Japanese Journal of Radiology

, Volume 32, Issue 4, pp 205–210

Spinal imaging features in Japanese patients with Marfan syndrome: a case-control study

Authors

  • Eri Hayashida
    • Diagnostic Imaging CenterSaiseikai Kumamoto Hospital
    • Diagnostic Radiology, Faculty of Life SciencesKumamoto University
    • Diagnostic Radiology, Faculty of Life SciencesKumamoto University
  • Akira Sasao
    • Diagnostic Imaging CenterSaiseikai Kumamoto Hospital
  • Tsuyoshi Yasuda
    • Diagnostic Imaging CenterSaiseikai Kumamoto Hospital
  • Toshinori Hirai
    • Diagnostic Radiology, Faculty of Life SciencesKumamoto University
  • Hideaki Yuki
    • Diagnostic Radiology, Faculty of Life SciencesKumamoto University
  • Seitaro Oda
    • Diagnostic Radiology, Faculty of Life SciencesKumamoto University
  • Joji Urata
    • Diagnostic RadiologyKumamoto City Hospital
  • Akihiko Arakawa
    • Diagnostic Imaging CenterSaiseikai Kumamoto Hospital
  • Yasuyuki Yamashita
    • Diagnostic Radiology, Faculty of Life SciencesKumamoto University
Original Article

DOI: 10.1007/s11604-014-0285-1

Cite this article as:
Hayashida, E., Utsunomiya, D., Sasao, A. et al. Jpn J Radiol (2014) 32: 205. doi:10.1007/s11604-014-0285-1
  • 109 Views

Abstract

Purpose

The purpose of our study was to evaluate the morphology of the lumbosacral spine, i.e. the dura and vertebral body shape, of Japanese patients with Marfan syndrome (MFS) by comparing it with sex- and age-matched controls.

Materials and methods

Spinal MR or CT images of 32 MFS patients and 32 controls were retrospectively reviewed. The anteroposterior dural sac diameter (DSD), anteroposterior vertebral body diameter (VBD), and vertebral body height (VBH) were measured from L1 to S1 levels and the dural sac ratio [DSR = (DSD/VBD)] and vertebral body aspect ratio [VAR = (VBH/VBD)] were calculated.

Results

At each level, mean DSD and DSR were significantly higher in MFS patients; VBD was not. The cutoff values for DSR to differentiate between MFS patients and the controls were 0.59, 0.46, 0.42, 0.45, 0.47, and 0.47 from the level of L1 to S1. At a sensitivity of 93.8 % and a specificity of 84.4 % the cutoff value at S1 was most diagnostic. In MFS patients VAR was significantly higher at L3 and L4.

Conclusion

Our cutoff value for DSR >0.47 at S1 may help to identify MFS in the Japanese population. A square-like appearance of the L3 and L4 vertebral bodies is a supplementary finding in MFS patients.

Keywords

Marfan syndromeDural ectasiaMagnetic resonanceComputed tomographyDural sac ratio

Introduction

Marfan syndrome (MFS) is an autosomal dominant inherited disorder of connective tissue. It is characterized by a highly variable manifestation of ocular, skeletal, cardiovascular, integumentary, pulmonary, and neurologic features [1, 2]. The incidence of MFS is one in 5,000–10,000 live births [3]. The life expectancy of MFS patients tends to be reduced primarily due to cardiovascular events secondary to progressive enlargement of the aortic root [4]. Major cardiovascular manifestations such as dilation and dissection of the ascending aorta were reported to be higher in Japanese MFS patients (98 and 24 %, respectively) [5]. Therefore, prophylactic medical therapy and surgical intervention are important [6] as is the early identification of MFS patients, especially for cardiologists and radiologists.

The diagnosis of MFS is based on major and minor features [1]. Dural ectasia, a ballooning or widening of the dural sac and a major criterion of MFS, is observed in 63–92 % of MFS patients [2, 711]. It is usually assessed on computed tomography (CT) and magnetic resonance (MR) images [1, 7, 9, 10, 12]. Oosterhof et al. [7] suggested that the index of the dural sac ratio (DSR), because of its 95 % sensitivity and 98 % specificity by applying the cutoff value (either 0.53 at L3 or 0.53 at S1 level), should be used for the diagnosis of dural ectasia in MFS patients. On the other hand, Weigang et al. [10] found that 94 % of patients with and 44 % without MFS fulfilled the criteria of dural ectasia when the Oosterhof cutoff values were used, and Akutsu et al. [5] reported that the method of Oosterhof et al. detected dural ectasia in only 20 % of Japanese patients with genetically proven MFS. We suggest that the criteria for dural ectasia must take into account differences in the basic physical features of Japanese and Western populations. In this study we compared the morphology of the lumbosacral spine, i.e. the dura and vertebral body, of Japanese MSF patients with sex- and age-matched controls.

Materials and methods

This retrospective review was approved by our institutional review board; informed patient consent was waived.

Study population

We enrolled 64 sex- and age-matched subjects, 32 MFS patients (20 men and 12 women, mean age 39.5 ± 11.8 years, range 20–75 years) and 32 control subjects (20 men and 12 women, mean age 39.9 ± 11.9 years, range 20–73 years). MFS was diagnosed based on the genetic and clinical criteria promulgated in the revised Ghent nosology [1]. The control subjects were chosen from a pool of patients in our radiology archives; for inclusion they had to be sex- and age-matched and free of spinal or connective tissue disease. None of our study population had lumbosacral transitional vertebrae. Of the 32 MFS patients, 5 presented with aortic dissection, 10 with annuloaortic ectasia, and 10 with both aortic dissection and annuloaortic ectasia. The other 7 patients had neither aortic dissection nor annuloaortic ectasia.

Data acquisition

MRI scans were acquired on a 1.5T unit (Excelart Vantage Power Plus Package; Toshiba, Tokyo, Japan). T2-weighted (TR/TE, 4000/120 ms) turbo spin-echo sequences were obtained in the sagittal plane; the section thickness was 4 mm and the matrix size was 320 × 416. The T2-weighted sequence was repeated in 5 angulated axial planes parallel to the 5 lumbar intervertebral discs.

CT scans were acquired on a multidetector CT scanner (Aquilion-16 or Aquilion-64; Toshiba, Japan). The data acquisition parameters were 5 mm detector collimation, 50 ms tube rotation time, 50 mAs tube current–time product, and 120 kVp tube voltage. For sagittal visualization of the lumbar spine, 3-mm-thick slices were contiguously reconstructed on a workstation (ZIO station NG1 Version 2.0.1.3; Ziosoft Inc., Tokyo, Japan).

Measurements and definitions

Two radiologists with 5 and 8 years of experience with spinal CT and MR imaging and blinded to all clinical information agreed on measurements on sagittal MR- or CT images by consensus. At levels L1 through S1 the dural sac diameter (DSD) at the midcorpus level, the vertebral body diameter (VBD) at the midcorpus- and superior endplate level, and the vertebral body height (VBH) at the posterior margin were recorded (Fig. 1). VBD and DSD at the midcorpus level were measured on the same line from L1 through S1. The dural sac ratio (DSR) and the vertebral body aspect ratio (VAR) were calculated using the equations:
$${\text{DSR = }}{{\text{DSD}} \mathord{\left/ {\vphantom {{\text{DSD}} {\text{VBD}}}} \right. \kern-0pt} {\text{VBD}}}\text{ }{\text{at the midcorpus level}}$$
$${\text{VAR = }}{{\text{VBH}} \mathord{\left/ {\vphantom {{\text{VBH}} {{\text{VBD at the}}\text{ }{\text{superior endplate level}}}}} \right. \kern-0pt} {{\text{VBD at the}}\text{ }{\text{superior endplate level}}}}$$
https://static-content.springer.com/image/art%3A10.1007%2Fs11604-014-0285-1/MediaObjects/11604_2014_285_Fig1_HTML.jpg
Fig. 1

Measurement of vertebral body diameter (VBD), height (VBH) and dural sac diameter (DSD). a DSR = DSD/VBD at the midcorpus level b VAR = VBH/VBD at the superior endplate level

We also visually assessed the presence of meningoceles of the nerve root sleeves.

Statistical analysis

Numerical data were expressed as the mean ± standard deviation. The Student t-test was used for continuous values to assess intergroup differences. Receiver operating characteristics (ROC) curves were evaluated to assess how DSR and VAR could be used to differentiate MFS patients from control subjects. Cutoff values for DSR and VAR for the diagnosis of MFS with a given sensitivity and specificity were derived from the ROC curve. Statistical analyses were achieved with commercial software (JMP9.0.2, SAS Institute, Cary, NC, USA). A p value of <0.05 was considered to indicate significant differences.

Results

DSR

The mean DSD and DSR values from L1 to S1 were significantly larger in the MFS patients than the controls; there was no significant difference in VBD (Table 1). DSD at S1 was larger than at L4 in 16 (50.0 %) of 32 MFS patients, and in 3 (9.4 %) of the 32 controls. Representative cases are shown in Figs. 2 and 3. ROC analysis showed that the area under the curve was 0.66, 0.70, 0.68, 0.73, 0.84, and 0.92, respectively, for L1 through S1 levels, and in MFS patients dural ectasia was primarily found at the S1 level (Fig. 4).
Table 1

Mean DSD, VBD, and DSR of the controls and MFS patients

Level

Controls

MFS

P value

DSD (mm)

 L1

13.8 ± 1.6

15.6 ± 1.6

<0.001*

 L2

13.2 ± 1.5

15.2 ± 1.8

<0.001*

 L3

12.6 ± 1.4

15.0 ± 2.5

<0.001*

 L4

12.7 ± 1.3

15.8 ± 2.9

<0.001*

 L5

12.7 ± 1.7

17.2 ± 3.4

<0.001*

 S1

9.7 ± 2.2

16.3 ± 4.5

<0.001*

VBD (mm)

 L1

29.5 ± 3.2

30.1 ± 3.8

0.42

 L2

30.9 ± 3.1

30.8 ± 4.2

0.93

 L3

31.8 ± 3.2

32.2 ± 4.0

0.66

 L4

32.3 ± 2.7

32.6 ± 3.7

0.63

 L5

31.4 ± 2.5

31.5 ± 3.4

0.99

 S1

24.1 ± 3.2

23.6 ± 4.5

0.59

DSR

 L1

0.47 ± 0.08

0.53 ± 0.10

0.007*

 L2

0.43 ± 0.07

0.51 ± 0.11

<0.001*

 L3

0.40 ± 0.06

0.48 ± 0.12

<0.001*

 L4

0.40 ± 0.06

0.49 ± 0.12

<0.001*

 L5

0.41 ± 0.06

0.55 ± 0.13

<0.001*

 S1

0.41 ± 0.09

0.73 ± 0.30

<0.001*

MFS Marfan syndrome, DSD dural sac diameter, VBD vertebral body diameter, DSR dural sac ratio

* Statistically significant

https://static-content.springer.com/image/art%3A10.1007%2Fs11604-014-0285-1/MediaObjects/11604_2014_285_Fig2_HTML.jpg
Fig. 2

Sagittal CT image of a 36-year-old woman (control). DSR was 0.42 and 0.28 at the level of L3 and S1, respectively

https://static-content.springer.com/image/art%3A10.1007%2Fs11604-014-0285-1/MediaObjects/11604_2014_285_Fig3_HTML.jpg
Fig. 3

Sagittal CT image of a 41-year-old female MFS patient. DSR was 0.58 and 1.13 at the level of L3 and S1, respectively. The shape of the vertebral bodies at L3 and L4 appeared “square-like”

https://static-content.springer.com/image/art%3A10.1007%2Fs11604-014-0285-1/MediaObjects/11604_2014_285_Fig4_HTML.gif
Fig. 4

Receiver operating characteristic curves for DSR at L1, L3, L5 and S1 levesl in MFS patients. For the S1 level, the area under the curve is 0.92, and cutoff value of 0.47 shows the highest diagnostic ability (sensitivity 93.8 % and specificity 84.4 %)

With the method of Oosterhof et al. [7] the sensitivity and specificity for the diagnosis of MFS at the level of S1 in our patients was 59.4 % (19/32) and 93.8 % (30/32), respectively; at the L3 level they were 43.8 % (14/32) and 81.3 % (26/32). Application at the S1 level of the cutoff value suggested by our ROC analysis (0.47) improved diagnostic capability (sensitivity = 93.8 %, sensitivity = 84.4 %). Table 2 shows our cutoff values at each spinal level and their sensitivity and specificity for dural ectasia as a marker for MFS. Meningocele was observed in 3 (9.4 %) of 32 MFS patients and none of the controls.
Table 2

Cutoff values for DSR as a marker for MFS at levels L1 to S1

Level

Cutoff value

Sensitivity (%)

Specificity (%)

L1

0.59

31.3

96.9

L2

0.46

65.6

71.9

L3

0.42

65.6

68.8

L4

0.45

59.4

84.4

L5

0.47

68.8

84.4

S1

0.47

93.8

84.4

When we applied our cutoff value at the S1 level, sensitivity and specificity for aortic dissection were 93.3 % (14/15) and 42.9 % (21/49), respectively; for annuloaortic ectasia they were 95 % (19/20) and 63.6 % (28/44).

VAR

Mean VBH from L1 to L4 was significantly higher in MFS patients than the controls (Table 3). Calculation of the VAR value for each level from L1 to S1 showed significant differences between MSF patients and the controls only at L3 and L4 (Table 3).
Table 3

Mean VBH and VAR for MFS patients and controls

Level

Controls

MFS

P value

VBH (mm)

 L1

28.0 ± 2.0

30.6 ± 3.1

<0.001*

 L2

28.9 ± 2.1

31.3 ± 3.4

<0.0013*

 L3

28.5 ± 2.0

31.1 ± 3.5

<0.001*

 L4

27.2 ± 2.5

30.0 ± 3.2

<0.001*

 L5

25.5 ± 2.6

25.2 ± 5.1

0.78

 S1

26.1 ± 2.3

26.1 ± 3.5

0.92

VAR

 L1

0.90 ± 0.08

0.95 ± 0.10

0.062

 L2

0.88 ± 0.07

0.93 ± 0.10

0.066

 L3

0.84 ± 0.07

0.89 ± 0.10

0.027*

 L4

0.80 ± 0.08

0.85 ± 0.10

0.018*

 L5

0.75 ± 0.08

0.72 ± 0.16

0.43

 S1

0.82 ± 0.09

0.77 ± 0.13

0.076

MFS Marfan syndrome, VBH vertebral body height, VAR vertebral body aspect ratio

* Statistically significant

Discussion

Mutation of the fibrillin-1 gene encoding for the fibrillin I protein causes disorder of connective tissue in MFS patients [13]. The defect in this protein elicits MFS features such as cardiovascular system pathology and skeletal, ocular, and central nervous system manifestations. The diagnosis of MFS has been based on the Ghent nosology [1]. Dural ectasia is one of the most important imaging findings in MFS patients [5]. Dural ectasia is considered to arise as hydrostatic cerebrospinal fluid pressure on constitutionally weak dura slowly balloons them outwardly and erodes the surrounding bone [14]. Although there are several methods for detecting dural ectasia [710, 15] there is no standardized quantification method. The Oosterhof cutoff values have been used most widely and are considered most sensitive for the diagnosis of dural ectasia in MFS patients [7, 10]. According to Oosterhof et al. [7], for a diagnosis of MFS in adults, the DSR at L1 through S1 is >0.64, 0.55, 0.47, 0.48, 0.48, and 0.57. However, we found that the adoption of their cutoff values resulted in inadequate sensitivity (59.4 % at S1 and 43.8 % at L3) in Japanese adult patients and our cutoff values were 0.59, 0.46, 0.42, 0.45, and 0.47 at levels L1 through S1. Habermann et al. [16] suggested that some of the differences in the cutoff values were secondary to age differences. Our findings suggest that the assessment of dural ectasia must take into account not only the age but also racial background. Previous studies, including Korean, Singaporean, and Japanese MFS populations showed racial differences in clinical features compared with Caucasian MFS populations [5, 17, 18]. Franken et al. [17] suggested that a more severely affected cardiovascular system in Asian populations might be related to under diagnosis of MFS by the specific diagnostic criteria for Caucasians. We think that the diagnostic criteria should be modified for Asian populations for accurate diagnosis of MFS.

Ahn et al. [9, 19] reported greater widening of the dural sac at the level of S1 than L4 as one of the major imaging findings of dural ectasia. We obtained similar results; in our patients DSR was significantly wider at S1, suggesting that hydrostatic pressure plays a role in the pathogenesis of dural ectasia [20]. The 0.47 cutoff value at S1 exhibited the best diagnostic ability (sensitivity of 94 %, specificity of 84 %). Based on our cutoff value, only 2 of our 32 MFS patients (6.3 %) did not show dural ectasia. Our study showed that evaluating dural ectasia in the lower lumbar and sacral regions helps to establish the clinical diagnosis in patients with suspected MFS.

VBH from L1 to L4 was significantly higher in our MFS patients than in the controls although the intergroup difference in VAR was significant only at L3 and L4. MFS patients are generally taller than average, and we posit that higher VBH/VAR might cause this height in MFS patients. However, Lundby et al. [21] reported that there were a few significant differences in VBHs between Caucasian MFS and non-MFS patients. The Japanese are generally smaller than Caucasians, and the difference of VBH might be significant between Japanese MFS and non-MFS patients. We believe that the “square-like” appearance at L3 and L4 on sagittal images may be a supplementary finding for the diagnosis of MFS in Japanese patients, and we are in the process of confirming our hypothesis in larger study populations. While meningocele/herniation of the nerve root sleeves may be another supplementary finding for dural ectasia, only 3 of our 32 MFS patients (9.3 %) harbored meningocele, suggesting that this finding is not so common in Japanese patients.

Our study has some limitations. First, the design of our single-center study was retrospective and the study population was relatively small. Multicenter prospective studies are needed. Second, we evaluated the presence of dural ectasia on CT or MRI scans. Ahn et al. [9, 19] evaluated both CT and MR images. Although we think that measurement differences between CT and MRI studies are small, we recommend that both imaging modalities should be applied. Third, we evaluated only the Oosterhof method [7] which, according to Weigang et al. [10] exhibited the highest sensitivity for the diagnosis of MFS. We plan to conduct studies to assess the usefulness of the methods of Ahn et al. [9] and Villeirs et al. [8] in Asian populations.

In conclusion, the incidence of dural ectasia in MFS patients may vary among different races. Therefore, we suggest that the cutoff value should be modified based on the race of the patient and that our cutoff value of DSR >0.47 at S1 is suitable for the diagnosis of MFS in the Japanese population. Also, a square-like appearance of the L3 and L4 vertebral bodies is a supplementary finding in Japanese MFS patients.

Conflict of interest

None.

Copyright information

© Japan Radiological Society 2014