Optimal Filum Terminale Thickness Cutoff Value on Sonography for Lipoma Screening in Young Children

Transcription

1 ORIGINAL RESEARCH Optimal Filum Terminale Thickness Cutoff Value on Sonography for Lipoma Screening in Young Children Hyun Joo Shin, MD, Myung-Joon Kim, MD, PhD, Hye Sun Lee, MS, Hyun Gi Kim, MD, PhD, Mi-Jung Lee, MD, PhD Objectives The purpose of this study was to evaluate the normal thickness of the filum terminale on sonography and suggest an optimal cutoff value for filum terminale lipoma screening in young children. Methods We retrospectively reviewed lumbosacral sonograms and magnetic resonance images from children younger than 36 months that were obtained between January 2013 and June The filum terminale thickness on sonography and the presence of fat in the filum terminale on magnetic resonance imaging were evaluated. Results From 111 children (mean age ± SD, 3.6 ± 3.0 months), 49 did not have abnormal lesions (normal group), and 62 had fat infiltration in the filum terminale (lipoma group). The filum terminale was thicker in the lipoma group than the normal group (1.5 ± 0.5 versus 0.9 ± 0.2 mm; P <.001). Filum terminale thickness also showed significance in a multivariable analysis with sex and age (odds ratio per 0.1-mm unit, 2.754; P <.001) and in propensity score matching for age (P <.001). The optimal cutoff value for filum terminale lipoma screening was 1.1 mm, with 94% sensitivity and 86% specificity. Conclusions The conventional cutoff value of 2 mm for a thickened filum terminale on sonography can be too thick. We suggest an optimal cutoff value of 1.1 mm for lipoma screening in young children. Key Words children; filum terminale; lipoma; magnetic resonance imaging; pediatric ultrasound; sonography Received October 29, 2014, from the Department of Radiology and Research Institute of Radiological Science, Severance Children s Hospital (H.J.S., M.-J.K., H.G.K., M.-J.L.), and Biostatistics Collaboration Unit (H.S.L.), Yonsei University College of Medicine, Seoul, Korea. Revision requested December 7, Revised manuscript accepted for publication February 3, Address correspondence to Mi-Jung Lee, MD, PhD, Department of Radiology and Research Institute of Radiological Science, Severance Children s Hospital, Yonsei University College of Medicine, 50-1Yonsei-ro, Seodaemun-gu, Seoul , Korea. mjl1213@yuhs.ac Abbreviations CI, confidence interval; MRI, magnetic resonance imaging; OR, odds ratio doi: /ultra Occult spinal dysraphism is a category of the disease that describes spinal dysraphism without a visible mass in the lumbosacral area in infants. It is more common than other types of spinal dysraphism that show a visible buttock mass regardless of skin coverage. 1 The term occult spinal dysraphism encompasses diagnoses such as a tight filum terminale, a dorsal dermal sinus, diastematomyelia, caudal regression syndrome, and spinal lipoma. 2 Of these, spinal lipoma is the most common type of occult spinal dysraphism, and filum terminale lipomas account for 12% of all spinal lipomas. 3 An accurate diagnosis of occult spinal dysraphism using imaging studies is important, since there are no visible masses or characteristic symptoms in young infants, and irreversible neurologic damage can occur as a result of delayed diagnosis. 1 Previous reports have suggested that most cases of a thickened filum terminale were due to the presence of a fibrofatty mass. 1 Although the clinical implications 2015 by the American Institute of Ultrasound in Medicine

2 of filum terminale lipoma in adults with respect to neurologic deficits are still controversial, many previous reports have suggested that an isolated filum terminale lipoma could be responsible for cord tethering and neurologic deficits in pediatric patients. 1,4,5 Therefore, detection of cord tethering including filum terminale lipoma is important, especially for screening purposes in pediatric patients. There is a possibility of developing neurologic sequelae from cord tethering in young infants, and the surgical procedure of detethering or removal of a lipoma is not complicated, compared with the risk of developing lifelong neurologic deficits. Sonography of the lumbosacral area is commonly performed in infants with sacral skin dimpling, which can be a skin marker of occult spinal dysraphism. 6,7 Sonography is a safe and easy diagnostic method compared with magnetic resonance imaging (MRI), especially when used for screening purposes in infants. In the literature to date, the cutoff value for a thickened filum terminale on sonography is 2 mm, with a normal range of 0.5 to 2 mm. 3,7 11 However, we have encountered infants with a diagnosis of filum terminale lipoma who had filum terminale thickness of less than 2 mm on lumbosacral sonography in clinical practice. Therefore, the purpose of this study was to evaluate the normal thickness of the filum terminale in young children on sonography and suggest an optimal cutoff value for thickened filum terminale screening to better identify patients with filum terminale lipoma. Materials and Methods Patients The Institutional Review Board at our hospital approved this retrospective study and waived the requirement of informed consent. We retrospectively reviewed the medical records of children younger than 36 months who underwent lumbar spine MRI between January 2013 and June 2014 in the Department of Neurosurgery at our tertiary care hospital. The reasons for performing imaging studies of the lumbosacral area were also reviewed. We excluded children who did not undergo lumbosacral sonography before MRI, as well as those with lesions other than filum terminale lipoma on MRI. Sex and age at the time of sonography and MRI were recorded. Sonography and MRI Sonography was performed with a 7 12-MHz linear transducer (iu22; Philips Healthcare, Bothell, WA). Two boardcertificated radiologists (M.-J.K. and M.-J.L.) with more than 20 and 10 years of experience in pediatric radiology, respectively, performed sonographic examinations of the lumbosacral area in the left decubitus position, including sagittal and axial views of the lumbosacral spinal canal in each infant. Filum terminale thickness was measured on sonography at the most clearly visualized area on sagittal or axial views, since the amount of fat infiltration in the filum terminale varied in its extent and location. Two radiologists (H.J.S. and M.-J.L.) reviewed the sonograms retrospectively again to confirm the filum terminale measurements. If there were repeated sonographic studies before MRI, the measurement only on the initial sonographic study was used for the statistical analysis. The echogenicity of the thickened filum terminale on sonography was also subjectively reviewed and compared with the adjacent cauda equina. Magnetic resonance imaging was performed at the discretion of the pediatric neurosurgeon either on an inpatient or outpatient basis. Lumbosacral spinal MRI was performed with a 1.5- or 3.0-T system (Intera Achieva; Philips Healthcare, Best, the Netherlands), including T1-weighted sagittal and axial images using the following parameters: repetition time, 647 milliseconds; echo time, 10 milliseconds; flip angle, 90 ; section thickness, 3 mm; matrix size, ; and field of view, 100 cm. The presence of fat in the filum terminale was diagnosed on T1-weighted MRI when high signal intensity was noted in the filum terminale, as described in previous studies. 4,12 Two radiologists (H.J.S. and M.-J.L.) with 10 and 4 years of experience in pediatric radiology retrospectively reviewed the medical records and imaging findings, including filum terminale thickness on sonography and the presence of filum terminale lipoma on MRI of the spine. Statistical Analyses Statistical analyses were performed with SAS version 9.2 software (SAS Inc, Cary, NC). The children were divided into a normal group and a lipoma group based on MRI findings for comparison of filum terminale thickness on sonography. A χ 2 test was used to compare the sexes between the groups. Student t tests were used to compare the age at the time of sonography and the filum terminale thickness between groups. Multiple logistic regression and propensity score matching using a Greedy 8-to-1 digit-matching algorithm were performed to compare filum terminale thickness with sex and age adjustments between the groups. A receiver operating characteristic curve analysis was performed before and after matching to obtain an optimal filum terminale thickness cutoff value for filum terminale lipoma screening on sonography. P <.05 was considered statistically significant. 1944

3 Results Patients During the study period, a total of 378 MRI examinations in 304 children were performed. The reasons for performing lumbar spine MRI in these children included evaluation of sacral skin dimpling in 197 children (65%), a back area mass in 46 (15%), an anorectal malformation in 40 (13%), postoperative follow-up in 14 (5%), and neurologic symptoms or abnormal radiographic findings in 7 (2%). Among the 304 children, 181 did not undergo sonography before MRI, and these children were excluded. Additionally, 12 of the remaining 123 children were excluded because they were found to have other masslike lumbo - sacral lesions such as lipomyelomeningocele (n = 9), lipomeningocele (n = 2), and sacrococcygeal teratoma (n = 1) on MRI. Therefore, a total of 111 children were included in this study. There were two reasons for performing lumbosacral sonography in these children: evaluation of sacral skin dimpling (97 [87%]) and screening in children with an anorectal malformation (14 [13%]). There were 64 boys and 47 girls in the final study population. The mean age ± SD at the time of sonography was 3.6 ± 3.0 months (range, 1 day 15 months). The interval between lumbosacral sonography and MRI ranged from 1 day to 21 months (mean, 4.3 ± 5.2 months). On MRI, 49 children had no evidence of lesions (normal group), and 62 children had fat infiltration in the filum terminale (lipoma group). In the lipoma group, 44 of 62 children underwent surgical detethering of the spinal cord during the study period and were pathologically confirmed to have filum terminale lipoma. There were 24 boys in the normal group and 40 boys in the lipoma group, without a sex difference (P= 0.100; Table 1). The mean age at the time of sonography was not different between the normal group (3.2 ± 2.8 months) and the lipoma group (3.8 ± 3.2 months; P = 0.324). Filum Terminale Thickness on Sonography The mean filum terminale thickness on sonography in the normal group was 0.9 ± 0.2 mm (range, of mm). The mean filum terminale thickness in the lipoma group (1.5 ± 0.5 mm; range, mm) was greater than that of the normal group (P <.001) before sex and age matching (Table 1 and Figure 1). In the lipoma group, the filum terminale thickness in the patients who underwent surgery (n = 44) was 1.6 ± 0.5 mm (range, mm), and 38 of 44 patients had thickness of less than 2 mm. The mean filum terminale thickness in the patients who did not undergo surgery (n = 18) was 1.4 ± 0.4 mm (range, mm), without a difference from that in the patients who underwent surgery (P=.193). Six of 111 children had follow-up sonography before MRI because of a filar cyst or a low-lying conus medullaris end level at L3 without another lesion. Among these children, the filum terminal thickness was unchanged in 2 and increased by 0.1 to 0.3 mm in 4 children during 1 to 6 months of follow-up (Figure 2). In the lipoma group, the echogenicity the of filum terminale was hyperechoic compared with the cauda equina in 60 of 62 children (97%). The other 2 were isoechoic compared with the cauda equina, with filum terminale thicknesses of 1.2 and 1.4 mm. In the multivariable analysis considering age, sex, and filum terminale thickness, the odds ratio (OR) for filum terminale thickness per 0.1-mm unit was (95% confidence interval [CI], ; P <.001). However, the ORs for age (1.570; 95% CI, ; P =.337) and male children (1.969; 95% CI, ; P =.295) were not significant. For propensity score matching, 38 children were selected in each group, with 20 boys and 18 girls. The mean age at the time of sonography was 3.0 ± 2.4 months for both groups. In this analysis, the mean filum terminale thickness in the lipoma group (1.4 ± 0.4 mm) was still greater than that in the normal group (0.9 ± 0.2 mm; P <.001; Table 1). Table 1. Clinical and Sonographic Comparisons Between the Normal and Lipoma Groups Before and After Age and Sex Matching Before Matching After Matching Normal Lipoma Normal Lipoma Characteristic (n = 49) (n = 62) P (n = 38) (n = 38) P Male/female 24/25 40/ /18 20/18 >.999 Age at sonography, mo 3.2 ± ± ± ± 2.4 >.999 Propensity score 0.5 ± ± ± ± 0.1 >.999 Filum terminale thickness, mm 0.9 ± ± 0.5 < ± ± 0.4 <.001 Data are presented as mean ± SD were applicable. 1945

4 In the lipoma group, 53 of 62 children (85%) had filum terminale thickness of less than 2 mm on sonography. On the receiver operating characteristic analysis without age and sex matching, the optimal cutoff value was 1.1 mm, with an area under the curve of (95% CI, ), 94% sensitivity, and 86% specificity. After age and sex matching, the receiver operating characteristic analysis also showed an optimal cutoff value of 1.1 mm, with an area under the curve of (95% CI, ), 92% sensitivity, and 84% specificity. Among the 65 children who had filum terminale thickness of greater than 1.1 mm on sonography, 58 (89%) had fat infiltration in the filum terminale on MRI. Among the 46 children with thickness of less than 1.1 mm on sonography, only 4 children (9%) had fat infiltration in the filum terminale on MRI. Discussion Fat accounts for greater than 90% of filum terminale thickening. 9 It is thought that an abnormality in the secondary neurulation process can lead to lipomatous changes within the filum terminale. 4,9,13 Filum terminale lipoma can result in a tethered cord, which can lead to traction of the spinal cord. 2 This condition can distort small vessels and nerve fibers, causing impairment of the microcirculation, electrophysiologic alterations, and cell membrane deformity, and result in metabolic or neurologic impairment, which can present as motor or sensory dysfunction, pain, urologic or sphincter dysfunction, abnormal reflexes, gait disturbances, scoliosis, or foot and hip deformities that are irreversible. 1,9,11,14,15 Figure 1. Images from a 2-month-old boy in the normal group. A, On lumbosacral sonography, the filum terminale (arrow) is well visualized as a tubular structure in the lumbosacral canal. The thickness of the filum terminale was 0.9 mm. B, On lumbosacral MRI, there is no high signal intensity indicative of fat in the filum terminale (arrow) on a sagittal T1-weighted image. 1946

5 The actual clinical impact of filum terminale lipoma on neurologic impairment remains controversial, however, especially in asymptomatic adults. Nevertheless, the longterm follow-up of clinical outcomes in patients with filum terminale lipoma is limited for ethical reasons, and the treatment can vary from observation to early dethetering according to the institution. 16 Therefore, to our knowledge, no investigation has clarified the definitive treatment strategy for patients with filum terminale lipoma. In addition, previous reports have suggested that the risk of developing neurologic deficits was high in pediatric patients compared with adults presenting with filum terminale lipoma. 4,5,9 The treatment for this condition, detethering, is not a technically complicated operation and can prevent the risk of irreversible neurologic damage by facilitating neuronal reparative mechanisms and metabolic changes that occur due to stretching of the filum terminale in growing children. 11 Therefore, we suggest that detection of abnormal filum terminale thickening is important, especially for screening purposes in children. For an accurate diagnosis of filum terminale lipoma, which is the most common cause of filum terminale thickening, it is crucial to know the optimal thickness cutoff value on lumbosacral sonography. However, in clinical practice, the generally accepted cutoff value of 2 mm may be too thick for filum terminal lipoma screening in young children. In our study, 53 children in the lipoma group (85%) were found to have fat infiltration in the filum terminale on MRI, with a thickness of less than 2 mm on sonography. The difference in cutoff values is likely due to advances in sonographic technology, which now features high spatial resolution that can detect submillimeter differences. To the best of our knowledge, no recent studies have defined a new cutoff value for filum terminale thickness on sonography. Moreover, our results showed that lumbosacral sonography in infants had good diagnostic value as a screening tool in young children with sacral dimpling. Therefore, sonography is a good first-line diagnostic approach, given its ease of use, accuracy, and lack of both radiation exposure and the need for sedation. Figure 2. Images from a 2-month-old boy in the lipoma group. A, On initial lumbosacral sonography, the thickness of the filum terminale (arrow) was 1.1 mm. B, Follow-up sonography was performed after 4 months, and the thickness of filum terminale (arrow) had increased to 1.5 mm (continued). 1947

6 In this study, although we included children younger than 36 months who underwent MRI of the spine, the oldest patient who underwent prior lumbosacral sonography was 15 months. This finding may have occurred because ossification of the vertebral arch progresses during the first year of life, and sonography can be limited after this age, as demonstrated in previous studies. 1,17 Additionally, the filum terminale thickness was measured on the sagittal or axial image in which it was most clearly visible and was not limited to the L5 S1 level or axial view only, since the sizes and locations of filum terminale lipomas vary. There could be debates concerning the risk of developing higher false-positive results when using a new cutoff value of 1.1 mm for filum terminale lipoma screening on sonography. In addition, prophylactic management of fat infiltration in the filum terminale in asymptomatic patients can be controversial. However, Xenos et al 5 suggested that prophylactic surgery could provide some protection from future neurologic deterioration, and most neurosurgeons Figure 2. (continued) C, Lumbosacral MRI was performed after the second sonographic examination and confirmed the presence of fat in the filum terminale (arrow) on a sagittal T1-weighted image. have been operating on spinal lipomas even in asymptomatic children. Our study has an implication in this regard because we first demonstrated a considerable falsenegative rate (85%) for fat infiltration screening in the filum terminale when using the traditional cutoff value of 2 mm. Selcuki et al 18 reported that the a radiologically normal filum terminale did not indicate a histopathologically normal finding and could result in urologic incontinence. They investigated patients with urinary incontinence who had filum terminale thickness of less than 2 mm and demonstrated that the filum terminale was composed of randomly distributed adipose tissues in densely compacted collagen fibers. We suggest the need to pay attention to children with filum terminale thickness of greater than 1.1 mm, not 2 mm, using recent sonography technology, and more careful clinical considerations such as follow-up sonography or further evaluations are needed for screening purposes in young children. Our study had several limitations. First, because our institution is a tertiary hospital, and MRI examinations of the spine were performed in the setting of suspected abnormalities, the children included in the study likely had a higher prevalence of filum terminale lipoma than the overall population, thereby resulting in a selection bias. Second, because of our retrospective study design, the timing of the sonographic and MRI examinations was different in each patient. Additionally, the number of children included in the study was relatively small. Moreover, we were unable to correlate neurologic results from these children, and pathologic data were not included as a standard reference, since not all of the children who had MRI underwent surgery, which is typically performed at the end of the first year of life. Therefore, we defined a filum terminale lipoma visible on MRI as a standard reference according to previous reports. 4,12 However, the sole presence of fat in the filum terminale may not always mean cord tethering and cause neurologic deficits. Further studies to discriminate meaningful lipoma from simple fat infiltration in the filum terminale are needed. In conclusion, the filum terminale thickness values on lumbosacral sonography in young children were 0.9 mm in the normal group and 1.5 mm in the lipoma group. Overall, 85% of the children in the lipoma group had filum terminale thickness of less than 2 mm on sonography, and the optimal cutoff value for filum terminale lipoma screening was 1.1 mm, with an OR per 0.1-mm unit of in this study. Therefore, we suggest that more careful clinical considerations are needed for children with filum terminale thickness of greater than 1.1 mm on sonography for screening purposes. 1948

7 References 1. Lode HM, Deeg KH, Krauss J. Spinal sonography in infants with cutaneous birth markers in the lumbo-sacral region: an important sign of occult spinal dysrhaphism and tethered cord. Ultraschall Med 2008; 29(suppl 5): Deeg KH, Lode HM, Gassner I. Spinal sonography in newborns and infants, part II: spinal dysraphism and tethered cord. Ultraschall Med2008; 29: Unsinn KM, Geley T, Freund MC, Gassner I. US of the spinal cord in newborns: spectrum of normal findings, variants, congenital anomalies, and acquired diseases. Radiographics 2000; 20: Al-Omari MH, Eloqayli HM, Qudseih HM, Al-Shinag MK. Isolated lipoma of filum terminale in adults: MRI findings and clinical correlation. J Med Imaging Radiat Oncol 2011; 55: Xenos C, Sgouros S, Walsh R, Hockley A. Spinal lipomas in children. Pediatr Neurosurg 2000; 32: Lowe LH, Johanek AJ, Moore CW. Sonography of the neonatal spine: part 1, normal anatomy, imaging pitfalls, and variations that may simulate disorders. AJR Am J Roentgenol 2007; 188: Lowe LH, Johanek AJ, Moore CW. Sonography of the neonatal spine: part 2, spinal disorders. AJR Am J Roentgenol 2007; 188: Korsvik HE, Keller MS. Sonography of occult dysraphism in neonates and infants with MR imaging correlation. Radiographics 1992; 12: Bulsara KR, Zomorodi AR, Enterline DS, George TM. The value of magnetic resonance imaging in the evaluation of fatty filum terminale. Neurosurgery 2004; 54: Yundt KD, Park TS, Kaufman BA. Normal diameter of filum terminale in children: in vivo measurement. Pediatr Neurosurg 1997; 27: Bao N, Chen ZH, Gu S, Chen QM, Jin HM, Shi CR. Tight filum terminale syndrome in children: analysis based on positioning of the conus and absence or presence of lumbosacral lipoma. Childs Nerv Syst 2007; 23: Uchino A, Mori T, Ohno M. Thickened fatty filum terminale: MR imaging. Neuroradiology 1991; 33: Brown E, Matthes JC, Bazan C III, Jinkins JR. Prevalence of incidental intraspinal lipoma of the lumbosacral spine as determined by MRI. Spine (Phila Pa 1976) 1994; 19: Tortori-Donati P, Rossi A, Cama A. Spinal dysraphism: a review of neuroradiological features with embryological correlations and proposal for a new classification. Neuroradiology 2000; 42: Cornette L, Verpoorten C, Lagae L, et al. Tethered cord syndrome in occult spinal dysraphism: timing and outcome of surgical release. Neurology 1998; 50: Kucera JN, Coley I, O Hara S, Kosnik EJ, Coley BD. The simple sacral dimple: diagnostic yield of ultrasound in neonates. Pediatr Radiol 2015; 45: Deeg KH, Lode HM, Gassner I. Spinal sonography in newborns and infants, part I: method, normal anatomy and indications. Ultraschall Med 2007; 28: Selcuki M, Vatansever S, Inan S, Erdemli E, Bagdatoglu C, Polat A. Is a filum terminale with a normal appearance really normal? Childs Nerv Syst 2003; 19: