The clinical and MRI effect of TNF-α inhibitor in spondyloarthritis patients with hip involvement: a real-world prospective clinical study

Background: Hip involvement is an important cause of disability and poor prognostic in patients with spondyloarthritis (SpA). TNF-α inhibitor treatment had been demonstrated to be effective in SpA patients with hip arthritis, however, quantitative assessment with magnetic resonance imaging (MRI) in the longterm follow-up still needs further application. Methods: A total of 239 patients were involved in this study, MTX and SSZ were given as basic treatment. 165 patients received TNF-α inhibitor plus basic treatment and 74 patients received basic treatment only as control. The clinical symptoms were assessed at baseline, week 12, week 24, and week 52. The MRI performances of hip arthritis including bone marrow edema (BME) and synovitis were quantitatively assessed using Hip MRI Inammation Scoring System (HIMRISS). Results: The clinical symptoms of ESR, CRP, BASDAI, Harris, and ASDAS-ESR of both groups got signicant clinical remission at week 52 (p<0.001). However, the disease activity change levels at week 52 in control group were signicantly worse than TNF group. In week 52, MRI performance showed signicant remission trend in TNF-α inhibitor group versus baseline of which total HIMRISS scores decreased signicantly ( 26.49±10.37 vs 20.59±9.41, p<0.001), but the control group only had slight improvement trend (p<0.05). Conclusions: TNF-α inhibitor could signicantly and better improve the clinical and MRI manifestation of hip involvement in SpA patients than controls. Quantitative MRI assessment combined with clinical assessment could accurately evaluate the treatment effect of TNF-α for SpA patients with hip involvement, and help to guide targeted treatment.

Retrospectively registered http://www.chictr.org.cn/showproj.aspx?proj=7701 Background Hip involvement is an important cause of disability and poor prognostics in approximately 10 to 50% of spondyloarthritis (SpA) patients [1,2], and 47%-90% with bilateral hip involvement [3]. Progressed hip involvement was reported to be associated with more severe spinal involvement [4,5], and seriously affect joint function and life quality. In addition, nearly 8% of patients with hip involvement suffers from intractable pain and disability. Even if they are treated with total hip arthroplasty, it is likely at high risk of revision surgery and high complication rates [6].
The principal clinical manifestations of hip involvement in SpA including in ammation of the subchondral BME and synovitis effusion [7]. MRI had been widely used as a sensitive tool to detect hip arthritis. It was proved to be helpful for early diagnosis and suppression of in ammation before structural damage [8,9]. HIMRISS was a new quantitative assessment method which was performed based on several MRI slices from one patient, and could systematically assess the synovitis and BME of femoral head and acetabular [10]. Although the association between HIMRISS and disease activity was validated in SpA [11], the sensitivity and accuracy of HIMRISS in assessing treatment response in SpA with hip involvement need further validation.
Previous treatment strategy often focused on SpA global symptom management rather than SpA related hip lesions or other peripheral joint diseases [12,13]. Increasing evidence had indicated that early detection and diagnosis of hip in ammation is bene cial to active treatment which was important to improve hip function and general prognosis of SpA [14]. TNF-α inhibitor had been demonstrated to be prompt and robust in improving the signs and symptoms of SpA as well as function and spinal mobility [15,16]. In recent clinical studies, TNF-α inhibitor including etanercept, adalimumab or in iximab were also indicated to be effective for SpA patients with hip involvement in by clinical and imaging assessment [1].
However, the detailed clinical effect of TNF-α inhibitor among long-term treatment cycle need to be further observed, and previous imaging assessment were limited to radiographic assessment of BASRI-hip or ultrasound [17,18]. More accurate imaging tools such as MRI were still not nely applied in assessing the acute and long-term followup treatment effect of TNF-α inhibitor.
Thus, the purpose of this prospective study was to quantitatively assess the detailed clinical remission and MRI changes of hip involvement under the treatment of TNF-α inhibitor versus controls. The hip in ammation was quantitatively assessed by HIMRISS based on MRI changes, as well as the systematically clinical evaluation. We aim to give a suggested assessment standard for SpA patients with hip involvement including MRI and clinical, and guide accurate and targeted treatment.

Study population
Our study was based on data from Xijing hospital in China. All the SpA patients were enrolled with clinical diagnose, and the long-term treatment and tracking were initiated by rheumatologists in Xijing hospital. Eligible patients were 18-65 years of age and had to fullfill the recently published Assessment of SpondyloArthritis international Society (ASAS) classi cation criteria [19] or the Modi ed New York criteria for axial spondyloarthritis [20]. Both acute and chronic in ammatory changes on MRI were considered positive signs for hip involvement and coxitis. All the patients were included and under the basic treatment of MTX (10mg every week) and SSZ (2g/day given orally) from baseline to 52 weeks. Of all the eligible participants, patients in TNF-α group were allocated to three kind of TNF-α inhibitor plus basic treatment (etanercept 50 mg given once weekly subcutaneously, adalimumab treatment [40 mg s.c. every other week], in iximab [3 mg/kg iv. at baseline, week 2 and week 6 and then every 8th week]), and patients in the control group received the basic treatment only. The cost of treatment was self-funded, and the therapeutic options were determined on the consent of the doctors and patients. The study was approved by the ethics committee of Xijing Hospital of China, and all the participated patients had written the informed consent before enrollment in this study.
ESR, CRP, and human leukocyte antigen HLA-B27 was measured with standard laboratory techniques.
Serum CRP and ESR were measured at baseline, 12 week, 24week and 52 week At the same time-point, evaluation of the disease activity levels (BASDAI, ASDAS-CRP and ASDAS-ESR) and hip function level (harris) were conducted by experienced rheumatologists .
Quantitative MRI assessment of changes HIMRISS was used to assess the MRI changes between baseline and week 52. MRI was performed with coronal short-tau inversion recovery (STIR) sequence with slice thickness of 4 mm, eld of view 400 × 400 mm. The baseline and at least once follow-up MRI images of both hips were recommended, and the time interval between twice MRI images should be greater than 3 months. The HIMRISS protocol was used to score the in ammatory changes in MRI images of the hip joints including BME of femoral head, BME of acetabular and synovitis [11]. The HIMRISS procedure was performed by two experienced readers (ZK and WwT) independently. The procedure was blinded to treatment arm and MRI time point, and analyzed by independent statistician. The standard HIMRISS scores were calculated based on 15 MRI image slides, therefore, in order to avoid bias, patients with MRI images of hip structure <5 slides were considered as missing samples.

Statistical analysis
The Kaplan-Meier test was used to assess the difference of clinical symptom remission(ASAS20, ASAS40, ASAS50 and ASAS70) and MRI improvement between different follow-up time points. The baseline status was taken into account and a non-parametric analysis of covariance (non-parametric ANCOVA) was used. The statistical analyses were performed using SPSS 22.0. P values <0.05 were considered as statistically signi cant.  (Table 1). In TNF-α group, 103 patients received etanercept treatment, 51 patients received adalimumab treatment and 11 patients received in iximab treatment. The SpA patients in both TNF-α inhibitor group and control group presented with 52-weeks observation period with clinical symptom and MRI evaluation. As shown in Table 1, the difference of demographic and clinical characteristics including ESR, CRP, Harris, BASDAI, morning stiffness time, and ASDAS-CRP between two groups were not signi cant as baseline, except for ASAS-ESR (p<0.01) and disease duration (p<0.001).

Results
Patients in TNF-α inhibitor group had relative shorter disease duration and worse disease activity than control group. All patients had serious disease activity with higher BASDAI score (5.76±1.20 and 5.53±1.29) and ASDAS score (2.78±0.62 and 2.53±0.57), as well as elevated CRP and ESR values. Quantitative magnetic resonance imaging assessment The MRI changes of hip in ammation of those SpA patients were assessed by HIMRISS scoring system including BME and synovitis. Compared with baseline, patients in TNF-α group had signi cantly MRI improvement of both BME and synovitis at week 52 (all the p<0.001). However, the MRI difference in control group between week 52 versus baseline was relatively less(BME of femoral head: p<0.05. total BME: p <0.05. synovitis: p <0.05. total mean/max HIMRISS: p <0.05), and the BME scores of acetabular were similar (p>0.01)( Table 2).
Inter-group analysis showed that there is little difference between the total HIMRISS scores of TNF group and control group in baseline (p>0.05) and week 52 (p>0.05). however, the treatment group with higher total HIMRISS scores as the baseline demonstrated signi cant improvement at week 52, and with lower scores compared with control group( TNF-α vs control at week 52: 20.59±9.41 vs 22.33±7.0). Additionally, BME scores of acetabular indicated that TNF-α group were signi cantly more serious in baseline (p=0.006), but got signi cantly improvement at week 52 (p<0.001) ( Table 2).  (Table 3). However, the mean BASDAI score, ASDAS_CRP, ASDAS_ESR, and harris scores of TNF-α group were signi cantly better than control group at week 52 (all the p values<0.000). Especially, the degree of ASDAS_CRP indicated a real-world clinical character that, patients in TNF-α group had relatively serious disease activity in baseline, but obvious better remission at week 52. ASAS remission rate of ASAS20, ASAS40, ASAS50, and ASAS partial were compared between TNF-α group and control group at week 52. TNF-α group had higher and better ASAS remission levels than control group (Figure 1). Moreover, the distribution of disease activity including BASDAI, ASDAS_CRP, and  (Figure 2).
In addition, a subgroup analysis for all patients in TNF-α group was performed to compare the treatment difference between two kinds of TNF-α inhibitors. Firstly, for all patients receiving treatment of TNF-α inhibitor, all the continuous parameters related to disease activity (i.e. CRP, ESR, BASDAI, and ASDAS) and hip joint function (i.e. harris) ameliorate signi cantly from week 12 (all the p values<0.001) compared with the baseline, and persist the signi cant remission level until week 52 (all the p values<0.001) ( Figure  3).
Moreover, the ASAS remission rate including ASAS20 remission rate, ASAS40 remission rate, ASAS50 remission rate, and ASAS partial remission rate were compared between different TNF-α inhibitors in follow-up time. Results indicated that the difference of ASAS remission rate between different TNF-α inhibitor groups during the 52 weeks were not signi cant (p values>0.01) (Figure 4).

Discussion
This study demonstrated that TNF-α inhibitor could achieve signi cant clinical remission and MRI manifestation changes for hip arthritis in SpA patients. In TNF-α group, the disease activity indexes including BASDAI, ASDAS, ASAS remission rate, and hip function indexes of harris scores demonstrated better improvement compared with the control group. Additionally, HIMRISS method detected the obvious better improvement of the hip in ammation based on overall MRI slices under TNF-α inhibitor treatment than control treatment. TNF-α blocking therapy were con rmed as a better treatment options for SpA patients with hip involvement.
Previous studies of therapy in SpA mainly focused on the axial spine radiology, and to some extent, on peripheral arthritis or enthesitis [5,21]. Hip involvement was di cult to cure, while it was often considered as an important cause of disability and poor prognostic in SpA, especially in AS [17,22]. More attentions turned to focus on the hip arthritis in SpA, and varies clinical studies including TNF-α inhibitor treatment or classical DMARDs treatment were conducted to observe the clinical and/or radiographic progress of hip joint and function. [23][24][25]. TNF-α inhibitor treatment had been demonstrated to be effect for SpA patients with hip involvement. However, the speci c effect difference between TNF-α inhibitor and other agents need to be evaluated in long-term follow-up.
In known therapeutic studies of SpA patients with hip involvement, BASRI-h method was often adopted to study radiological changes of hip lesions [24]. However, most of the hip involvement in AS belongs to early coxitis characterized by in ammation of the subchondral bone marrow edema and synovitis [7,26], while BASRI-h method were performed based on structural damages which may not sensitively re ect the treatment effect to those patients in early and acute in ammation stages in short-time follow-up.
Comparatively, MRI was considered as a more sensitive assessment tool for hip in ammation, and is especially helpful in early diagnosis[8, 23,27].
Recent studies had conducted that synovial enhancement by heperintense signals on contrast-enhanced T1 weighted images was the most frequent imaging nding in early-stage hip involvement in AS [14]. Joint effusion of femoral neck and subchondral bone marrow edema were the frequent ndings by heperintense signals on T2WI or STIR images [8,27]. However, lesion frequency analyze or character description could not quanti cationally re ect MRI changes under treatment. HIMRISS could systematically quantify the in ammatory changes including BME of femoral head, BME of acetabular, and synovitis effusion of hip based on global MRI slices of patient [10]. HIMRISS method had been previously applied in SpA patients diagnosed with hip arthritis and indicated signi cant correlation between MRI manifestation and harris score or disease activities [11].
In known clinical studies, the clinical manifestations assessed in SpA patients with hip involvement mostly refer to global outcome indexes including disease activity scores and in ammatory markers.
Despite of the fact that the association between hip impairment and axial disease of in ammatory back pain were still argumentative [22], more evidence pointed to the fact that hip disease more frequently occur in SpA patients with severe axial disease [28]. However, the treatment e cacy changes of hip involvement in early-stage and long-term followup were still not fully understood, more detailed assessment of clinical and MRI manifestation need to be performed.
In this study, the clinical e cacy of TNF-α inhibitor for hip involvement was compared with control agents in a one-year follow-up periods. TNF-α was demonstrated to be signi cantly effective to improve disease activity and in ammation and the effect was signi cantly better than control. All the clinical indexes of TNF-α group including ESR, CRP, BASDAI, and ASDAS-CRP/ASDAS-ESR signi cantly ameliorated at different follow-up time compared with baseline, this result kept consistent with previous studies. Harris score was also signi cantly increased which re ected an outstanding improvement of hip function.
In this study, HIMRISS method was used to semi-quantify the acute and chronic in ammatory changes of hip arthritis of SpA under treatment of TNF-α inhibitor or control. HIMRISS results showed a signi cantly better MRI improvement of TNF-α group than control group at week 52. All HIMRISS parameters of TNF-α group including BME of femoral head, BME of acetabular, synovitis effusion and total HIMRISS scores got signi cant improvement with p values < 0.001.
The result indicated that, to some extent, TNF-α inhibitor could simultaneously achieve to signi cant longterm MRI remission and clinical remission, rather than control group could not notably achieve MRI remission. TNF-α inhibitor indicated the outstanding advantages for SpA patients with hip involvement.
An addition subgroup analyses further compared the clinical effect between different TNF-α inhibitors. Previous study had indicated that etanercept and adalimumab treatment could got similar ASDAS-CRP improvement in AS patients [16]. In this study, different ASAS remission rate were compared between etanercept/in iximab and adalimumab in different follow-up time points. The Kaplan-Meier test result indicated that the difference of remission rate including ASAS20, ASAS40, ASAS50, and ASAS partial were not signi cant between different TNF-α inhibitors. TNF-α inhibitors had similar high effect on SpA patients complicated with hip involvement.
The follow-up cycle of hip joint function and clinical characters varies from 12 weeks to 24 months in previous studies. In this study, we observed the treatment response from 12 weeks to 52 weeks. Lian et al and Huang et al [23]demonstrated that TNF-α inhibitor could maintain stable hip function by harris score and clinical remission of hip arthritis in SpA patients for 6 months. In this study, we simultaneously compared the effectiveness of TNF-α inhibitor and SSZ therapy on the clinical outcome of hip arthritis in SpA from baseline to 52 weeks. Our result further con rmed that, the clinical remission levels at week 52 of TNF-α inhibitor group were signi cantly better than SSZ group in all clinical parameters including ESR, CRP, harris, BASDAI, and ASDAS. TNF-α inhibitor treatment got a quick clinical remission from 12 weeks and hip function improvement, and maintained the signi cant improvement tendency until week 52.
Based on the fact that the economic cost of TNF-α inhibitor was relatively higher than classical DMARDs, the results indicated a real-world characteristic at baseline that patients in TNF-α group had relatively more serious disease activity and shorter disease duration than control group which was consistent with previous studies [29]. Those SpA patients with active hip arthritis were suggested to receive TNF-α inhibitor therapy to get quick pain reduction and symptom remission based on the global effectiveness of TNF-α inhibitor [30]. Similarly, the baseline result of MRI analysis also did not ful ll the criteria of no difference between two groups. Especially of the parameters about BME score of acetabular, total BME score, and effusion-synovitis. The real-world data revealed that, both disease activity and MRI features of patients under treatment of TNF-α inhibitor were relatively more serious. Those patients were suggested to TNF-α inhibitor therapy in order to quickly improve clinical outcome, relief pain and reduce adverse reactions. As a consequence of those real-world characteristics, the results must be interpreted with caution.
However, some limitations of this study should be acknowledged. Firstly, one-year treatment and observation period is short, prolonged followup need to be observed. Next, This is a single-centre study, and population bias may limit the e ciency. Moreover, although HIMRISS method was performed based on accurate MRI slices, it was not commonly used and need more application and con rmation in the future.

Conclusion
The quantitative assessment results demonstrated that TNF-α inhibitor could signi cantly simultaneously improve clinical outcome and MRI response of hip arthritis in SpA patients. This improvement could quickly achieve in 12 weeks treatment period and maintain in long-term follow-up, which achieved advanced curative effect than non-TNF-α agents.

Declarations
Ethics approval and consent to participate The study was approved by the ethics committee of Xijing Hospital of China (Instruction number: 20110303-7).

Consent for publication
Not applicable.
Availability of data and materials The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
Competing interests Figure 1 The distribution of ASAS remission rates between TNF-α and control group at week 52. The clinical performance of TNF-α inhibitor group from baseline to week 52.