus, five diastematomyelia, five lipomyelomeningocele, two caudal regression syndrome); 41 control patients had no dysraphic lesions; 17 patients had associated renal anomalies. Two neuroradiologists reviewed MR images from conventional and fast screening protocols. Diagnostic performance parameters included sensitivity, specificity, and area under the receiver operating characteristic curve (Az value).

RESULTS: The sensitivity of conventional and fast screening MR studies was 97.1% and 98.5%, respectively; the specificity was 90.9% and 84.8%, respectively. The Az value was 0.973 for the fast screening and 0.976 for the conventional MR studies (P = .83). Interobserver agreement was very good for fast screening images ({kappa} = 0.68) and excellent for conventional images ({kappa} = 0.75). For renal anomalies, the Az value was 0.786 and 0.853 for fast screening and conventional MR imaging, respectively (P = .28).

CONCLUSION: Conventional three-plane lumbosacral MR imaging in children and young adults suspected of having occult dysraphism provides better diagnostic information than does fast screening two-plane MR imaging because of its higher specificity and interobserver agreement.

Index terms: Magnetic resonance (MR), in infants and children • Spinal cord, developmental defect, 33.145, 33.148, 33.361 • Spinal cord, MR, 33.121411 • Spinal cord, neoplasms, 33.361, 33.368, 33.369


   Introduction
TOP
 Abstract
 Introduction
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 
Occult spinal dysraphism is one of the most common indications for spinal magnetic resonance (MR) imaging in children and young adults. The clinical findings range from skin stigmata such as a dimple, hair patch, or hemangioma to dysfunctions of gait, bladder, or bowel (1–3). In addition, occult dysraphism commonly is associated with genitourinary tract anomalies (4). Spinal MR imaging, therefore, has the important role of helping determine the presence of a dysraphic myelodysplasia and associated urinary tract anomalies so that appropriate surgical and medical treatment can be instituted.

The authors of several descriptive studies (1–3,5–7) have illustrated the importance of spinal MR imaging for the characterization of a variety of occult dysraphic lesions, including filar lipoma, intradural lipoma, dermal sinus, lipomeningocele, and diastematomyelia. However, these studies have had two limitations. First, varying imaging protocols were used in these studies. Hence, the number of MR imaging sequences used per protocol ranged from two to six, depending on the institutional preference. Second, most of these studies lacked an adequate control group. Therefore, the diagnostic performance (ie, sensitivity, specificity, and receiver operating characteristic [ROC] curve) for the different lumbosacral MR protocols is not known. To our knowledge, no large analytic study has been performed to determine the diagnostic performance of different MR protocols in children and young adults suspected of having occult dysraphism. The determination of optimal MR imaging standards that are based on diagnostic performance is of prime importance to patient care, because of the ongoing restructuring of the field of health care.

We hypothesized that a fast screening MR imaging protocol that includes acquisition of only axial and sagittal T1-weighted images would have the same diagnostic performance as a conventional MR imaging examination and could be used as a screening examination for lumbosacral dysraphism. To test this hypothesis, a case-control study was conducted with children and young adults suspected of having lumbosacral occult dysraphism. A blinded comparative analysis of the results of a fast screening MR imaging protocol versus those of a conventional MR imaging protocol was performed to determine the diagnostic performance of both protocols in relation to the outcomes of dysraphism and associated renal anomalies.


   MATERIALS AND METHODS
TOP
 Abstract
 Introduction
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 References
 
Study Design
All patients in the study were referred for lumbosacral spinal MR imaging for evaluation of possible occult dysraphic myelodysplasia. The specific clinical findings are shown in Table 1. Patients with a history of nonoccult dysraphism (ie, spina bifida aperta or myelomeningocele), intracranial abnormalities, or central nervous system surgery were excluded from the study.


View this table:
[in this window]
[in a new window]
  		TABLE 1. Major Clinical Findings in 101 Patients Suspected of Having Occult Dysraphia
 
The study population of 101 patients (36 male and 65 female patients; mean age ± SD, 4.9 years ± 5.2; age range, 1 day to 26 years) was divided into a case group and a control group. The case group included 60 patients with conditions that represented a broad spectrum of occult dysraphic myelodysplasias (Table 2). The control group included 41 children and young adults with no myelodysplasia (Table 3). A neuroradiology database of cases from 1990 to 1996 was stratified, and a representative case mix for each study group was selected by one of the authors (L.S.M.).


View this table:
[in this window]
[in a new window]
  		TABLE 2. Final Diagnosis in 60 Patients with Suspected Occult Spinal Dysraphia
 

View this table:
[in this window]
[in a new window]
  		TABLE 3. Final Diagnosis in 41 Control Patients
 
The final diagnosis was determined on the basis of a combination of clinical, surgical, histopathologic, imaging, and laboratory findings. The mean follow-up was 5.4 months (range, 5 days to 1.8 years). Thirty-eight (63%) patients in the case group underwent surgery, and all had a demonstrable tethering lesion.

All patients underwent MR imaging with a 1.5-T system (Signa; GE Medical Systems, Milwaukee, Wis) equipped with a surface spine coil. A complete conventional lumbosacral MR imaging study was obtained in all patients. The MR imaging studies were subsequently classified in two protocol sets: conventional studies and fast screening studies (Table 4). The conventional four-sequence MR imaging protocol included the following pulse sequences: a thick-section sagittal spin-echo T1-weighted localizer sequence followed by thin-section sagittal, axial, and coronal spin-echo T1-weighted sequences. The fast screening two-sequence protocol included thin-section sagittal and axial spin-echo T1-weighted sequences.


View this table:
[in this window]
[in a new window]
  		TABLE 4. Lumbosacral MR Imaging Sequences
 
Definition of Important Variables
Definitions for the position of the conus medullaris included (a) normally positioned conus termination, which is defined as the tip of the conus medullaris at or above the L2-3 level, and (b) low conus termination, which is defined as the tip of the conus medullaris below the L2-3 level (8).

Study Analysis and Statistical Tests
The fast screening and conventional MR images were read independently by two experienced pediatric neuroradiologists (T.Y.P., P.D.B.), who had pediatric spinal MR imaging experience of 7 (T.Y.P.) and 13 years (P.D.B.). Both reviewers were blinded to the study population case mix, clinical data, and results of other imaging studies. All patient-identifying marks, including age, sex, and history, were covered with a black mask. In all cases, the fast screening MR study was read first, followed by the conventional MR study. The cases were read in random order with at least 4 weeks between readings of the two sets of studies. This was done to prevent test-interpretation and memory-recollection biases (9).

For occult dysraphic myelodysplasias and renal abnormalities, rating scores for each reader were dichotomized as positive or negative and were compared with the final diagnosis to determine the sensitivity and specificity. The 95% CIs were derived by using the Pratt approximation (10).

ROC curve analyses were performed because these analyses provide a description of disease detectability independent of the effects of disease prevalence and decision threshold (11). ROC curves were obtained by using a questionnaire with a five-point confidence-rating scale: score of 1, definitely not present; score of 2, probably not present; score of 3, indeterminate; score of 4, probably present; and score of 5, definitely present. ROC curve analyses were performed by using the ROC ANALYZER program for Windows (Centor RM, University of Alabama, Birmingham). The area under the curve (Az) and the standard error of Az were calculated by using the nonparametric trapezoidal rule (12). The Az values for the screening and conventional MR imaging studies were compared for each reader by using the Wilcoxon statistic (13). The questionnaire also included information about imaging findings, differential diagnosis, number of kidneys, specific renal abnormalities, and the technical quality of the imaging study.

Results of the independent readings were analyzed with the {kappa} statistic to measure the degree of interobserver agreement (14). Two-tailed P values were calculated, and a value of less than .05 was considered to indicate a statistically significant difference.


   RESULTS
TOP
 Abstract
 Introduction
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 References
 
Characteristics of the Study Population
The study included 101 patients suspected of having occult dysraphism. The mean age (±SD) was 4.9 years ± 5.2, and the age range was 1 day to 26 years. Forty-six (46%) patients were aged younger than two years. There were 36 male (36%) and 65 female (64%) patients.

The case group included 60 (59%) patients with a broad spectrum of occult myelodysplastic lesions (Table 2), and the control group included 41 (40%) patients with no dysraphic disorders (Table 3). The 60 patients in the case group had a total of 68 lesions, because eight patients had two abnormalities each (Table 2). Seventeen patients had renal abnormalities, 15 (88%) of whom also had a dysraphic lesion (Table 2).

Protocol Analysis
Occult dysraphic myelodysplasias.—Table 5 shows the sensitivity and the specificity associated with the MR imaging protocols and the readers. The sensitivity of fast screening MR imaging for occult myelodysplasias was 98.5% (95% CI = 92.1%, 99.9%) for both readers. The sensitivity of conventional MR imaging was 97.1% (95% CI = 89.8%, 99.7%) for reader 1 (T.Y.P.) and 95.6% (95% CI = 87.6%, 99.1%) for reader 2 (P.D.B.). The specificity of fast screening MR imaging was 84.8% (95% CI = 68.1%, 94.9%) for reader 1 and 75.8% (95% CI = 57.7%, 88.9%) for reader 2. The specificity of conventional MR imaging was 90.9% (95% CI = 75.7%, 98.1%) for both readers.


View this table:
[in this window]
[in a new window]
  		TABLE 5. Sensitivity and Specificity for Fast Screening and Conventional MR Imaging in Patients with Occult Dysraphic Myelodysplasia
 
There were no indeterminate fast screening or conventional MR image readings. False-positive results obtained with the fast screening protocol included questionable thickened conus filum terminale, low conus medullaris, and scoliosis with questionable small syringomyelia; these findings were not confirmed with results from additional MR imaging sequences, follow-up MR imaging, or conventional radiography. False-negative results obtained with the fast screening protocol included images interpreted as normal in patients with documented scoliosis and low conus medullaris. False-positive results obtained with the conventional MR imaging protocol included questionable dural ectasia and low conus medullaris; these findings were not confirmed at follow-up MR imaging. False-negative results obtained with the conventional MR imaging protocol included images interpreted as normal in patients with documented low conus medullaris. None of the patients with false-positive or false-negative fast screening or conventional MR imaging results underwent surgery.

With regard to interobserver agreement for occult myelodysplasias, {kappa} values of 0.68 for fast screening MR images and 0.75 for conventional MR images were obtained.

Figures 1 and 2 show the ROC curves for the readers and MR imaging protocols. Table 6 shows that the Az values for both readers were greater than 0.9 for both MR imaging protocols. Comparison of Az values revealed no significant differences (reader 1, P = .83; reader 2, P = .24).



View larger version (13K):
[in this window]
[in a new window]
  		Figure 1. ROC curve for detection by reader 1 of occult dysraphism on conventional and fast screening MR images.

 


View larger version (14K):
[in this window]
[in a new window]
  		Figure 2. ROC curve for detection by reader 2 of occult dysraphism on conventional and fast screening MR images.

 

View this table:
[in this window]
[in a new window]
  		TABLE 6. Results of ROC Curve Analyses for Detection of Occult Dysraphia and Renal Abnormalities on Fast Screening and Conventional MR Images
 
Renal abnormalities.—Twelve fast screening MR imaging studies and four conventional MR imaging studies were interpreted as indeterminate because the kidneys could not be assessed adequately. Among the studies where the kidneys could be assessed (Table 7), the sensitivity of fast screening MR imaging was 61.5% (95% CI = 31.5%, 86.2%) for reader 1 and 76.9% (95% CI = 46.2%, 95.1%) for reader 2. The sensitivity of conventional MR imaging was 81.3% (95% CI = 54.4%, 96.0%) for reader 1 and 70.6% (95% CI = 44.0%, 89.7%) for reader 2. The specificity, including the 95% CI, was between 82.8% and 99.2% for both protocols and readers (Table 7).


View this table:
[in this window]
[in a new window]
  		TABLE 7. Sensitivity and Specificity for Fast Screening and Conventional MR Imaging in Patients with Renal Abnormalities
 
False-positive results obtained with both imaging protocols included cases of mild to moderate pelvicaliceal dilatation that were determined at renal ultrasonography (US) to be normal variants. False-negative results obtained with the fast screening protocol included studies interpreted as normal in patients who had ureteral dilatation, horseshoe kidney, or mild or moderate pelvicaliectasis at renal US and/or MR imaging performed with additional sequences. False-negative conventional MR imaging studies included images interpreted as normal in patients with mild pelvicaliectasis or horseshoe kidney.

The ROC curve analysis demonstrated Az values of 0.786–0.874 for both protocols and readers (Table 6). There were no significant differences between the ROC curves (reader 1, P = .28; reader 2, P = .29).

Time Analysis
The acquisition time for the conventional MR imaging protocol ranged from 15 minutes 27 seconds to 16 minutes 44 seconds (Table 4), and the total examination time (including patient positioning, study localization, and study acquisition) was 25–30 minutes. The acquisition time for the fast screening MR imaging protocol ranged from 8 minutes 11 seconds to 9 minutes 28 seconds (Table 4), and the total examination time was 18–20 minutes.


   DISCUSSION
TOP
 Abstract
 Introduction
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
References
 
The conditions in the 101 patients included in this study represented a wide spectrum of occult dysraphic myelodysplasias and associated renal anomalies (Table 2). For the outcome of dysraphic myelodysplasia, the sensitivity of fast screening MR imaging (98.5%) was similar to that of conventional MR imaging (95.6%–97.1%). The specificity of conventional MR imaging (90.9%) was greater than that of fast screening MR imaging (75.8%–84.8%). However, the ROC curve analysis demonstrated excellent diagnostic performance with both protocols, with Az values greater than 0.9 and no significant differences (P >= .24). None of the patients with false-positive or false-negative results obtained with either imaging protocol had a tethering lesion that necessitated surgery. Interobserver agreement was very good for the fast screening MR images ({kappa} = 0.68) and excellent for the conventional MR images ({kappa} = 0.75) (15).

For the outcome of renal abnormalities, the sensitivity of fast screening MR imaging (61.5%–76.9%) was less than that of conventional MR imaging (70.6%–81.3%). The specificities of conventional and fast screening MR imaging were the same (90.9%). Diagnostic performance, determined on the basis of ROC curve analyses, showed Az values of 0.786–0.874 for both imaging protocols. No significant differences were noted when the Az values were compared (P >= .28). However, 12 fast screening and four conventional MR imaging studies were read as indeterminate for renal abnormalities because of incomplete imaging.

The diagnostic performance of fast screening MR imaging for the detection of occult myelodysplasia may allow its use as a screening examination. The 5-mm sagittal spin-echo T1-weighted localizer images and the 3-mm coronal spin-echo T1-weighted images from the conventional protocol did not help improve the sensitivity of MR imaging for the detection of occult dysraphic myelodysplasias. However, the availability of these two additional sequences as part of the conventional MR imaging protocol resulted in an increase in the specificity of MR imaging and a decrease in the number of indeterminate renal evaluations.

For both protocols, the diagnostic performance (ie, Az value) for renal abnormalities was less than that for the dysraphic lesions. This result suggests that additional renal MR imaging sequences (ie, axial and coronal fast spin-echo T2-weighted imaging) should be performed in patients with occult dysraphism if commonly associated renal anomalies are to be adequately assessed. The additional MR imaging sequences may be more time efficient and cost-effective than a separate renal US examination. This consideration may be of particular importance in markets where capitation plays a major role.

In the event that an occult dysraphic lesion is identified on either the fast screening or the conventional lumbosacral MR images, neither study may provide all the pertinent information needed to fully characterize dysraphic myelodysplasia. Lumbosacral MR imaging studies in children should be monitored regardless of the protocol, so that preoperative sequences such as T2-weighted and fat-suppressed contrast material–enhanced T1-weighted MR imaging in patients with diastematomyelia and dermal sinus can be ordered appropriately. This is especially true in the child who is sedated. MR imaging of the entire spine may be necessary in patients with clinical findings above the lumbosacral level or with symptoms and signs that may be due to conditions other than dysraphic myelodysplasia despite a negative lumbosacral fast screening MR study.

By design, this case-control study involved a population with a high prevalence of occult dysraphism. Because sensitivity and specificity are characteristics of a test, they are independent of the prevalence of disease (11,16). This, therefore, allowed us to determine the sensitivity and specificity of spinal MR imaging protocols by using a broad case mix of occult dysraphic myelodysplasias.

Additional analytic studies are needed to confirm the findings we have described. Multicenter prospective randomized trials could be performed to compare different imaging modalities and to determine the best approach for the outcomes of occult dysraphic myelodysplasias and associated renal anomalies. These studies could include newer MR imaging techniques such as spinal cord motion sequences, the clinical effectiveness of which has not yet been fully determined (17). In addition, decision-analysis and cost-effectiveness studies are necessary to determine optimal imaging strategies in children and young adults suspected of having occult dysraphism.

In conclusion, recent changes in the field of health care have emphasized the importance of determining the exact diagnostic performance of different MR imaging protocols to ensure optimal patient care. Our results suggest that conventional three-plane lumbosacral MR imaging in children and young adults suspected of having occult dysraphism provides better information than fast screening two-plane MR imaging because of the higher specificity and interobserver agreement of the former.


   Footnotes
 
Abbreviation: ROC = receiver operating characteristic

Author contributions: Guarantors of integrity of entire study, L.S.M., P.D.B.; study concepts, L.S.M., M.A.O., D.Z.; study design, L.S.M., M.A.O.; definition of intellectual content, L.S.M.; literature research, L.S.M., M.A.O.; clinical studies, L.S.M., M.A.O., T.Y.P., P.D.B.; data acquisition, T.Y.P., P.D.B.; data analysis, L.S.M., M.A.O., D.Z., J.D.; statistical analysis, L.S.M., D.Z., J.D.; manuscript preparation, L.S.M., M.A.O., D.Z.; manuscript editing, L.S.M., D.Z., T.Y.P., P.D.B.; manuscript review, L.S.M.


   References
TOP
 Abstract
 Introduction
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 References
 

  1. Raghavan N, Barkovich AJ, Edwards M, Norman D. MRI imaging in the tethered spinal cord syndrome. AJR 1989; 152:843-852.[Abstract/Free Full Text]
  2. Brophy JD, Sutton LN, Zimmerman RA, Bury E, Schut L. Magnetic resonance imaging of lipomyelomeningocele and tethered cord. Neurosurgery 1989; 25:336-340.[Medline]
  3. Moufarrij NA, Palmer JM, Hahn JF, Weinstein MA. Correlation between magnetic resonance imaging and surgical findings in the tethered spinal cord. Neurosurgery 1989; 25:341-346.[Medline]
  4. Appignani BA, Jaramillo D, Barnes PD, Poussaint TY. Dysraphic myelodysplasias associated with urogenital and anorectal anomalies: prevalence and types seen with MR imaging. AJR 1994; 163:1199-1203.[Abstract]
  5. Barnes PD, Lester PD, Yamanashi WS, Prince JR. Magnetic resonance imaging in infants and children with spinal dysraphism. AJNR 1986; 7:465-472.
  6. Barkovich AJ, Edwards MSB, Cogen PH. MR evaluation of spinal dermal sinus tracts in children. AJR 1991; 156:791-797.
  7. Altman NR, Altman DH. MR imaging of spinal dysraphism. AJNR 1987; 8:533-538.[Abstract]
  8. Wilson DA, Prince JR. MR imaging determination of the location of the normal conus medullaris throughout childhood. AJR 1989; 152:1029-1032.[Abstract/Free Full Text]
  9. Begg CB, McNeil BJ. Assessment of radiologic tests: control of bias and other design considerations. Radiology 1988; 167:565-569.[Abstract/Free Full Text]
  10. Blyth CR. Approximate binomial confidence limits. J Am Stat Assoc 1996; 81:843-855.
  11. Metz CE. Basic principles of ROC analysis. Semin Nucl Med 1978; 7:283-298.
  12. Centor RN. The use of ROC curves and their analysis. Med Decis Making 1991; 11:102-106.
  13. Hanley JA, McNeil BJ. The meaning and use of the area under the receiver operating characteristic (ROC) curve. Radiology 1982; 143:29-36.[Abstract/Free Full Text]
  14. Agresti A. Categorical data analysis New York, NY: Wiley, 1990; 366-370.
  15. Rosner B. Fundamentals of biostatistics New York, NY: Duxbury, 1995; 426.
  16. Weinstein MC, Fineberg HV. Clinical decision analysis Philadelphia, Pa: Saunders, 1980; 84-91.
  17. Quencer RM, Post MJ, Hinks RS. Cine MR in the evaluation of normal and abnormal CSF flow: intracranial and intraspinal studies. Neuroradiology 1990; 32:371-391.[Medline]



This article has been cited by other articles:


Home page
 RadiologyHome page
A. J. Khanna, T. R. Yoon, M. A. Mont, D. S. Hungerford, and D. A. Bluemke
Femoral Head Osteonecrosis: Detection and Grading by Using a Rapid MR Imaging Protocol
Radiology, October 1, 2000; 217(1): 188 - 192.
[Abstract] [Full Text]

This Article
Right arrow Abstract Freely available
Right arrow Figures Only
Right arrow Full Text (PDF)
Right arrow Submit a response
Right arrow Alert me when this article is cited
Right arrow Alert me when eLetters are posted
Right arrow Alert me if a correction is posted
Right arrow Citation Map
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Download to citation manager
Right arrow reprints & permissions
Citing Articles
Right arrow Citing Articles via HighWire
Right arrow Citing Articles via Google Scholar
Google Scholar
Right arrow Articles by Medina, L. S.
Right arrow Articles by Barnes, P. D.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Medina, L. S.
Right arrow Articles by Barnes, P. D.

HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS
RADIOLOGY RADIOGRAPHICS RSNA JOURNALS ONLINE