Ultrasound-detected bone erosion is a relapse risk factor after discontinuation of biologic disease-modifying antirheumatic drugs in patients with rheumatoid arthritis whose ultrasound power Doppler synovitis activity and clinical disease activity are well controlled

Forty patients with RA who fulfilled the 2010 RA classification criteria [15] and discontinued bDMARD therapy during the period from November 2010 to March 2015 were consecutively recruited to participate this study from three rheumatology centers in Nagasaki Prefecture (Nagasaki University Hospital, Isahaya General Hospital, and Japanese Red Cross Nagasaki Genbaku Hospital). All of the patients satisfied the following inclusion criteria. All patients (1) had been treated with a bDMARD for at least 6 months, (2) were in sustained low disease activity (LDA) or remission (i.e., with a Disease Activity Score in 28 joints based on erythrocyte sedimentation rate [DAS28-ESR] <3.2) for at least 3 months, and (3) and had not taken oral glucocorticoids at the discontinuation of bDMARD therapy. Each patient had undergone an ultrasound assessment of 22 joints of bilateral hands at discontinuation, and the maximum grade of PD at each joint was ≤1. All patients underwent clinical and laboratory assessments every month after discontinuation for 12 months. This study was a retrospective observational investigation using anonymized information. The participants had discontinued bDMARD therapy depending on the clinical decision before we approached to take part in this study. The patients gave their informed consent to be subjected to the protocol, which was approved by the institutional review board (reference number 10062546) of Nagasaki University Hospital.

Clinical and laboratory assessments included 28-joint swollen and tender joint counts; the evaluator global assessment (EGA); the patient global assessment (PtGA); the Health Assessment Questionnaire-Disability Index; the ESR; and the levels of C-reactive protein (CRP), matrix metalloproteinase 3 (MMP-3), rheumatoid factor (RF), and anticyclic citrullinated peptide antibody (ACPA). Clinical disease activity was evaluated by the DAS28-ESR, the Simplified Disease Activity Index (SDAI), and the Clinical Disease Activity Index (CDAI).

The ultrasound examinations were performed by Japan College of Rheumatology (JCR)-certified sonographers (SK, KF, AN, TA, and TS). SK is also involved in the JCR Committee for the Standardization of Musculoskeletal Ultrasonography. A systematic multiplanar grayscale (GS) and PD examination of joints was performed with different scanners (Aplio 500, Toshiba, Tokyo, Japan; Ascendus, Hitachi, Tokyo, Japan; Noblus, Hitachi; or LOGIQ E9, GE Healthcare, Wauwatosa, WI, USA), using a multifrequency linear transducer (12–14 MHz). All of the scanners are high-end models. On each scanner, the factory setting for superficial musculoskeletal assessment was used. The settings were also adjusted for increasing Doppler sensitivity by decreasing the pulse repetition frequency (800 Hz) or velocity range (3.1 cm/second) and adjusting the Doppler gain to a level just below random noise. Thus, appropriate scanning conditions have been considered to be achieved even if different scanners areobtained. Articular synovitis was assessed by ultrasound at the 22 joints, including the bilateral wrist joints (radiocarpal, intercarpal, and distal radioulnar joints) as well as the first to fifth metacarpophalangeal (MCP) and proximal interphalangeal (PIP) joints. Each joint was scored for GS as well as PD on a scale from 0 to 3 [16], and the sum of the GS or PD scores was used as the indicator of ultrasound disease activity. In previous studies, we demonstrated that the present synovitis scoring system was useful for evaluating the early diagnosis and therapeutic effectiveness of medications for patients with RA [5, 17‐19]. Positivity of GS and PD in articular synovitis was defined as a score ≥1 for GS and PD, respectively.

We also examined the presence or absence of bone erosion in the present study. In addition to synovitis, ultrasound bone erosion appears to be predictive of further radiographic progression [4]. Erosion was defined by a cortical break seen in two perpendicular planes [20]. Clearly visible erosions in both longitudinal and transverse scans were considered for the study. We scanned the dorsal, volar, and lateral aspects of the involved joints (wrist, MCP, or PIP). The sonographer paid particular attention in assessing unifocal bony breaks of small size to avoid misinterpretation with anatomical necks or vascular bone channels [21, 22]. Vascular bone channels were distinguished from bone erosion on the basis of anatomical location, insertion of feeding vessels detected by PD, and absence of adjacent synovial lesion. All joint regions were sonographically examined in a standardized manner according to the EULAR [23] and JCR guidelines on a semiquantitative scale as described previously [16].

Patients were assessed monthly by JCR-certified rheumatologists and received routine clinical management after the discontinuation of bDMARD therapy as described above. Patients were considered to have had a relapse when their antirheumatic treatment was escalated because the physician concluded that the patient was exhibiting an increase in RA disease activity. Although the physicians had recognized the ultrasound findings at discontinuation of bDMARD therapy, they did not decide on whether a relapse had occurred solely on the basis of ultrasound findings.

Within-group comparisons were made using the Mann-Whitney U test or the χ² test. We performed a logistic regression analysis to investigate the relationships between the variables at the initiation or discontinuation of the bDMARDs and relapse during the 12 months after the discontinuation of the patient’s bDMARD treatment. For the multivariate logistic regression analysis, we selected the variables that showed p values <0.2 (Table 1). The overall significance level for the statistical analysis was 5% (two-sided). p Values <0.05 were considered significant.

Forty patients (33 females, 7 males) were enrolled in this study. The median (IQR [Q_1-4-Q_3/4]) patient age was 54.5 (45.8–61.3) years, and the median disease duration was 3.5 (5.5) years. The median (IQR) values were as follows: tender joints count 0 (0), swollen joint count 0 (0), PtGA 4 (2–8) mm, EGA 3 (2–5) mm, ESR 10 (5–15) mm/h, CRP 0.05 (0.03–0.06) mg/dl, MMP-3 33 (26–55) ng/ml, DAS28-ESR 1.63 (1.25–2.01), SDAI 0.9 (0.5–1.4), and CDAI 0.8 (0.4–1.3). The rate of remission (DAS28-ESR less than −2.6) was 92.5%. The rate of absence of PD (total PD score 0) was 80%. Bone erosion was detected in 24 joints (17 wrist and 7 MCP joints) and in 13 patients. It was not detected by conventional radiography in 10 (41.7%; 5 wrist and 5 MCP joints) of 24 joints and 6 (46.2%) of 13 patients. Thirty patients had been treated with tumor necrosis factor (TNF) inhibitors, and the other ten patients had been treated with non-TNF inhibitors.

Nineteen (47.5%) patients relapsed during the 12 months after discontinuation of their bDMARD therapy. The time course of the relapse-free periods by Kaplan-Meier survival estimate is shown in Fig. 1. Among the patients who relapsed, most (15 of 19 patients [78.9%]) relapsed within 6 months after bDMARD discontinuation. Treatment with a bDMARD was reintroduced in 13 patients. Treatment with conventional disease-modifying antirheumatic drugs (cDMARDs) was increased for three patients and added for two patients. A low-dose corticosteroid was added for one patient.

As shown in Table 1, no clinical variables, including seropositivity, composite measurements, and treatments at initiation and discontinuation, were significantly different between the relapse patients and nonrelapse patients.

We performed multivariate logistic regression analysis to identify any variables that could be used to predict relapse during the 12 months after discontinuation of bDMARD therapy (Table 2). Only the presence of bone erosion detected by ultrasound at bDMARD discontinuation was an independent predictive variable for relapse (OR 8.35, 95% CI 1.78–53.2, p = 0.006).