Magnetic resonance imaging phantom-based S1 vertebral scores are indicators of fat–water-like osteoporotic changes in postmenopausal women: a pilot study
Article information
Abstract
Study Design
A prospective study.
Purpose
To assess fat–water-like tissue changes on the 1st sacral vertebra using novel magnetic resonance imaging (MRI) phantom-based F- and W-scores and evaluate their diagnostic performances in osteoporosis detection.
Overview of Literature
Using an uncommonly advanced MRI technique, previous studies have found that fat–water changes were consistent with osteoporosis. The role of routine MRI sequences can be extended in this regard. The S1 vertebra is considered a crucial anatomical site in spine surgeries because it seldom suffers from fractures. Thus, S1 could indicate osteoporotic fat–water changes.
Methods
Forty-two female volunteers (aged 62.3±6.3 years) underwent spine examination with both MRI (including a phantom) and dual-energy X-ray absorptiometry (DXA) following ethical approval. MRI phantom-based F- and W-scoreS1 were defined by normalizing S1 vertebral signal intensities (SIs) by coconut oil and water SIs of the phantom on T1- and T2-weighted imaging, respectively. Using receiver operating characteristic analysis, the diagnostic performances of the new scores for evaluating osteoporosis and vertebral fractures were investigated against standard areal bone mineral density measured with DXA (DXA-aBMD).
Results
The F-scoreS1 and W-scoreS1 were greater (4.11 and 2.43, respectively) in patients with osteoporosis than those without osteoporosis (3.25 and 1.92, respectively) and achieved areas under the curve (AUCs) of 0.82 and 0.76 (p<0.05), respectively, for osteoporosis detection. Similarly, the mean F-scoreS1 and W-scoreS1 were higher (4.11 and 2.63, respectively) in patients with vertebral fractures than in those without fractures (3.30 and 1.82, respectively) and had greater AUCs (0.90 for W-scoreS1 and 0.74 for F-scoreS1) than DXA-aBMD (AUC, 0.26; p<0.03). In addition, the F- and W-scoreS1 demonstrated a strong correlation (r=0.65, p<0.001).
Conclusions
The new S1 vertebral-based MRI scores were developed to detect osteoporotic changes and demonstrated improvements over DXA-aBMD in differentiating patients with vertebral fractures.
Introduction
The first sacral vertebra (S1) is of considerable clinical importance, particularly in lumbar spine surgery [1]. The bone mineral density (BMD) of S1 determines the success or failure of many spinal surgeries [2]. A previous study highlighted the importance of BMD assessment before any spinal surgery [3]. Osteoporosis, a skeletal disease characterized by low bone mass and reduced bone strength, is the most common cause of low BMD [4]. Globally, approximately 200 million women have osteoporosis [5]. A higher risk of vertebral fractures is the main problem associated with osteoporosis [6], which has increased globally by 38% between 1990 and 2019 [7]. The prevalence is increasing with the aging of the global population. Consequently, identifying patients with osteoporosis may help improve surgical outcomes and prevent associated problems.
Dual-energy X-ray absorptiometry (DXA), the gold standard imaging for osteoporosis, measures areal BMD (aBMD) based on a T-score of <−2.5 at the hip or L1–L4 vertebrae [8]. However, 49.5% of patients who had previously sustained a spinal fracture had normal aBMD [9]. Moreover, misdiagnosis may result from an increase in aBMD caused by degenerative structures [10]. Therefore, researchers have attempted to develop alternative opportunistic approaches for osteoporosis screening. With the 10% annual growth of referral to magnetic resonance imaging (MRI), it is frequently used for spinal problems including low back pain [11]. A previous study indicated inverse relationships between BMD and increased signal intensity (SI) on T1-weighted imaging caused by increased fatty bone marrow content with aging or osteoporosis [12].
The vertebral bone quality (VBQ) score was developed to detect osteoporosis using T1-weighted images alone [13]. Huang et al. [2] introduced a modified VBQ score when the assessment of BMD becomes challenging because of L1–L4 deformities. S1 VBQ was found to be a promising indicator for distinguishing patients with poor bone quality and degenerative lumbar disease and those with adolescent idiopathic scoliosis [2,14]. However, the VBQ score uses cerebrospinal fluid (CSF) as a control. Studies have suggested that it is affected by physiological and pathological factors [13,15]. Moreover, fat–water changes were found to be associated with osteoporosis [16]. Thus, to detect fat–water-like tissue changes, a new approach could be employed using MRI phantom with constant external reference controls. In this study, we aimed to develop MRI phantom-based S1 vertebral scores (F-scoreS1 and W-scoreS1) in routine T1- and T2-weighted imaging (T1WI and T2WI, respectively) and check their diagnostic performances in detecting osteoporosis, low bone mass, and osteoporotic vertebral fracture using DXA-aBMD as a reference standard.
Materials and Methods
Study design
This prospective study enrolled a total of 61 human participants. Only female patients who underwent an MRI of the lumbar spine and DXA scanning were included. This single-center study was conducted between November 2022 and April 2023, as per the Declaration of Helsinki. The ethical approvals were obtained from the Lady Reading Hospital Medical Teaching Institution Institutional Review Board (LRH-MTI IRB) (approval numbers: 319/LRH/MTI & 01/LRH/MTI). Written consent was obtained from only women aged ≥55 years.
The average waiting time between MRI and DXA was approximately 1 hour. All MRI examinations were performed with a 1.5-Tesla superconducting MRI scanner (Toshiba Excelart Vantage, Tustin, CA, USA) using an in-built spine coil. All patients who had previous fractures, metabolic bone diseases, and chronic immobilization and were taking medications (such as thyroxine, estrogen, and corticosteroids) were excluded. Fig. 1 shows the flow chart of patient recruitment.
Patient selection diagram: 42 female volunteers consented to undergo both magnetic resonance imaging (MRI) and dual-energy X-ray absorptiometry (DXA) examinations following the inclusion and exclusion criteria.
Magnetic resonance imaging of the human participants with a phantom
A phantom consisting of coconut oil (T1 reference control) and deionized water (T2 reference control) was used in the MRI of the lumbar spine (Fig. 2). The wedge-shaped acrylic phantom was constructed, which accommodated the lumbar lordotic space with a length of 12 cm and width of 15 cm and caudal end corner angles of 30° and cranial end of 25°. Other dimensions and characteristics are shown in Fig. 2. The two reference controls, i.e., T1 and T2, were each filled with a separate tube, with a length of 10 mm, wall thickness of 2 mm, and diameter of 12 mm. In MRI, safety was crucial. Supplement 1 shows the materials used for the phantom, including the specification and composition of the T1 control (coconut oil) and T2 control (deionized water).
An acrylic phantom was designed and developed consisting of coconut oil as a good T1 reference control and deionized water as a good T2 reference control. The phantom shape was consistent and well-fit for the lumbar lordotic space and was used during magnetic resonance imaging (MRI) scanning.
Routing sagittal T1WI and T2WI sequences were obtained using the imaging parameters outlined in Table 1. The images acquired were subsequently transferred from the MRI console to the reporting workstation. All images were analyzed separately by two radiologists, each with >25 years of clinical experience in musculoskeletal reporting. They were blinded to the clinical information and independently analyzed the phantom-based T1WI and T2WI for S1. Values of SI parameters were derived for a total of 30 S1 vertebrae that were assessed to obtain SI values on both T1WI and T2WI. A region of interest (ROI) was drawn in the upper part of S1, avoiding the cortical bone, endplate, posterior venous plexus, focal lesions, or any degeneration, if present. The ROIs were obtained using MircoDicom DICOM viewer software (Fig. 3). Similarly, ROIs of 30–50 (±5) pixels were drawn to measure the SI values in respective T1WI and T2WI (phantom reference controls) using the following equations:
Imaging parameters used for routine magnetic resonance imaging sequences
Magnetic resonance (MR) images were analyzed with free MircoDicom software. (A) To calculate the F-scoreS1, regions of interest (ROI) indicated by white arrowheads were selected on S1 vertebra and on coconut oil to calculate the signal intensity (SI) using T1-weighted images, while a small ROI was selected for cerebrospinal fluid (as a control) at L3 level to calculate the vertebral bone quality (VBQ) score; (B) similarly, to calculate the W-scoreS1, T2-weighted images were used to measure SI at S1 vertebra and SI of deionized water (T2 reference control), indicated by white arrowheads.
For T1WI,
where SI refers to the signal intensity, S1 signifies the first sacral bone, and SICO represents the SI of the T1 control (coconut oil) of the phantom. To get rid of the fractional value, the resultant ratio was multiplied by 10. F-scoreS1 may reveal the relative proportion of fat-like tissue within the vertebral ROI.
For T2WI,
where SIDW refers to the SI of the deionized water (T2 control), because pure water was used as the reference control and had plenty of protons for T2 and W-scoreS1 could show how much water-like tissue is present in the S1 vertebra.
For comparison, the VBQS1 was calculated, and S1 was divided by the CSF ROI, which was obtained at the L3 level [13].
Dual-energy X-ray absorptiometry scanning
All patients consented to undergo DXA of the lumbar spine using a standard Hologic (Horizon A System; Hologic Inc., Marlborough, MA, USA) as per the guidelines of the International Society of Clinical Densitometry [17]. The aBMD and T-score were calculated based on the DXA results [8]. Patients were classified based on the T-score: a standard deviation of ≥−1 presented as normal, between −1 and >−2.5 as osteopenia, and ≤−2.5 as osteoporosis. To identify vertebral fractures, Genant’s semiquantitative method (24) was used, which analyzed changes in the vertebral shapes at the anterior, middle, or posterior, and reduced vertebral height was considered.
Statistical analyses
All statistical analyses were performed using IBM SPSS Statistics for Windows ver. 26.0 (IBM Corp., Armonk, NY, USA). The intra- and inter-reader reproducibilities of the newly defined F- and W-scoreS1 were assessed between the readers using the intraclass correlation coefficient (ICC), whereas Cohen’s kappa was used for vertebral fractures. To determine the differences between these groups, an independent-sample t-test was used, and p<0.05 was considered significant. Violin plots were generated to illustrate the distribution of new F- and W-scoreS1 in osteoporosis, low bone mass, and fracture groups. The diagnostic performances of F-scoreS1, W-scoreS1, and VBQS1 in osteoporosis and vertebral fractures were evaluated using receiver operating characteristic analysis and area under the curve (AUC). Moreover, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy in diagnosing osteoporosis were determined.
Results
A total of thirty postmenopausal female patients (average age, 62.33±6.25 years) with a body mass index (BMI) of 28.78±6.15 kg/m2 were included in this study (Fig. 1). No significant difference in height, age, weight, or BMI was found between the osteoporotic (13/30) and non-osteoporotic (17/30) groups. The mean aBMD was 0.69 g/cm2 in the osteoporotic group compared with 0.93 g/cm2 in the non-osteoporotic group. Patients with low bone mass had an aBMD of 0.74 g/cm2 compared with 1.02 g/cm2 in patients with normal bone mass. Similarly, patients with vertebral fractures had an aBMD of 0.74 g/cm2 compared with 0.88 g/cm2 in patients without fractures (all p<0.05). Fig. 4A–C shows negative correlations between the F-scoreS1 and the aBMD (−0.64 with p<0.001), W-scoreS1 and aBMD (−0.60 with p=0.001), and VBQS1 and aBMD (−0.54 with p=0.002). Similarly, a positive correlation was found between the F-scoreS1 and W-scoreS1 (r=0.65 with p<0.001) (Fig. 4D).
(A, B) Negative and strong correlations were found between F-scoreS1, W-scoreS1, and areal bone mineral density (aBMD), compared to (C) the vertebral bone quality score (VBQ) showed similar negative correlation; however, the strongest correlation was reported between the two new scores, (D) Additionally, a positive and significant correlation was found between the S1 F-score and W-score, which shows that with an increase in fat content, there could be an increase in water-like contents. CI, confidence interval.
S1-based magnetic resonance imaging scores as osteoporosis indicators
The mean F- and W-scoreS1 were higher in participants with osteoporosis (4.11 and 2.43) than in those without osteoporosis (3.25 and 1.92), and the median and interquartile S1-based F- and W-scoreS1 were also higher in participants with osteoporosis (Table 2). The scores were higher in patients with osteoporosis (F-scoreS1=20.92%, W-scoreS1=20.98%, and VBQS1=14.15%) than in those without osteoporosis. When detecting osteoporosis (Fig. 5), the F- and W-scoreS1 resulted in AUCs of 0.82 (95% confidence interval [CI], 0.67–0.97; p=0.003) and 0.76 (95% CI, 0.58–0.94; p=0.016), respectively, whereas VBQS1 had an AUC of 0.74 (95% CI, 0.56–0.92; p=0.025), which is lower than those of F-scoreS1 and W-scoreS1. Furthermore, a median F-scoreS1 of 3.50 had a diagnostic performance with a sensitivity of 84.6%, specificity of 70.6%, PPV of 68.8%, NPV of 85.7%, and accuracy of 76.67%, and a median W-scoreS1 of 2.08 resulted in a sensitivity of 76.9%, specificity of 70.6%, PPV of 66.7%, NPV of 80%, and accuracy of 73.33% (Table 3). The ICC values of 0.99 and 1.00 for the F- and W-scoreS1, respectively, demonstrated the high degree of agreement between the two readers in the score calculation. A strong level of reproducibility in ROI positioning on S1 was achieved, which demonstrated a remarkably high ICC for ROI on both T1WI- and T2WI-based SI, with values of 0.996 (p<0.001) and 0.998 (p<0.001), respectively.
Number of subjects (and percentages) and mean values of various scores for different groups
The performances of the F-scoreS1 and W-scoreS1 were assessed for osteoporosis (A), low bone mass (B), and vertebral fractures (C). The receiver operating characteristic area under the curve (AUC) for F-scoreS1 and W-scoreS1 shows significant differences in F-scoreS1, W-scoreS1, and areal bone mineral density (aBMD) identified between non-osteoporosis and osteoporosis, normal and low bone mass, as well as between non-fracture and fracture cases. Compared to the vertebral bone quality (VBQ) score and the current standard dual-energy X-ray absorptiometry-aBMD in detecting osteoporosis, low bone mass, and vertebral fractures, the new scores showed improved results. CI, confidence interval.
Diagnostic accuracy of the new scores in osteoporosis detection
Performances of the S1 scores in detecting low bone mass
Similarly, 21 out of 30 patients were found to have low bone mass (osteopenia and osteoporosis) based on the DXA T-score. The mean F- and W-scoreS1 in patients with low bone mass were 3.93 and 2.35, respectively, compared with patients with normal bone mass (2.91 and 1.65). The scores were higher in patients with low bone mass or osteopenia (F-scoreS1=25.95%, W-scoreS1=29.78%, and VBQS1=23.70%) than in healthy patients; however, using the aBMD as a standard of reference, the difference was 38.88%. The AUCs were higher for the newly defined scores than for the VBQS1, whereas the highest AUC was reported for W-scoreS1 (0.86; 95% CI, 0.72–0.99; p=0.002) (Fig. 5). The AUC of F-scoreS1 was 0.84 (95% CI, 0.69–0.99; p=0.004), and that of VBQS1 was 0.80 (95% CI, 0.65–0.96; p=0.009).
Performances of S1 scores in detecting vertebral fractures
The 12 of 30 participants (40%) had osteoporotic fractures, according to Genant’s semiquantitative approach, and the inter-reader agreement results showed a Cohen’s κ of 0.93. The mean F- and W-scoreS1 were higher in patients with osteoporotic vertebral fractures (4.11 and 2.63, respectively) than in those without fractures (3.30 and 1.82, respectively) with p<0.05. These scores were higher in patients with fractures (F-scoreS1=19.70%, W-scoreS1=30.79%, VBQS1=24.08%, and BMD=18.91%) than in those without fractures (Table 2). When used to distinguish vertebral fractures, W-scoreS1 obtained in the highest AUC of 0.90 (95% CI, 0.78–1.00; p<0.001), and F-scoreS1 had an AUC of 0.74 (95% CI, 0.55–0.92; p<0.033) and VBQS1 had 0.73 (95% CI, 0.58–0.95; p<0.03), respectively (Fig. 5). The violin plots showed a wider overlapping distribution for aBMD than for S1-based F- and W-scoresS1 between participants with and without fractures (Fig. 5).
Discussion
The S1 vertebra is clinically important, particularly in assessing bone quality in spine surgeries. This study demonstrated that S1 is an independent site to indicate osteoporotic changes on MRI. Our results are in agreement with the findings of earlier literature that identified S1 as an indicator of osteoporosis [2,14,18]. However, no studies have evaluated the role of S1 in osteoporosis for fat–water-like tissue changes in postmenopausal women utilizing routine MRI sequences. In this study, F-scoreS1 and W-scoreS1 demonstrated diagnostic performance with AUCs of 0.82 and 0.76, respectively, compared with VBQS1 when detecting osteoporosis. Compared with aBMD, the new scores also performed better in differentiating osteoporotic vertebral fractures. This implies that our new method could offer an alternative diagnostic approach to osteoporosis assessment. Moreover, osteoporosis evaluation based on S1 has several benefits. First, the intra- and interrater reliabilities are better than when evaluating the lumbar vertebrae (L1–L4). Some patients also remain immobile owing to spinal compression fractures, which primarily occur around the thoracolumbar joint; as a result, the patient may be unable to undergo DXA. In addition, the DXA score of patients with L1 and L2 injuries could be challenging and erroneous. Moreover, S1 assessment is considered crucial for spine surgeries because it is unaffected by spinal injury.
The most commonly applied technique to diagnose osteoporosis is the DXA T-score [8]. However, the lower sensitivity and effect of degenerative structures limits its role [9]. In addition, current guidelines recommend that women aged >65 years and men aged >70 years should receive DXA-BMD screening [19]. However, DXA referrals decreased by 27.2% [20]. Owing to the underutilization of DXA and underdiagnosis of individuals at risk, researchers have been working to seek other opportunistic methods for osteoporosis screening. MRI offers better tissue contrast and qualitative information than other imaging modalities [21]. Routine MRI-based scoring methods are emerging as new alternative tools for osteoporosis evaluation because they can determine changes in the bone marrow that are linked to osteoporosis. In detecting vertebral fractures, our scores performed better than aBMD. A reason could be the increase in BMD, which was found to occur as an early change compared with BMD loss, or the aforementioned limitations of DXA and lower sensitivity [9,10]. A study indicated that bone loss in osteoporosis is largely caused by excessive synthesis of the BMD [22]. Similarly, the BMD is thought to be the main reason for initiating bone loss in postmenopausal women, and recent studies are targeting the BMD to treat conditions associated with bone loss [23,24].
In this study, the F-scoreS1 was used to quantify bone marrow changes linked to osteoporosis, which were found to negatively correlate with aBMD (r=−0.64, p=0.001). The newly defined F- and W-scoreS1 were higher in female patients who had osteoporosis, low bone mass, and vertebral fractures. The current study also found the highest AUC of 0.90 for W-scoreS1 in patients with vertebral fractures, indicating that the presence of osteoporosis is associated with water-like tissue changes, as reported previously [25]. In the present study, an increase in water-like content is consistent with fatty replacement in osteoporosis, as was shown by a positive correlation between F- and W-scoreS1 (r=0.65, p<0.001). This study also addressed the role of T2WI in osteoporosis, which was recommended by Hammood et al. [26].
Furthermore, the statistically significant p-values further support the reliability of our results, which were for ROI-based SI calculation from S1. We believe that these outcomes validate the accuracy and consistency of our approach to delineating ROI, thus ensuring the reliability of our findings. Similarly, an excellent inter-reader agreement was reported in calculating the F- and W-scoreS1. The higher ICC showed higher reproducibility for our method. In addition, S1 has better bone quality and bone strength than the lumbar spine [27]. However, osteoporosis affects the entire skeleton, including the sacrum, and osteoporotic changes could be assessed at S1. In this study, our S1-based scoring method can distinguish between patients with and without osteoporosis, low bone mass, and even vertebral fractures. S1 exhibits superior bone quality metrics than the lumbar spine and can be a good site for assessing these changes. Comprehensive assessments of bone condition at S1 could be considered for a more holistic understanding [2,28].
Some limitations must be discussed. First, this was a relatively small-sample study. Because we mainly focused on the technical development and testing of this idea, this study demonstrated the feasibility of this method, which provides better diagnostic outcomes. Second, some of the cross-sectional images of the phantom reference controls were missing in the mid-sagittal view; however, the ROIs were selected next to the mid-sagittal view or any available on sagittal scans. Third, the S1 BMD was unavailable, and the DXA-based BMD from L1–L4 was considered to determine osteoporosis. We should also consider cost-effectiveness in clinical decision-making, particularly when using MRI. The present study contributes to the growing body of literature supporting the potential role of MRI in serial osteoporosis assessment. However, in our proposed method, we are utilizing routine sagittal T1WI and T2WI sequences, which take less than 5 minutes, which also means less scanning time and less cost. In our future research, we intend to further explore the cost-effectiveness of MRI-based follow-up spine assessments by conducting sensitivity analyses and incorporating relevant economic modeling approaches. Future larger studies are recommended to further validate the performance of these new modified methods for generalizability and clinical utility in osteoporosis.
Conclusions
In conclusion, the proposed S1 vertebra-based F- and W-scoreS1 were developed based on routine T1WI and T2WI sequences. These new scores showed that S1 could be the sole site to indicate osteoporosis-related changes. The new phantom-based scores demonstrated improved abilities over DXA-aBMD in differentiating patients with vertebral fractures. Furthermore, the use of routine T1WI and T2WI sequences to identify vertebral bone water–fat changes in osteoporosis could be implemented for a wider clinical purpose.
Key Points
Osteoporosis is a public health musculoskeletal disorder consistent with fat–water-like tissue changes in the spine.
The novel S1-based magnetic resonance imaging scoring system utilizing routine T1 and T2 sequences demonstrated capabilities in detecting osteoporosis and discriminating vertebral fractures.
S1 was identified as an indicator of osteoporosis, and the novel scoring system may provide an alternative tool other than bone mineral density measurement for wider opportunistic use in clinics.
Notes
Conflict of Interest
No potential conflict of interest relevant to this article was reported.
Funding
Funding from the National Natural Science Foundation of China (12272017) and Beijing Natural Science Foundation (L232058) is acknowledged.
Author Contributions
Conceptualization: RD, HY; data curation: RD, TN, XC; formal analysis: all authors; funding acquisition: HY; methodology: RD, HY; project administration: all authors; visualization: RD, HY; writing–original draft: RD, HY; writing–review & editing: all authors; and final approval of the manuscript: all authors.
Supplementary Materials
Supplementary materials can be available from https://doi.org/10.31616/asj.2023.0116.
Supplement 1. Specifications of the materials used with relevant magnetic resonance imaging safety measures.
asj-2024-0116-Supplementary-1.pdf