To compare ultrasound-verified joint inflammation between elderly-onset rheumatoid arthritis (EORA) and younger-onset rheumatoid arthritis (YORA) patients.
To compare ultrasound-verified joint inflammation between elderly-onset rheumatoid arthritis (EORA) and younger-onset rheumatoid arthritis (YORA) patients.
We conducted a retrospective analysis of 145 consecutive rheumatoid arthritis patients routinely assessed by sonography of wrists, metacarpophalangeal joints, and proximal interphalangeal joints, including semiquantitative scoring of synovial hypertrophy/effusion (SH/E) and power Doppler (PD) signals. Global ultrasound (GU) scores were calculated adding SH/E and PD results. EORA was defined by disease onset at age ≥60 years. Number of tender joints and swollen joints, global assessment of disease activity by physician or patient, Disease Activity Score in 28 joints (DAS28), Clinical Disease Activity Index (CDAI), and Simplified Disease Activity Index (SDAI) scores were recorded. Respective values for disease activity were accounted for in group comparisons using SPSS statistical software (version 18.0).
Seventy patients were diagnosed with EORA (mean ± SD age 71.0 ± 7.3 years, 81.4% women) and 75 patients with YORA (mean ± SD age 46.8 ± 10.2 years, 86.7% women). EORA patients had higher GU scores (median 18.5 [interquartile range (IQR) 17.0] versus 12.0 [IQR 15.0], P = 0.009) and SH/E scores (median 12.0 [IQR 10.0] versus median 9.0 [IQR 9.0], P = 0.004) than patients with YORA. Patients with EORA were more likely to show PD signals in at least 1 joint than YORA patients (85% versus 72%; odds ratio 3.9 [95% confidence interval 1.3–11.5], P = 0.015). DAS28, CDAI, and SDAI scores did not differ between the groups. The sonographic pattern of joint involvement was similar in both groups, with active inflammation most commonly presenting at the wrists.
Ultrasound examination indicated higher inflammatory burden in EORA patients than in YORA patients despite similar clinical disease activity.
It can be expected that the number of elderly patients suffering from rheumatoid arthritis (RA) rises during the next years (1). The clinical presentation of elderly-onset RA (EORA; defined by disease onset at age ≥60 years) patients frequently differs from that of younger-onset RA (YORA) patients (2), i.e., EORA patients more often present with acute onset of disease, constitutional symptoms, and inflammation of large joints compared to YORA patients (2). The overall clinical disease activity as measured by composite scores, however, does not differ between the groups (3).
Musculoskeletal ultrasound is used more and more as a biomarker for diagnosis and followup of RA patients (4). Sonography has a high sensitivity to detect joint inflammation and RA-specific bone lesions, as well as a high specificity to exclude synovitis in patients with noninflammatory joint pain (4). In patients with established RA, it was demonstrated that up to 50% of patients still have active synovitis at ultrasound examination despite clinical remission (5) and, conversely, sonography excluded active inflammation in up to 20% of cases with painful joint swelling (6).
The specific knowledge about the value of ultrasound in the management of elderly RA patients is scarce, although the importance of sonography has been increasing for diagnosis and monitoring of RA (4). In the present work we compared ultrasound findings between EORA and YORA patients.
Sonography-detected joint inflammation is higher in elderly-onset rheumatoid arthritis (EORA) patients than in younger-onset rheumatoid arthritis (YORA) patients despite comparable clinical disease activity.
The pattern of ultrasound-verified inflammatory changes is similar in EORA and YORA patients.
We performed a retrospective analysis of 149 consecutive RA patients attending the Hospital of the Elisabethinen, Klagenfurt (Austria), as described previously (6, 7). Four patients were excluded as age at disease onset was not available. EORA and YORA were defined by disease onset at age ≥60 years and age <60 years, respectively. Patients underwent complete clinical assessment by the same trained nurse for the number of tender and swollen joints using the 28-count index, physician's and patient's global assessment of disease activity on a visual analog scale (range 0–100 mm), and the Health Assessment Questionnaire (HAQ). Composite index scores, including the Simplified Disease Activity Index (SDAI), the Disease Activity Score for 28 joints (DAS28), and the Clinical Disease Activity Index (CDAI), were calculated. Posteroanterior radiographs of both hands were taken within 2 weeks of ultrasound investigations and were assessed by an experienced radiologist (HW). The presence of ≥1 erosion and/or ≥1 osteophyte was considered as erosive disease and/or concomitant osteoarthritis, respectively. The retrospective analysis was accepted by the Carinthian Institutional Review Board.
Sonographic examination of 22 joints (wrists, metacarpophalangeal joints 1–5, proximal interphalangeal [PIP] joints 2–5, and interphalangeal joint 1 of both hands) was performed by 3 rheumatologists (C. Dejaco, C. Duftner, and EW-F) according to a standard ultrasound protocol as described previously (7). An Acuson Antares Siemens ultrasound machine (linear probe at 5–13 MHz and Doppler frequency of 5–8.9 MHz) was used.
Synovial hypertrophy and/or joint effusion (SH/E) were subjectively graded from 0 to 3, where 0 = no SH/E, 1 = minimal SH/E, 2 = moderate SH/E, and 3 = extensive SH/E. Power Doppler (PD) signals were also semiquantitatively assessed on a scale from 0 to 3 (where 0 = no PD signal, 1 = up to 3 single or 2 confluent vessels, 2 = less than half of the synovia, and 3 = half or more of the synovia covered by PD signals) (see Supplementary Figure 1 for examples, available in the online version of this article at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)2151-4658) (7). SH/E and PD were independently scheduled at dorsal and volar views, and the higher value was counted for the score. Scores were expressed per joint (possible range 0–6), and a global ultrasound (GU) score was calculated by the addition of all joint scores (possible range 0–132).
Statistical analysis was performed with SPSS software (version 18.0). Descriptive statistics were used to summarize the data. The Kolmogorov-Smirnov test was used to check normality of metric data. For data with a normal distribution, the mean ± SD is shown, and we applied Student's t-tests for comparisons. For non-normally distributed data, we indicate the median (range) and used the Mann-Whitney U test. Correlations were analyzed by Spearman's rank correlation coefficient and proportions by chi-square test. Multivariate linear (method of least squares) and logistic regression (maximum likelihood method) models were developed to investigate the possible influence of demographic and clinical factors (including sex, erosive disease, concomitant osteoarthritis, current therapy, disease duration, and DAS28 score) on ultrasound-verified inflammation.
Backward stepwise regression analysis was conducted, excluding parameters failing the prespecified significance level of 0.1. A square root transformation of GU score or SH/E (dependent variables) and disease duration (covariates) was used to normalize the distribution of the variables. Square root–transformed data were nearer to normality than data with other, particularly logarithmic, transformations. An interaction term (EORA*DAS28) was introduced in linear regression analysis to investigate the association between DAS28 scores and GU scores in the elderly RA group. Normality, linearity, and homoscedasticity of the regression model were checked, and goodness-of-fit was tested by Hosmer-Lemeshow statistics.
The following sensitivity analyses were performed: 1) construction of an inclusive model forcing parameters with a high likelihood of confounding (i.e., square root–transformed disease duration, therapy, and DAS28 score) irrespective of its individual significance level in addition to the predictor of primary interest (EORA/YORA) into the final model; 2) exclusion of 10–20% of influential cases with lowest/highest dfbeta statistics; 3) or exclusion of 10–20% of influential cases with largest Cook values; and 4) omission of patients treated with anti–tumor necrosis factor α (anti-TNFα) agents.
Out of the 145 RA patients, 70 patients were classified as having EORA and 75 patients as having YORA. Clinical characteristics are depicted in Table 1. Complete clinical, serologic, and ultrasound data were available in 54 patients in each group.
|Age at diagnosis, years||71.0 ± 7.3||70||46.8 ± 10.2||75||< 0.001|
|Female, no. (%)||57 (81.4)||70||65 (86.7)||75||0.39|
|Disease duration, months||2.0 (0–156)||70||26.0 (0–530)||75||0.001|
|Newly diagnosed, no. (%)||28 (40.0)||70||17 (22.7)||75||0.011|
|DAS28 score||4.5 ± 1.3||54||4.5 ± 1.3||56||0.79|
|CDAI score||19.1 ± 12.6||55||20.5 ± 12.1||62||0.55|
|SDAI score||21.2 ± 13.7||55||21.4 ± 12.1||62||0.95|
|Swollen joint count||1 (0–22)||55||2.5 (0–17)||62||0.25|
|Tender joint count||6.0 (0–26)||55||6.5 (0–26)||62||0.49|
|Patients' global assessment, VAS||45.2 ± 30.7||55||45.6 ± 29.8||63||0.94|
|ESR, mm/hour||28 (2–110)||69||18 (0–90)||70||0.003|
|CRP level, mg/dl||11.5 (0–169.0)||70||4.0 (0–94.0)||73||0.002|
|Erosive disease, no. (%)||37 (52.9)||70||43 (60.6)||71||0.36|
|Concomitant osteoarthritis, no. (%)||64 (91.4)||70||43 (58.9)||73||< 0.001|
|HAQ score||1.0 ± 0.75||55||1.0 ± 0.76||60||0.97|
|Current DMARD treatment, no. (%)|
|Methotrexate||42 (60.0)||145||46 (61.3)||145|
|Anti-TNFα therapy||2 (2.9)||145||13 (21.3)||145||0.005|
|Rituximab||0 (0)||145||2 (2.7)||145|
|Tocilizumab||0 (0)||145||2 (2.7)||145|
The prevalences of ultrasound and clinical findings at different joint regions are detailed in Table 2. PD signals were most frequently observed at wrists in both EORA and YORA patients, whereas SH/E, as well as clinical findings, were most prevalent at PIP joints in both groups. Wrists and the second and third PIP joints most frequently revealed sonographic signs of inflammation (SH/E, as well as PD signals); the prevalence was 67.1–92.9% in EORA cases and 57.3–84.0% in YORA cases (differences not significant). Symmetric sonographic joint involvement was more common at wrists from EORA than from YORA patients (75.7% versus 50.7%; P = 0.008 for SH/E and 28.6% versus 13.4%; P = 0.06 for PD), whereas the frequency of symmetric ultrasound findings at other joints was similar in both groups.
|%||Median (range)||%||Median (range)|
|MCP||70.0||2 (0–10)||56.0||1 (0–10)|
|PIP||90.0||3 (0–8)||76.0†||3 (0–10)|
|MCP||32.9||0 (0–8)||30.7||0 (0–10)|
|PIP||65.7||1 (0–8)||54.7||1 (0–8)|
|MCP||63.0||1 (0–10)||64.5||1 (0–10)|
|PIP||72.2||2 (0–10)||72.6||2 (0–10)|
|Wrists||34.3||0 (0–9)||46.9||0 (0–5)|
|MCP||35.2||0 (0–10)||51.6||1 (0–10)|
The median number of joints per patient revealing SH/E and/or PD signals was higher in EORA than in YORA patients (8.0 [range 0–20.0] versus 5.0 [range 0–20.0]; P = 0.001, and/or 3.0 [range 0–18.0] versus 2.0 [range 0–14.0]; P = 0.05, respectively, according to the Mann-Whitney U test). Also, median GU and SH/E scores were increased in EORA compared to YORA patients, as shown in Figure 1. EORA cases were more likely to have sonographically active disease than patients with YORA (85% versus 72%; odds ratio [OR] 3.9 [95% confidence interval (95% CI) 1.3–11.5], P = 0.015) as indicated by the presence of PD signals (grade ≥1) in at least 1 joint.
No difference was found between EORA and YORA patients concerning clinical measures of disease activity as outlined in Table 1. Particularly, the DAS28, CDAI, and SDAI scores were comparable in both groups.
In correlation analyses, an association of GU, SH/E, and PD scores with DAS28 (corrcoeff 0.40, 0.29, and 0.38, respectively, P < 0.05 for all), as well as a linkage between the GU and HAQ scores (corrcoeff 0.27, P < 0.05) were found in EORA but not in YORA patients. Ultrasound scores of inflammation were not linked with the number of tender and swollen joints (28- and 22-joint count), the physician's global assessment, or the patient's global assessment in either group.
A stratified analysis according to DAS28 categories revealed higher GU scores and SH/E scores in EORA than YORA patients with high clinical disease activity (see Supplementary Figure 2, available in the online version of this article at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)2151-4658). Additionally, we found higher ultrasound scores of inflammation in EORA patients with moderate or high clinical disease activity compared to patients in remission or low disease activity.
Multivariate linear and multipredictor logistic regression analyses were performed in those 108 patients with complete data sets to prove the validity of the above reported results. For linear regression, either the square root–transformed GU score or square root–transformed SH/E score was applied as the dependent variable and EORA/YORA status as the predictor of primary interest. After backward stepwise exclusion of nonsignificant parameters (see Methods section for a full list of variables), the final models included erosive disease and DAS28 as covariates. Regression analysis confirmed our unadjusted result of higher sonographic scores in EORA than in YORA patients (regcoeff for GU score 0.70, P = 0.039; SH/E: regcoeff 0.67, P = 0.026) and yielded an association between GU score and DAS28 (regcoeff 0.39, P = 0.002) that was particularly evident in the EORA group as indicated by the introduction of the interaction term EORA*DAS28 (regcoeff 0.65, P = 0.002).
According to logistic regression analysis with the presence of PD signals (grade ≥1) in at least one joint as the dependent variable, and with osteoarthritis and DAS28 as covariates (included into the final model after backward exclusion of nonsignificant parameters), EORA patients were at an OR of 3.0 (95% CI 1.1–8.2, P = 0.029) to have hypervascularized lesions compared to YORA patients. Supplementary Table 1 (available in the online version of this article at http://onlinelibrary.wiley.com/journal/10.1002/(ISSN)2151-4658) details the results of our sensitivity analyses that did not significantly alter the primary results.
This is the first report on rheumatologic ultrasound in EORA and YORA patients, demonstrating that the burden of joint inflammation is higher in EORA than in YORA patients despite similar clinical disease activity. This finding was robust to adjustments by several covariates including disease duration, disease activity, radiographic damage, and current therapy in multivariate regression analyses, and was not influenced by the exclusion of patients under current treatment with anti-TNFα agents.
Disease activity of EORA patients is often underestimated in clinical practice, although these patients commonly present with severe illness, including arthritic and constitutional symptoms leading to transient or permanent immobility (8). Our data support this observation, suggesting that EORA patients have even higher inflammatory scores at sonographic than at clinical examination. Also, many physicians fear possible side effects and toxicity of disease-modifying antirheumatic drugs in the elderly, and therefore EORA patients frequently receive less aggressive therapies than YORA patients aimed at symptom control rather than inflammatory control (3). This observation was reflected also in our own EORA and YORA patients, when focusing on the rate of biologic therapies (Table 1).
In contrast to previous studies suggesting a similar grade of disease activity in EORA and YORA cohorts according to clinical composite scores (3, 9), we found more inflammation in the former group, compared to the latter group, using ultrasound. A relative disparity between clinical and ultrasound findings was already reported in previous studies in patients with RA, indicating that clinical composite scores lack sensitivity and specificity to detect joint inflammation (4). Also, these indices are influenced by osteoarthritis and other noninflammatory joint pain that are common in the elderly (10). The prevalence of radiographic signs of osteoarthritis was (as expected) higher in the EORA group compared to the YORA group; hence, ultrasound may be a useful tool to objectify joint inflammation in these patients.
The impact of ultrasound results on clinical and structural outcomes of EORA patients is elusive so far. Data from earlier EORA cohorts indicate more functional limitations and greater radiographic damage in elderly-onset compared to younger-onset patients, despite similar levels of clinical disease activity and comparable disease duration (9). However, as ultrasonography was not performed in these studies, and followup radiography was not available in our cohort, the possible linkage between ultrasound-verified inflammation and structural lesions in EORA patients remains unclear. The slightly higher prevalence of erosive disease in the YORA group compared to the EORA group cannot be explained by ultrasound results; rather, it is linked to longer disease duration in the YORA patients than in the EORA patients. Disease duration, on the other hand, did not influence ultrasound results as indicated by regression analysis.
The pattern of clinical and sonographic joint involvement was comparable in our EORA and YORA patients. Most patients showed active ultrasound lesions at the wrists, whereas clinical symptoms were most frequently noted at PIP joints. In previous RA cohorts, a similar distribution of clinical and sonographic findings was reported (11). It can be assumed that the sonographic patterns of EORA and YORA patients are different, if large joints, particularly shoulders, are examined (8). Besides, involvement of metatarsophalangeal joints is often asymptomatic in RA and no data exist on the prevalence of ultrasound findings at these sites in elderly RA patients. These issues need to be clarified by future studies.
The most important limitations of our study are the retrospective design, the missing clinical records in some patients, the differences between EORA patients and YORA patients concerning disease duration and anti-TNFα treatment, and the absence of data on inter- and intraobserver agreement. We know from other studies that reliability of ultrasound in RA is excellent, whereas reproducibility of clinical joint examination appears to be poor (4). Radiographs were not routinely rated according to established radiographic scores, and no systematic reassessment of available films was performed. We were therefore unable to correlate sonographic data with radiographic scores in EORA and YORA patients, and focused rather on the relation between clinical and sonographic findings, adjusting for the presence of radiographic damage in regression analyses.
In conclusion, this is the first report of ultrasound objectifying joint inflammation in EORA patients. Ultrasound-detected levels of inflammation are higher in EORA patients compared to YORA patients, despite similar clinical disease activity.
All authors were involved in drafting the article or revising it critically for important intellectual content, and all authors approved the final version to be submitted for publication. Dr. Schirmer had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study conception and design. Dejaco, Duftner, Wipfler-Freissmuth, Weiss, Schirmer.
Acquisition of data. Dejaco, Duftner, Wipfler-Freissmuth, Weiss, Schneider.
Analysis and interpretation of data. Dejaco, Duftner, Schirmer.