Detailed bone assessment of the sacroiliac joint in a prospective imaging study: comparison between computed tomography, zero echo time, and black bone magnetic resonance imaging

Changes in the sacroiliac (SI) joints are frequently related to degenerative disease, secondary to rheumatologic (e.g., axial spondyloarthropathies) or other nonrheumatologic disorders such as infectious, drug-, trauma-, or pregnancy-related [1]. Imaging findings of the SI joints are important since clinical decision-making heavily relies on these findings. Notably, in the context of seronegative spondyloarthropathies, imaging findings are particularly important as these diseases are not diagnosed using specific biochemical markers. Furthermore, imaging findings are included in the Assessment of Spondyloarthritis (ASAS) axial Spondyloarthritis (axSpA) classification system [2]. Although MRI is key for the detection of early changes like bone marrow edema, osteitis, enthesitis, or capsulitis [3], CT and conventional radiography can better identify specific structural osseous changes of the SI joints occurring at a later stage of the disease, e.g., erosions, subcortical sclerosis, and ankylosis, and help to differentiate post-inflammatory from degenerative changes [4]. Early diagnosis of these osseous changes is of paramount importance to correctly diagnose patients, prevent therapeutic delays, and improve overall outcome.

Recently, new MRI sequences have become available that may allow for accurate cortical and trabecular bone assessment. ZTE (zero echo time) [5‐7] and UTE (ultrashort echo time) [8‐11] MRI sequences belong to a group of novel 3D MRI pulse sequences that use an echo time (TE) of 1 ms or less enabling the acquisition of sufficient signal from rapidly decaying short-T2 tissue components which are abundant in calcified cartilage such as at the transition of articular cartilage to the subchondral bone, tendons, and menisci (Fig. 1) [12‐14].

In particular, in these short echo time sequences, signal can be provided from short T2 tissues to provide a contrast mechanism that affords visualization of the cortical bone of the SI joint [5, 6, 12]. Importantly, ZTE and UTE feature comparable contrast-to-noise and signal-to-noise ratios (CNR/SNR) [15].

Another promising established MRI sequence that may potentially enable bone assessment is the black bone (BB) sequence. This 3D low flip angle gradient-echo MRI sequence features high contrast between bone and other tissues while providing a low contrast between those tissues [16‐18].

As previously described, it would be desirable to detect both early (e.g., bone edema) and more chronic (subchondral sclerosis) changes in the SI joints by a single MRI scan as an “all-in-one” diagnostic step. A holistic MRI protocol would be more time and cost-efficient, optimizes the diagnostic workflows, and would potentially even allow for the replacement of CT imaging in certain circumstances reducing healthcare costs and ionizing radiation exposure and allowing for more efficient and timely clinical interpretation. Though low-dose CT is broadly implemented, the ALARA (as low as reasonably achievable) principles should also be considered with avoidance of any ionizing radiation if possible. Especially, since younger patients are predominately affected by rheumatological changes in the SI joint. Thus, the goal of this prospective study was to evaluate the diagnostic performance of ZTE and black bone MRI sequences for bone assessment of the SI joint and demonstrate the non-inferiority compared to CT as the reference standard.

There is no prior work, to our knowledge, that has explored the value of ZTE and BB MRI sequences intra-individually for the osseous assessment of the SI joints compared to CT as the reference standard.

Ninety-seven patients were initially included in the study and examined by 3-Tesla MRI. Eight patients were excluded due to incomplete ZTE/BB or CT datasets or poor image quality due to motion artifacts. Eventually, 89 patients (55 men and 34 women with a mean age of 55.7, standard deviation of 15 years, and median of 57 years) were enrolled in the final analysis. The average body mass index (BMI) was 25.8 with a standard deviation of 3.1 kg/m². The mean time gap between the CT and MRI scan was 5.2 months (with a standard deviation of 5.4 months, a median of 4 months, a minimum of 0, and a maximum of 12 months).

A detailed overview of the data is provided in Fig. 3 and Table 2. Inter-reader agreement was moderate to excellent (k values: 0.5–0.83). Except for the feature gas in SI joints, there were no significant differences between the three imaging modalities for the remaining 6 features (p > 0.52 for all 6 features, Figs. 4, 5, and 6). For the feature gas in SI joints, ZTE exhibited significantly lower scores than both BB and CT (both p < 0.001), while there were no significant differences between BB and CT (p = 0.1).

Furthermore, except for the feature gas in SI joints on ZTE (alpha = 0.179), Krippendorff’s alpha coefficients indicated a substantial or almost perfect agreement for all other features of both MRI sequences (ZTE and BB) compared to CT. Lastly, when combining the diagnostic performance of all parameters, ZTE achieved a sensitivity and specificity of 64.1% and 98.8%, respectively, while BB achieved a sensitivity of 78.3% and a specificity of 98.9% compared to CT. If the single feature “gas in SI joints” was excluded from the calculation of the diagnostic performance, there was an improvement particularly for ZTE with 76.7%/98.6% sensitivity/specificity, while BB achieved 80.8%/99.1% sensitivity/specificity compared to CT. Sensitivity and specificity are listed in Table 3.

Our study showed that a ZTE and BB sequence provides similar diagnostic performance to CT with a high inter-reader and inter-modality agreement for six of the seven investigated features (except for the feature “gas in the SI joints”). However, the low prevalence of two of the investigated features in our study group (erosions, 1/178 of SI joints and ankylosis, 2/178 of SI joints) limited the meaning of the calculated sensitivity and specificity values for these two features.

Adding a ZTE or BB sequence to the regular SI joints MRI protocol would allow for the simultaneous detection of both osseous and non-osseous (e.g., cartilage abnormalities and enthesitis) changes in the SI joints with only one single MRI scan. The results show the value of ZTE and BB MRI sequences in displaying precise osseous morphology and offering detail beyond what traditional MR sequences can provide. Streamlining the diagnostic process can improve time to treatment, especially critical in inflammatory etiologies.

Notably, scan time is prolonged only by approximately 4 min by adding a ZTE or BB sequence to the MRI protocol and due to the 3D nature of the datasets, customized multiplanar reformations can be performed.

Only in terms of the detection of gas in the SI joints, ZTE and to a lesser extent BB showed significantly lower sensitivity than CT. In general, the presence of gas in the SI joints is, analogous to other joints, most frequently considered to be a nonspecific finding or a sign of degeneration or recent trauma. This imaging finding corresponds to a vacuum phenomenon primarily caused by the intraarticular accumulation of nitrogen [20, 21]. Gas bubbles lead to susceptibility artifacts on MR images, to which ZTE sequences are less prone owing to a very low echo time and flip angle [22, 23]. On the other hand, due to the nature of gradient-echo MR sequences, gas accumulation becomes more apparent in the BB sequence [24].

Some of the different SI joint pathologies arise from etiologies (e.g., traumatic and rheumatic) which are present in younger patients where ionizing radiation is a particular concern. Especially, since radiosensitive organs, like the ovaries and testes, are located in close proximity to the SI joints [25], the substitution of a CT scan by ionizing radiation-free ZTE or BB MRI sequences would remedy this concern.

The combination of an image acquisition that is not associated with the exposure to ionizing radiation, precise bone depiction, and detection of early sacroiliac joint changes like bone marrow edema makes SI joint MRI augmented by a ZTE or BB sequence a plausible “all in one” diagnostic procedure and may render CT imaging for some assessments obsolete. This is in line with other MRI studies about ZTE in other body regions, e.g., the shoulder and the spine, where a successful evaluation of cortical and trabecular bone without CT has been demonstrated [7, 8, 26, 27].

Our study has limitations. First, the period between CT and MRI ranged from 0 to 12 months. However, the average time gap between the CT and MRI scan was only 5.2 ± 5.4 months. Within this short period, rapid changes to the SI joints in our random patient cohort were unlikely. Second, no analysis of the ZTE and BB sequences in the semicoronal planes was performed even though these are efficient planes for the detection of osseous SI joint changes, since there were no adequate CT datasets for the custom reformations available for the majority of scans. Third, to compute sensitivity and specificity, we elected to dichotomize the assessment scores and hence erased possible small differences in the grading of the two readers (between Likert scales of 0 and 1 or 2 and 3). Lastly, the prevalence of pathologic changes in the SI joints in our patient cohort was rather low and less distinct, because the included patients were randomly selected mirroring the general population as described in the “Materials and methods” section. Additional prospectively acquired studies dedicated to the SI joints, especially in patients with SI joint pathology (e.g., erosions), would be helpful to reconfirm the results of our study.

Overall performance of the detailed bone assessment of the SI joints using ZTE and BB sequences was comparable to CT and hence may serve as an alternative to CT for the assessment of cortical and trabecular bone alterations.