DISEASE ACTIVITY PREDICTS WHOLE BODY AND REGIONAL LEAN TISSUE IN RHEUMATOID ARTHRITIS – A CROSS-SECTIONAL STUDY

Aim. The study aims to assess the potential infl uences of rheumatoid arthritis (RA) and its specifi c disease measures on lean body composition phenotypes of female patients. Methods. The study was cross-sectionally designed to include Caucasian postmenopausal female RA patients and age-matched postmenopausal female controls. All the subjects gave written informed consent and the study was approved by the local ethics committee. Each subject underwent in the same day a clinical examination, laboratory tests, whole body dual X-ray absorptiometry (DXA) composition and physical activity estimation using a self-administered questionnaire. Correlations, differences and predictive power were analyzed with appropriate statistical tests. Results. The study included 107 RA patients and 104 controls. Compared to the normal subjects, who recorded higher levels of physical activity, the RA patients had signifi cantly lower appendicular lean tissue absolute and relative indices and higher prevalence of sarcopenia. The whole body and appendicular lean tissue indices showed signifi cant negative correlations with measures of disease severity (duration, infl ammation, quality of life and radiographic progression), independent of age, levels of physical activity, body mass index and smoking. Conclusions. The measures of disease activity and severity independently predict lean tissue phenotypes in RA patients, behaving as risk factors for sarcopenia and rheumatoid cachexia. The diagnosis of RA in itself is a signifi cant predictive factor of sarcopenia.


INTRODUCTION
Rheumatoid arthritis (RA) is a progressive autoimmune disease characterized by chronic infl ammation which can lead to permanent joint deformity, disability, distress and socio-economic costs. This disease is also associated with profound modifi cations of the body composition since the RA-lean tissue interaction leads to sarcopenia and rheumatoid cachexia (1). The main pathogenic factors involved in the RA-associated muscle loss seem to be inadequate physical activity (2) and chronic infl ammation-induced catabolism via cytokines such as tumor necrosis factor α (3). The study of these conditions is very relevant since sarcopenia and rheumatoid ca-chexia are associated with low quality of life, chronic fatigue, muscle weakness and adverse outcomes (4). Classical anthropometric measures, such as body mass index (BMI), do not give information on body composition and are not able to discriminate between the proportions of tissue types, which can vary widely within the same BMI ranges. The assessment of body composition in RA patients was mostly done using bioimpedance methods (5;6), which are relatively inexpensive and fast. The use of the more appropriate dual X-ray absorptiometry (DXA) technique for body composition in RA is limited (7)(8)(9)(10), especially regarding the whole body and regional muscle mass which are highly correlated with more sensitive estimations methods such as magnetic resonance imaging (11). Most of the studies included both male and female RA patients, although on one hand, RA is more prevalent among women and on the other hand that body composition differs signifi cantly among genders. In this context, the present study aims to evaluate the whole body and regional lean mass in Romanian female RA patients using the DXA technique and to assess the potential infl uences of RA and its specifi c disease measures on lean body composition phenotypes of female patients.

Patients and criteria
The study was cross-sectionally designed to include all the patients admitted to our Rheumatology Department (Research Centre of the Pathology and Treatment of Systemic Rheumatic Diseases -RCRD, Bucharest) in the random order of presentation between May and August 2013, who fulfi lled the following criteria: female sex (since RA is more frequent in women and since men have largely different body composition phenotypes than women); Caucasian race; postmenopausal status; 2010 ACR/EU-LAR classifi cation criteria for RA (12) and more than 6 months disease duration. Age-matched healthy Caucasian post-menopausal female subjects were randomly selected and invited to participate in the study using the records of general practitioners associated with RCRD from the same geographic area. The following exclusion criteria were applied: age under 18 years; pregnancy; weight > 150 kg (DXA table weight limit); muscular and neuromuscular disease (myositis, muscular dystrophy, myasthenia gravis); current cancer; HIV-AIDS; digestive pathology (gastrectomy, bariatric or intestinal surgery, primary biliary cirrhosis, celiac disease, malabsorption); chronic obstructive lung disease; severe heart failure (New York Heart Association classifi cation ≥ 3); endocrine abnormalities (hyperthyroidism, hyperparathyroidism, hypogonadism); moderate-severe chronic kidney disease (glomerular fi ltration rate < 60 mL/min/1.73 m 2 ); psychiatric eating disorder; therapy with any of the following drugs in the last 6 months: estrogen-replacement therapy, glucocorticoids exceeding 7.5 mg/day oral prednisone equivalent, antipsychotics, anticonvulsants, heparin, orlistat, lorcaserin, exogenous insulin. Each subject gave written informed consent and the study was approved by the local ethics committee. The clinical examination, laboratory tests and DXA body composition analysis were done in the same day for each patient.

Clinical examination
A clinical interview and a review of the medical history recorded age, smoking status, disease duration, duration of morning stiffness, patient global self assessment of general health and disease activity (visual analog scale), treatment regimes and extraarticular manifestations (rheumatoid nodules, vasculitis, neurologic, pulmonary, cardiac, renal or ophthalmologic involvement). The patients were given a self-administered modifi ed health assessment questionnaire (MHAQ). (13) A single senior author rheumatologist (DO) performed systematical clinical examinations, which focused on anthropometrics and RA variables. Height and weight were measured in upright anatomical position, light clothes, without shoes, using a mechanical scale (0.1 kg maximal error) and a stadiometer (0.3 cm maximal error). Obesity and overweight were defi ned using the World Health Organization (WHO) cutoffs of BMI ≥ 30 kg/ m 2 and 25 kg/m 2 respectively. The clinical examination recorded the number of painful and swollen joints (bilateral proximal interphalangeal, metacarpophalangeal, wrist, elbow, shoulder, knee joints) and identifi ed the patients with RA joint deformities of their hands (henceforth designated clinical structural damage): fi xed fl exion contracture, ulnar deviation, "swan neck", "boutonniere", "Z" thumb, other RA luxations and subluxations, impaired range of motion, arthritis mutilans. A visual analog scale indication of the global disease activity was recorded by the evaluator. A single senior author rheumatologists (VB) identifi ed radiological damage on standard postero-anterior X-ray images of the hands. The Simple Erosion Narrowing Score (SENS) was calculated using the method proposed by van der Heijde et al. (14).

Laboratory measures
All the included patients underwent morning venipuncture and blood samples were tested with commercial kits for routine blood chemistry and complete blood count, erythrocyte sedimentation rate (ESR; Westergren method), C-reactive protein (CRP; nephelometric method), IgM rheumatoid factor (RF) and anti-citrullinated peptide antibodies (ACPA; enzyme-linked immunosorbent assay). In-fl ammation was classifi ed if either CRP or ESR were above the upper limit of normal (5 mg/L and 30 mm/h respectively), in the absence of other causes than RA. Using the clinical and laboratory measures, the RA activity was assessed using two composite tools: disease activity score (DAS28; remission ≤ 2.6; low disease activity -LDA 2.6-3.2; moderate disease activity -MDA 3.2-5.1; high disease activity -HDA > 5.1) (15) and clinical disease activity index (CDAI) (16).

DXA whole body composition
Body composition was evaluated by whole body less head DXA with a Lexxos C05LX223 densitometer. All the scans were performed by a single certifi ed clinical densitometrist (CP; 0.48% variation coeffi cient). Daily calibration and quality control tests were performed according to the manufacturer's recommendations and different regions of interest were manually checked for maximal reliability (17). The patients were required to wear light clothing, without metal or plastic, and were scanned in the morning, after nocturnal fast, micturition and 5-10 minutes of supine rest on the examination table, in the absence of pregnancy and radioactive or radiocontrast investigations in the last week. Data records included whole body and regional (arms, legs) variables such as lean tissue density/mass/area/percent (wbLT D/M/A/P) and whole body adipose tissue mass (wbATM). Appendicular lean mass (ALM) was calculated as the sum of the lean masses of all four members, while the appendicular lean density (ALD) was calculated as the arithmetic mean of the lean tissue densities of the four members. The skeletal muscle mass index (SMI) was calculated in three ways: ALM divided by body mass (SMI1), by square height (SMI2) and by wbLTM (SMI3). The free-fat mass index (FFMI) was calculated in three ways: wbLTM divided by body mass (FFMI1), by square height (FFMI2) and by wbATM (FFMI3). The fat mass index (FMI) was calculated by dividing wbATM to square height. Since the defi nitions of sarcopenia (18,19) and cachexia (20) include muscle strength, a variable which was not included in the study design, the terms "sarcopenia" and "cachexia" will be used as classifi cation labels for the whole body phenotypes based on low muscle mass and high adipose mass. Defi ning sarcopenia as SMI2 below 2 standard deviations (SD) of the Rosseta study young female population mean (7.3 ± 0.9 kg/m 2 ), or as FFMI2 below the predicted value for Caucasian females (21), resulted in 100% prevalence of sarcopenia in our sample (18,22). Another strategy was to defi ne sarcopenia as SMI2 < 20 th percentile of the study group (23,24), but unlike the cited authors we did not include male subjects in the study. In the absence of specifi c Romanian population cutoffs for whole body composition, we used our non-RA group as reference population. Accordingly, sarcopenia was defi ned by two principles: as SMI2 or FFMI2 < 20 th percentile of the non-RA group (corresponding to 3.453 kg/m 2 and 6.874 kg/m 2 respectively); and as T-scores of SMI2 or FFMI2 below 2 SD of the reference group's mean. The T-scores were calculated by subtracting each of the SMI2 or FFMI2 values from the mean of the reference group (4.002 kg/m 2 and 8.075 kg/m 2 respectively) and dividing by the standard deviation of the reference group (0.608 kg/m 2 and 1.357 kg/m 2 respectively). Rheumatoid cachexia was defi ned as FFMI2 below the 10 th percentile and as FMI above the 25 th percentile of our reference population (25)(26)(27).

Physical activity estimation
Physical activity was estimated using the selfadministered Global Physical Activity Questionnaire version 2 (GPAQ2), developed by WHO, with a total of 16 questions in 4 activity sections (work, travel, recreation, sedentary behavior) (28,29). The English questionnaire was translated into Romanian by the authors and back-translated by an independent authorized translator. The fi nal version of the Romanian version was pre-tested on a random sample of 15 healthy employees and 15 RA patients from RCRD.

Statistics
Data distribution normality was assessed using descriptive statistics, normality, stem-and-leaf plots and the Lillefors corrected Kolmogorov-Smirnov test. Qualitative data were expressed as "absolute value (percentage of group)" and were studied using cross-tabs with χ 2 or Fisher tests. Non-normally distributed scale data were reported as "median (interval)" or "mean (interquartile range)" and their correlations and differences were assessed using non-parametric tests: bivariate Spearman and partial correlations of 2 scale variables; Mann-Whitney U and Kruskal-Wallis for differences of scale variables in groups with 2 (e.g. smoking) or more categories (e.g. RA disease activity). To determine which categories of multi-level nominal variables produced signifi cant χ 2 or Mann-Whitney tests, one-way ANOVA was used with post-hoc analysis (Tukey and Bonferroni multiple comparisons). To assess the independent predictive capacity of RA-variables, logistic regression models were created using lean tissue phenotypes (binary logistic regression) or lean tissue scale variables (standard multivariate linear regression) as dependents and RA variables and confounding variables as covariates; the scale variables included in the models were normalized using arithmetic functions (indicated in text). All tests were considered signifi cant if p < 0.05 and were done us-ing Statistical Package for the Social Sciences v.20 (SPSS Inc., Chicago, USA, 2008) for Windows.

Comparison of groups
The study included 107 RA postmenopausal female patients and 104 postmenopausal non-RA female subjects ( Table 1). The 49 (45.8%) patients on low-dose glucocorticoids were taking ≤ 7.5 mg/day prednisone for a median period of 9 months (3-15 months). Compared to the normal subjects, who recorded higher levels of physical activity, the RA patients had signifi cantly lower appendicular lean tissue  absolute and relative indices and higher prevalence of sarcopenia (22.4-38.3% according to the defi nition method).

RA-specifi c variables
Compared to the non-sarcopenic RA patients (Table 2), the sarcopenic RA patients, regardless of the defi nition method, had longer disease duration, higher prevalence of glucocorticoid treatment and rheumatoid cachexia and a more severe and active disease (as measured clinically and radiographically). Table 2 reports only the differences of FFMI-defi ned sarcopenia by the 20 th percentile method, but the other three defi nitions of sarcopenia behaved the same (data not shown). The sarcopenic patients also displayed lower levels of physical activity and higher levels of sedentary behavior (Table 3). Similarly, compared to the RA patients without cachexia, the cachectic RA patients had longer disease duration and signifi cantly higher radiographic progression (as measured by SENS). Of note, the RA patients with infl ammation, clinical structural damage, glucocorticoids and MDA had lower whole body and appendicular lean tissue ( Table 4).
The whole body and appendicular lean tissue indices showed signifi cant negative correlations with  the measures of disease severity (duration, infl ammation, quality of life and radiographic progression), independent of age, the levels of physical activity and with the BMI and smoking status (Table 5). In fact, these disease measures were capable to independently predict lean tissue phenotypes in RA patients, behaving as risk factors for sarcopenia and cachexia (Tables 5 and 6). The diagnosis of RA is a signifi cant predictive factor of sarcopenia defi ned by SMI, but not by FFMI.

Unique study fi ndings
Our results suggest that RA becomes a signifi cant predictor of the low muscle mass through three inter-twined mechanisms: anatomic joint damage (clinical structural damage and SENS), disease activity (DAS28 and CDAI activity classes) and infl ammation (CRP, ESR). The RA patients with anatomical joint damage, high disease activity and chronic infl ammation exhibit lower whole body and appendicular lean tissue indices and higher prevalence of sarcopenia and rheumatoid cachexia. These three factors might have a direct effect on muscle mass  (generalized muscle catabolism in chronic infl ammation and periarticular appendicular muscle atrophy) and an indirect infl uence on body composition (lower levels of physical activity, secondary osteoarthritis). These fi ndings have implications in fundamental research (e.g. the relationship of muscle mass and insulin resistance and metabolic syndrome in RA) and translational clinical potential (e.g. the relationship between the risk of falling and fragility fractures and the whole body or appendicular lean tissue; the effect of targeted treatment on body composition). A prospective study which would observe the evolution of DXA-measured lean tissue under treatment is recommended in order to confi rm our results
and to document the dynamic extent of the body composition alteration in RA.

Comparing with other studies
There are few data regarding DXA-estimated whole body and regional lean tissue in women with RA. Generally, recent studies reported no signifi cant difference of absolute whole body lean tissue between RA patients and controls (7,9,30), although there are reports which fi nd the opposite (31,32). In the later studies, the RA patients had lower whole body lean tissue than controls, but the difference was not signifi cant in the study setting, as we also found. Signifi cant whole body lean tissue differences appear when other variables are used. Giles et al. (30) and Dao et al. (7) reported that whole body lean mass divided by square height does not differ sig-nifi cantly (FFMI2 in our study), but that the ratio of whole body lean and adipose content is signifi cantly lower in RA patients (FFMI3 in our study, which we found to be non-signifi cant, probably because our normal subjects had a signifi cantly higher BMI and equivalent whole fat mass compared to the RA patients). The loss of muscle mass seems to be more pronounced in the members, as studies report in accordance with our data. Female RA patients have signifi cantly lower appendicular lean mass than controls (9,31), although there are reports in which this difference failed to reach statistical signifi cance (30). Whether regional and selective or whether whole body and nonselective, the lean mass loss is more prevalent among the female RA population as indicated by the higher frequency of sarcopenia compared to controls (7,30). Our data showed that a SMI2-derived defi nition of low muscle mass is more appropriate for classifi cation purposes and that the female RA patients have a 3-fold higher risk of low muscle mass than controls, as Giles et al. also found (30). Our data showed that RA-specifi c variables correlate and are signifi cant predictors of lean mass (Tables 5 and 6). Disease duration was a strong predictor of low muscle mass, both whole body and appendicular, in female RA patients, a fi nding which confi rms some literature results (7,9,26,33), and contradicts others (30,31). Since disease duration is associated with radiologic progression, disability and co-morbidity, one would have expected to observe a negative infl uence on lean mass in RA, the differences in reports being probably caused by the extent of disease duration and sampling methods. Similar observations were made with HAQ (7,26,30), infl ammation markers (8,26), clinical joint deformity (30), which had a signifi cant negative effect on lean mass, while auto-antibody seropositivity and DMARD treatment produced no signifi cant differences (9).
The points where our results differ from the literature are disease activity and glucocorticoid treatment. As with other reports (9,30), the absolute values of composite disease activity scores, DAS28 namely, did not correlate nor did it predict lean mass indices, but at a nominal level analysis of our data notable results began to appear (for example patients with CDAI-remission had a lower prevalence of sarcopenia and a higher lean mass body percent than CDAI-MDA patients - Tables 2 and 4). Our glucocorticoid treated RA patients had signifi cantly lower whole body and appendicular lean tissue indices, sarcopenia was more frequent among these patients and this status was a signifi cant predictor of low muscle mass and lean tissue indices, while the majority of reports deny any cross-sectional effect of glucocorticoid treatment on lean body composition (7)(8)(9)31). Clinical and fundamental research has shown that exogenous glucocorticoids are associated with skeletal muscle catabolism, atrophy and wasting (34)(35)(36), therefore one would expect that long-term glucocorticoids contribute to the loss of muscle mass in the already at-risk RA patients. An explanation of this report discrepancy would be on one hand the disease duration (Book et al. studied early RA patients with disease duration below 12 months) (9), and on the other hand a less clear history of the exposure to glucocorticoid treatment (exusers, current-users or both, and never-users).

Study limitations
The cross-sectional design of the study did not allow follow-up of patients and dynamic observations of the whole body lean tissue. Data regarding diet were not included in the study design. For a thorough comparison between RA patients and controls regarding lean tissue indices, a BMI-match may have increased the objectivity of the observation. The cumulative dose of glucocorticoids and their treatment periods could not be objectively recorded.

CONCLUSION
In comparison to normal subjects, women with RA have signifi cantly lower appendicular lean tissue DXA indices, higher prevalence of sarcopenia and lower levels of physical activity. Disease duration, infl ammation, disease activity, quality of life, clinical and radiographic progression and glucocorticoid treatment are negatively correlated with lean body composition, predict whole body and appendicular lean tissue and are independent risk factors for sarcopenia and rheumatoid cachexia. The diagnosis of RA in itself is a signifi cant predictive factor of sarcopenia.