Axial spondyloarthritis (axSpA) is a chronic inflammatory rheumatic disease affecting the spine and the sacroiliac (SI) joints. The prominent feature of axSpA in the new bone formation is ankylosis in the SI joints and bridging syndesmophytes in the spine (1). This feature is associated with irreversible spinal immobility and functional limitation. Overall, new bone formation is a slow process in axSpA. Magnetic resonance imaging (MRI) is superior to radiography because it can demonstrate active inflammation and structural damages, even in patients in early axSpA with normal radiography findings. Fat metaplasia in an erosion cavity, also known as backfill, is a feature that can be detected on an MRI and is associated with new bone formation in the SI joints (1, 2). It has been defined as the presence of high signal intensity on T1-weighted images (T1WI) within the SI joint space and is hypothesized to represent reparative tissue refilling the eroded subchondral bone (2). Thus, identifying patients with a high likelihood of backfill progression is critical to adjusting the treatment strategy early in the disease process.

Inflammatory and chronic structural damages are objective signs of axSpA. Several studies have used the Spondyloarthritis Research Consortium of Canada (SPARCC) scoring system to evaluate the inflammatory status of SI joints (3). It is the most popular imaging standard of inflammation with high accuracy and reproducibility (4). Previous studies have established a correlation between SPARCC scores and clinical disease activity (5). Another MRI-based sacroiliitis activity index is the apparent diffusion coefficient (ADC) value from diffusion-weighted imaging (DWI), which detects the diffusion of water molecules and has been applied in the diagnosis and efficacy evaluation of sacroiliitis (6, 7). Fat metaplasia is frequently observed in the bone marrow after the resolution of inflammation in patients with axSpA. Previous studies have demonstrated that fat metaplasia is associated with an increased risk of spinal progression (8, 9). The mDixon-Quant sequence is one of the most accurate techniques to quantify bone marrow fat content by measuring the fat fraction (FF). FF is considered an alternative to invasive biopsy to study the distribution and quality of fat deposition (10, 11). Previous studies have investigated the usefulness of quantifying fat deposition by measuring fat fraction (FF) in patients with axSpA patients (12). In addition, a study suggests that FF estimation in the post-inflammatory area indicates the chronicity of axSpA and it is valuable in determining radiologic progression (13).

Only a few studies have explored predictors for backfill (14) and no studies have explored using quantitative MRI, which may provide important information for predicting new bone formation and prognosis in patients with axSpA. Therefore, this longitudinal study aimed to quantitatively assess the inflammatory and chronic structural damages and identify the predictors of backfill progression in axSpA patients.

A total of 52 patients with axSpA who underwent baseline and follow-up MRI were evaluated (Figure 1). The mean age and disease duration were 25.94 ± 6.60 and 2.31 ± 1.55 years, respectively. 19/52 (36.5%) patients had “backfill” at baseline. After follow-up, new backfill progressed in 30/52 (57.7%) patients (Figure 2). One of them showed partial ankylosis during the 12-month follow-up. Fat metaplasia was observed in the periarticular bone marrow of 22 patients (Figure 3) but there was no backfill progression.

Table 1 summarizes the baseline characteristics of the patients. The average follow-up periods were 5.9 and 17 months in the progressive and non-progressive groups. The mean age of the patients with backfill was slightly lower than those without backfill progression. However, our study group showed no statistical difference (P=0.274). The mean disease duration was higher in patients with backfill progression than in those without (P<0.001). No statistically significant differences were observed in CRP and ESR levels between the progressive and non-progressive groups.

The mean SPARCC score was 27.47 ± 17.57 (range: 7-66) and 14.00 ± 8.53 (2-30) in the progressive and non-progressive groups, respectively. The mean baseline SPARCC score was significantly different between patients with and without backfill progression (P=0.003). Moreover, the patients with backfill progression had a higher frequency of backfill at baseline, a higher SSS score of fat metaplasia, and a higher SSS score of backfill than those who did not develop new backfill (P=0.001, P<0.001, and P=0.002, respectively). No significant differences were detected in the baseline SSS score of erosion between the two groups (Figure 4).

The mean FF values of the progressive and non-progressive groups in the fat metaplasia area were 77.45 ± 10.44% and 68.26 ± 15.05%, respectively. A higher FF in the fat metaplasia area at baseline was more frequent in patients than in those without backfill progression (P=0.019). The mean ADC and FF values in the BME area of the progressive group were 114.6 ± 23.2 × 10^-5 mm²/s and 19.9 ± 13.6%, respectively. Moreover, the mean ADC and FF values of the non-progressive group were 111.1 ± 34.9 × 10^-5 mm²/s and 18.2 ± 10.7%, respectively. No significant differences were observed in the ADC and FF values in BME areas (Figure 4).

Univariate logistic regression analysis was performed to identify the factors associated with backfill progression of SI joints in patients with axSpA. The results showed that SPARCC score of inflammation [odds ratio (OR) 1.115, 95% confidence interval (CI): 1.031-1.205, P=0.006] and SSS score of fat metaplasia (OR 1.136, 95% CI: 1.042-1.240, P=0.004) differed significantly (Table 2).

MRI evaluation revealed a strong to excellent agreement between the two readers. An excellent correlation was achieved for the SPARCC and SSS assessment of inflammation, backfill, and fat metaplasia at baseline (ICC 0.83-0.91, 0.88-0.93, and 0.95-0.97, respectively). A strong to excellent correlation was established to identify erosion at baseline (ICC 0.78-0.90).

This study attempted to investigate the backfill progression in the SI joints and identify the potential risk factors of backfill in axSpA patients. About 57.7% of patients showed backfill progression over a 5.9-month follow-up interval. The results indicated that a high SPARCC score of inflammation and a high SSS score of fat metaplasia at baseline are predictors of the progression of backfill in axSpA patients. Furthermore, the mean disease duration, SPARCC score, SSS scores of fat metaplasia and backfill, frequency of backfill, and FF values in the fat metaplasia area at baseline were significantly associated with progressive backfill.

In 2019, the ASAS MRI working group introduced the new concept of backfill (2). Backfill was present in almost 66% of patients with spondyloarthritis (SpA) and has shown a high diagnostic value (17, 18). According to previous studies, fat metaplasia and backfill are early post-inflammatory changes and might reflect the early stage of bone remodeling (19, 20). The histology is not clear at this time. A previous study showed that the fat signal detected by spinal MRI of AS patients is related to a high content of adipocytes. Therefore, the disturbance of the homeostasis between osteoblasts and osteoclasts may be a crucial reason for new bone formation (21). A multivariate analysis of new bone formation in ankylosing spondylitis (AS) showed that fat metaplasia and backfill were significantly correlated with ankylosis (14). Typically, backfill indicates the initiation of new bone formation (1). However, no study has explored the correlation between fat metaplasia and backfill. Interestingly, we found that a high SSS score of fat metaplasia is significantly associated with the development of backfill. High FF values in fat metaplasia areas were frequent in patients with backfill progression, suggesting that fat metaplasia is a preceding stage of backfill. However, the histopathological analysis is not precise at this moment, which could be attributed to the effect of fat deposition on osteoblastogenesis in the bone marrow (22).

Subchondral bone destruction occurs after inflammation and reparative tissue fills the eroded area and progresses to ankylosis. A prospective study reported that baseline SI joints inflammation and backfill scores independently predict the 5-year progression of ankylosis (23). However, this study did not indicate the impact of the correlation between backfill and SPARCC scores. In the present study, a high SPARCC score predicted backfill progression in axSpA. This finding was consistent with a multivariable analysis that showed a correlation between backfill and SPARCC scores (24). This might be related to the expression of bone morphogenetic protein (BMP) genes. BMP gene is involved in chondrocyte differentiation and bone formation and the BMP-2 gene promotes new bone formation. Another study of AS patients demonstrated that the expression level of the BMP-2 gene was regulated by pro-inflammatory cytokines (such as TNF-α or interleukin (IL) -1β) and the higher the expression of the BMP-2 gene, the faster the radiological progression of AS patients (25). Therefore, it is speculated that severe sacroiliitis, as detected on MRI, upregulates the level of the BMP-2 gene, which further leads to the formation of new bone. Therefore, the degree of inflammation reflected by the SPARCC score is closely related to clinical treatment and prognostic stratification. Therefore, in patients with high SPARCC scores, targeted and aggressive therapy may be required to prevent further radiological progression. Both SPARCC score and ADC value are MRI sacroiliitis activity indices. The lack of correlation between ADC value in the BME area and the backfill progression may be because the SPARCC score reflects the extent, intensity, and depth of inflammation comprehensively, while ADC value reflects only the intensity of BME in a single area or it could be attributed to the absence of a uniform method for ROI delineation.

Delaying or preventing imaging progression is a major goal in the axSpA treatment process. Current studies have found that TNFi treatment can significantly improve inflammation and a study has shown that patients treated with TNFi have a lower rate of structural progression in the SI joints (26). All patients in this study were treated with TNF-α inhibitors. NSAIDs have been internationally recognized for the treatment of peripheral joint symptoms in axSpA; however, there is a lack of evidence for the drug’s therapeutic effect on axial skeleton disease and improving disease prognosis. Recently, IL-17 has become a hotspot research target in the field of axSpA treatment. IL-17 inhibitor secukinumab may have advantages in controlling and delaying the structural progression of axSpA (27), which needs to be verified by large-scale studies.

Nevertheless, the study has several limitations. Firstly, the sample size was small. Secondly, the patients in this cohort were followed up for only an average of 17 months. In the current study, the average follow-up period was 5.9 months in the progressive group and a previous study suggested that the change in backfill can be observed in 5 months (28). Long follow-up intervals would be influenced by several factors, such as self-discontinuation of medication by patients and recurrence of inflammation. Lastly, the ROI delineation was only a small, circular area, and defining ROI is methodologically challenging. Thus, the delineation of the full lesion is essential in future studies.

The long disease duration, high SPARCC score, high SSS scores of fat metaplasia and backfill, high frequency of backfill at baseline, and high FF value in fat metaplasia area at baseline were frequent in patients with backfill progression. The severity of sacroiliitis and extensive fat metaplasia at baseline indicated a poor prognosis and an increased tendency of new bone formation in the SI joints.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Institutional Review Board of Tianjin First Central Hospital Medical Ethics Committee. The patients/participants provided their written informed consent to participate in this study.

YMZ, ZG, and XL contributed to the conception and design of the study. All authors performed the acquisition, analysis, and interpretation of data. YMZ wrote the manuscript. All authors reviewed the manuscript, and all authors approved the submitted version.

Funded by Tianjin Key Medical Discipline (Specialty) Construction Project.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.