Micro-structural bone changes in early rheumatoid arthritis persist over 1-year despite use of disease modifying anti-rheumatic drug therapy

Participants attended baseline and 12-month in-person evaluations. Physical Evaluations: Measures of height (cm) and weight (kg) [Body Mass Index (BMI): Kg / m2] were collected for all participants and a 28-joint active (Tender and Swollen) joint count for RA participants [32]. Self-Reported Measures: Participants completed a General Health and Lifestyle questionnaire. RA participants also completed a Stanford Health Assessment Questionnaire - Modified (mHAQ) [33]. HR-pQCT Imaging: The imaging protocol has been described in detail elsewhere [19]. Briefly, we acquired HR-pQCT images with a Scanco XtremeCT imaging system [Scanco Medical AG, Switzerland] using standard manufacturer recommended parameters [19]. For the radius, the reference line was the medial/distal radius (Fig. 1a – Scout View). The scan started 3 mm proximal to this reference line and extended 9.02 mm (110 slices) proximally (Fig. 1a – Scout View). For the metacarpal head the reference line was the tip of the most distal second or third metacarpal head. The scan started 4 mm distal to this reference line and extended 18.04 mm (220 slices) proximally [19, 20]. (Fig. 1a -Scout View). Each 110-slice scan takes 2.8 min with an effective dosage of less than 2 μSv [19].

At baseline an internal medicine resident extracted information from rheumatologists’ electronic medical records including; RA diagnosis date, timing and type of prescribed RA medication(s), and RA blood markers [anti-citrullinated peptide autoantibodies (ACPA) and/or Rheumatoid Factor (RF) Positive] at time of diagnosis. At 12-months, RA participants completed a log of RA medications prescribed and taken in the previous 6 months. The cumulative dosage of any DMARD or GC medications were not collected at either time point.

One of three trained operators [intra-rater reliability, 10 scans measured twice, Pearson’s r > 0.9] analyzed all images using manufacturer’s evaluation software (SCANCO, V 6.0) [19]. Prior to analysis, each image was graded for motion artifact using a 5-point grading scale and only images rated 3 or higher were used [19]. Regions of Interest included the ultra-ultra-distal radius (UUD: 110 slices, starting 3 mm proximal to radius reference line and running proximally), metacarpal head two (MH2: 110 slices starting at the distal tip of MH2 running proximally), and metacarpal head three (MH3: 110 slices starting at the distal tip of MH3 running proximally) (Fig. 1a – Scout View) [19]. We ran standard manufacturer and direct transformation image analyses scripts to segment the cortical and trabecular bone regions and measure bone density and microstructure (Fig. 1b, c – Cortical/Trabecular Compartment Segmentation) [19].

Density measures included apparent bone mineral density (BMD - mgHA/cm³) for cortical and trabecular bone regions, and cortical bone material bone density (TMD - mgHA/cm³). Measures of cortical microstructure included: thickness (CtTh - mm), thickness variability CtThSd - mm), volume fraction (BV/TV_cort - %) and porosity (CtPo - %). Measures of trabecular microstructure included: volume fraction (BV/TV_trab - %), number (TbN – 1/mm), thickness (TbTh - mm), separation (TbSp - mm), separation variability (TbSpSd - mm), connective density (TbCD - mm⁴) and structural model index (SMI _0–3; lower values indicate more plate-like verses more rod-like structure) [19].

We examined differences in baseline anthropometrics (Age, Sex, BMI) and Fracture Risk between RA and NRA participants using Paired Student T-tests (two tailed, p < 0.05). For the longitudinal analyses of micro-structural bone quality, we conducted a General Linear Modelling 2 × 2 repeated measures analyses for the HR-pQCT image analyses. We compared disease status (RA vs NRA) over Time [Baseline vs 12 Month] using a two-tailed analysis with a Sidak multiple comparison adjustment. We set alpha at p = 0.05. We examined main effects for disease (RA vs NRA, independent of time), time (change over 1-year, independent of disease status) and interaction [disease (RA vs NRA) x time (baseline vs 12- months)]. We did not correct for separate statistical analyses for multiple HR-pQCT imaging outcomes. All statistical analyses were completed using SPSS software v. 23 (IBM Corp, Armonk, NY).

Forty-nine individuals diagnosed with RA in the previous year by one of nine rheumatologists were screened for eligibility, with 19 excluded (39% excluded). Reasons for exclusion were RA diagnosis at the time of screening greater than 1 year (n = 10), not able to commit to study (n = 6), or language barrier (n = 3). Of 43 NRA individuals screened eight were excluded (19% excluded) because they were unable to commit to the study (n = 5) or they reported a co-morbid inflammatory health condition (n = 3). Of the 35 eligible NRA participants five were ultimately excluded as they were not matched with an RA study partner. The 60 participants were serially recruited to the study and evaluated at baseline over an 11-month time period. Fifty-six participants (26 RA/30 NRA) completed the 12-month evaluation (93% completion); 54 were evaluated at 52-weeks (+/− 4 weeks) and 2 were evaluated at 60 weeks. Four RA participants withdrew [1 death, 2 serious illness (cancer, cardiac disease), 1 no longer interested] and their 4 sex/age-matched NRA partners were excluded from the final longitudinal analyses. Five images were excluded from analyses due to motion artifact at baseline (4.2%; 2-MH, 3-UUD) and at 12-months (9.6%; 4-MH, 1-UUD). Of the 26 matched pairs of participants at 12 months (n = 52); 23 pairs (88.5%; n = 46) had baseline and 12-month UUD images and 22 pairs (84.6%; n = 44) had baseline and 12-month MH images for comparison.

See Table 1 for further details of participant demographics at baseline. In summary, of the 60 participants, 48 were females with a mean age of 53 years varying from 21 to 74 years. The RA and NRA pairs were well matched by age and sex, with exact matching for sex and no statistically significant difference in age, with the RA participants on average 53 years old, compared to the NRA participants 52 years old. On average, female RA and both RA and NRA males were overweight (BMI: 25–29.9 kg/m²), whereas, NRA females were of high normal body weight (BMI 24.8 kg/m²). [34] The mean BMI was significantly higher in the RA group for women, but not for men. The mean Fracture Risk Assessment Tool (FRAX®_Canada) 10-year major fracture and hip fractures risk scores were both significantly higher among participants with RA [35]. Some other differences at baseline included, 6 RA participants smoked compared to 1 NRA participant, 7 RA participants compared to 1 NRA participant had been told they may have poor bone health (i.e. osteopenia or osteoporosis) and 3 individuals with RA reported taking a bone antiresorptive or anabolic medication in the last 5 years [36]. As well, 19 RA participants reported taking calcium or vitamin D nutritional supplements compared with 13 of NRA participants.

See Table 2 for further details of RA participant characteristics. In summary, at baseline the 30 RA participants were on average 7.7 months (varying from 1 to 15 months) since diagnosis. Notably, one RA participant was scanned at 15-months post diagnosis which was a protocol violation. This occurred as the participant did attend the baseline evaluation at 12-months post diagnosis, however, they could not complete the evaluation due to physical illness. Unfortunately, the re-evaluation was subsequently delayed due to a 3-month planned vacation. Twenty-two of the RA participants were ACPA and/or RF positive at the time of diagnosis. At baseline, 29 of 30 RA participants received one or more non-biologic DMARD, including Methotrexate, Hydroxychloroquine, or Sulfasalazine. Of these, 14 received single, 7 received double and 8 received triple DMARD therapy. Thirteen individuals received only DMARD therapy, 15 received DMARDs in combination with oral glucocorticoid (at least one episode of Prednisone, > 5 mg/day, > 3 weeks), 1 received DMARD in combination with an anti-TNF (Tumor Necrosis Factor) biologic medication (Adalimumab) started 8-months post diagnosis and 1-week prior to baseline imaging and 1 person having only received glucocorticoid medication. DMARD therapies were started on average 0.1 month after diagnosis, whereas, oral glucocorticoid (GC) medication was initiated on average 1.8 months prior to final RA diagnosis. The negative value for onset of GC medication can be explained in part by a typical 4 to 6-week time interval between the initial visit to the rheumatologist and final diagnosis of RA, when GC medications are often started prior to definitive RA diagnosis. Additionally, one participant had received a GC medication prescription (> 5 mg/day for 3 weeks) from their primary care physician 11-months prior to the referral to a rheumatologist. At 12-months, 24 of 26 RA participants were still receiving DMARD therapy. However, only 2 had taken any oral glucocorticoid medications in combination with DMARDs in the previous 6 months and 19 had tapered down to single DMARDs therapy. Whereas, one person was taking only an anti-TNF biologic medication (Adalimumab), while another had opted out of taking all prescribed RA medications as a personal choice.

See Table 2 for further details of RA participant clinical characteristics. This study used a 28-joint active count as a measure for evidence of active joint inflammation (i.e. no or some active joints) at the time of imaging, with active inflammation of any joint defined as both swollen and tender. At baseline, 9 of 30 RA participants presented with no active joints. Of the remaining 21 RA participants with at least one active joint, the median active joint count was 4. At 12-months, 15 of 26 RA participants presented with no active joints. Of the remaining 11 participants with at least one active joint, the median active joint count was 1. At baseline RA participants reported low levels of functional disability [DI_1–3: Mean 0.59 (+/−0.60)], pain (VAS_0–100: Mean 21.2 (+/− 16.4) and impact on well-being [VAS_0–100: Mean 23.7 (+/− 18.8)] [33]. Notably, RA participants also reported minimal improvement in their self-reported functional disability, pain or impact on well-being over 1-year [37].

See Table 3 for details of the values and results of the longitudinal statistical analyses for the HR-pQCT disease main effect, representing the overall differences between RA and NRA participants independent of time at all three sites. In summary, differences in density included: RA participants had significantly lower trabecular bone density at all three sites (varying from 15.4 to 10.9% lower), significantly lower cortical bone apparent density at both MH sites (MH2 8.7% and MH3 9.9% lower) and significantly lower cortical bone material density at the MH2 site (MH2 3.9% lower). Differences in cortical bone micro-structure included: At the UUD site RA participants demonstrated significantly greater variability in cortical thickness (9.3% greater variability in thickness) and significantly less cortical bone volume at both MH sites (MH3: 3.8% MH2: 4.5% lower). Trabecular bone micro-structure differences included: At all three sites RA participants had significantly larger and more variable sized spaces between trabeculae (varying from 11.9% to 16.0% larger spaces and 22.9% to 35.2% more variable sized spaces). At all three sites RA participants trabecular matrix was also significantly more rod- verses plate-like shaped trabecular matrix (SMI varied from 20.8% to 84.1% greater) with significantly lower trabecular bone volume (varying from 12 to 8% lower). At the UUD and MH2 sites, RA participants also had significantly fewer trabeculae (UUD 7.2% lower; MH2 9.0% lower). Whereas, at the UUD site RA participants also had trabecular that were less connected (13.0% less trabecular connectivity). Figure 2 shows plots (mean +/− SEM) for selected density and microstructural variables with consistent differences between RA and NRA participants across the three ROIs examined in this study, as well as, notable differences in the density and micro-structure values of the two MH head ROIs and the UUD ROI. To illustrate further the differences between RA and NRA participants, Fig. 3 shows the typical visual differences in trabecular bone micro-structure seen in 3-Dimensional reconstructed images of the UUD and MH sites in age- and sex-matched RA and NRA study partners.

See Table 4 (Additional file 1) for details of the data values and results of the statistical analyses for the longitudinal time main effect analysis, representing the aging effect over 12-months for all participants independent of disease status. In summary, density changes over time included: Cortical bone material density at the MH3 site and UUD trabecular bone apparent density site were significantly reduced over 12-months (MH3 0.6% lower, UUD 2.5% lower). Changes in cortical bone micro-structure over12-months included: UUD cortical thickness and porosity significantly increased over 12-months (Thickness 4.7% greater; Porosity: 29.6% greater). Trabecular bone micro-structure changes over 12-months included: At the UUD site, trabecular number, connectivity and SMI all increased significantly across one year (Number: 1.9% greater, Connectivity 2.7% greater and SMI 3.4% greater). Whereas, trabecular thickness, trabecular spacing and variability were significantly lower (Thickness: 1% lower, Spacing 1.7% lower and variability 1.8% lower). Additionally, at the MH3 site, trabecular bone volume, thickness and thickness variation all increased significantly (volume 0.9% higher, thickness 0.7% higher and thickness variation 1.8% higher).

See Table 5 (Additional file 1) for full details of the data values and results of statistical analyses for the RA vs NRA by Baseline vs 12-months (interaction) main effect analyses. There was no significant interaction main effects for any cortical or trabecular bone density or microstructure outcomes examined, which indicates that the rate of change over 12-months was not different for the RA and NRA participants. Or alternately, that the RA participants did not show either an increased or decreased rate of change in bone micro-structure over the 12-months relative to the NRA participants, indicating as well that any underlying difference in bone micro-structure at baseline in the RA participants relative to the NRA participants persisted over the 12-months.