Patients and study design
Two open-label investigator-initiated clinical trials (NCT01717859, NCT02321930) recruited RA patients from two university-based rheumatology clinic sites following institutional review board approval (IRB#12–001547, IRB#14–001148) and appropriate patient consent obtained. While the primary endpoint of these two studies aimed to examine early changes in PDUS, this post hoc study aimed to evaluate a subset of RA subjects who underwent prior joint surgery and quantify ultrasound synovitis of these operated joints. Participants’ joints were pooled into a single cohort and characterized as replaced joints, operated joint areas (OJA), or native joints (no prior surgery). OJA included joint arthroscopies, joint fusion, synovectomies, and tendon surgeries; these interventions were grouped together given their individual small sample sizes. Carpal tunnel surgeries and nerve transpositions were excluded. Patients at baseline were at least 18 years of age, met 1987 ACR criteria for RA, demonstrated disease activity score/erythrocyte sedimentation rate (DAS28/ESR-4 item) ≥ 3.2, and had cumulative Power Doppler score > 10 over 32 joints (see below for scoring). Patients also had received ≤10 mg prednisone and maintained stable concomitant DMARDs for at least 1 month. The 6-month tocilizumab study began with an infusion of 4 mg/kg of drug every 4 weeks and escalated to 8 mg/kg if DAS28/ESR-4 item was > 3.2 at 12 weeks, with patient and sonographer blinded to dosage step-up. The 3-month tofacitinib trial administered 5 mg of drug twice-daily by mouth. Only patients with replaced joints and/or OJA that completed the trial were included for analysis.
Ultrasound assessments
MSUS scanning was performed for tocilizumab patients (baseline and months 1, 3, 4, 6) and tofacitinib (baseline, 2 weeks, and 3 months) to assess disease activity and synovitis at pre-specified joints. MyLab70C US machine (Biosound Esaote, Fishers, IN) was used for image acquisition in the tocilizumab trial (12–18 MHz linear probe), whereas tofacitinib images were obtained using GE LogicE9 US machine (GE Healthcare, Chicago, IL) (6–15 MHz linear probe), as mandated in the respective parent clinical trials. MSUS assessments were conducted by one of two independent, experienced sonographer-rheumatologists by enrollment site (GK, VKR).
Sonography of some joints can be challenging due to lower incidence of involvement, lack of standardization of the optimal views in RA, and increases in depth that decrease sensitivity of PDUS detection. As such, there is no consensus on the number of joints to scan in RA [9]. Our ultrasound protocol included bilateral GSUS and PDUS images of 16 joints commonly assessed in other RA MSUS studies: dorsal long, dorsal short, and volar long views of metacarpophalangeal (MCP) 1–5, proximal interphalangeal (PIP) 2–5, and interphalangeal (IP); dorsal long midline views of radiocarpal-intercarpal wrist joints and dorsal long and short views of radioulnar wrist joints (wrist); dorsal long views metatarsophalangeal (MTP) 2–5; and medial/lateral parapatellar axial oblique views of the knees. B-mode scanning of replaced joints and OJA, including joint position and depth, were similar to that performed on native joints. In subjects with joint replacement, hardware artifacts could be recognized as well as material in the pseudocapsule. In most circumstances, Doppler signal was present within the visualized intracapsular material. Joints that could not be assessed by ultrasound (e.g. severe anatomical deformation) were excluded from the MSUS joint-level analyses.
Each joint view was scored on a previously standardized semiquantitative scale ranging from 0 to 3 [10, 11]. The maximum score of all views was selected for each joint. Images were de-identified to patient and date and scored independently by two experienced sonographer rheumatologists by enrollment site (GK, VKR) who were blinded to the sequence of visits, the patient, and the clinical assessment. PDUS inter-rater reliability was 0.77, and intra-rater reliability ranged from 0.82–0.89 (weighted Kappa). GSUS inter-rater reliability was 0.57, and intra-rater reliability ranged from 0.65–0.76 (weighted Kappa).
Clinical examinations
To capture clinical correlates for joints within and outside of the ultrasound protocol, a comprehensive, 68-joint count for tenderness and 66-joint count for swelling was collected by established convention at each visit and scored as absent (0) or present [1] for each joint. The clinical assessors were blinded to US data. A subset of 28 joints from these assessments also facilitated calculation of patient-level disease activity measures (DAS28 and Clinical Disease Activity Index [CDAI]). Replaced joints were excluded when calculating the DAS28/ESR-4 item and the CDAI. Joints not assessed for tenderness or swelling were excluded from joint-level analyses but were input as a score of zero for patient-level joint counts.
Statistical analysis
Measures of disease activity were computed at the patient- and joint-level. Joint-level GSUS and PDUS measures were compared between baseline and follow-up time points using Wilcoxon signed rank tests. McNemar’s test was used to compare joint-level dichotomous measures of PDUS≥1, GSUS≥1, tenderness, and swelling. Logistic regressions for PDUS≥1, GSUS≥1, tenderness, and swelling, adjusted for anatomical joint site, were used to verify that the models above were not affected by different distributions of anatomical joint sites across the three joint types. Mixed effects ordinal logistic regression models were used to compare joint-level GSUS and PDUS scores over time between joint types. These models included terms for joint type, time, and the time by joint type interaction. A random effect term was also included to account for clustering of joints within subjects. Spearman correlation was used to test for association between the time since surgical procedure and both the baseline and end-point PDUS and GSUS scores. In addition, subgroup sensitivity analysis was performed for each of the clinical trials (tocilizumab vs tofacitinib), as well as for small vs large/medium (wrists, knees [ultrasound was not performed on shoulders, hips, elbows and shoulders]) joints to examine differential effects.