Article Text
PDF
Abstract
Background Characterisation of the long-term outcome of patients with ‘difficult to treat’ (D2T) rheumatoid arthritis and factors contributing to its evolution are unknown. Herein, we explored the heterogeneity and contributing factors of D2T long-term outcome.
Methods Patients included from a prospective single centre cohort study. The EULAR definition of D2T was applied. Longitudinal clustering of functional status (modified Health Assessment Questionnaire (mHAQ)) and disease activity (Disease Activity Score-28 (DAS28)) were assessed using latent-class trajectory analysis. Multiple linear mixed models were used to examine the impact of comorbidities and their clusters on the long-term outcome.
Results 251 out of 1264 patients (19.9%) were identified as D2T. Younger age, fibromyalgia, osteoarthritis, DAS28-erythrocyte sedimentation rate (ESR) at first biological or targeted synthetic disease-modifying antirheumatic drug (b/ts-DMARD) initiation and failure to reduce DAS28-ESR scores within the first 6 months of b/ts-DMARD therapy were significant predictors of patients becoming D2T. Long-term follow-up (total of 5872 person-years) revealed four groups of functional status evolution: 18.2% had stable, mildly compromised mHAQ (mean 0.41), 39.9% had gradual improvement (1.21–0.87) and two groups had either slow deterioration or stable significant functional impairment (HAQ>1). Similarly, four distinct groups of disease activity evolution were identified. Among the different clusters of comorbidities assessed, presence of ‘mental-health and pain-related illnesses’ or ‘metabolic diseases’ had significant contribution to mHAQ worsening (p<0.0001 for both) and DAS28 evolution (p<0.0001 and p=0.018, respectively).
Conclusion D2T patients represent a heterogeneous group in terms of long-term disease course. Mental-health/pain-related illnesses as well as metabolic diseases contribute to long-term adverse outcomes and should be targeted in order to optimise the prognosis of this subset of rheumatoid arthritis.
- Rheumatoid Arthritis
- Outcome Assessment, Health Care
- Patient Reported Outcome Measures
Data availability statement
Data are available upon reasonable request. The data and analytical methods that support the findings of this study are available to qualified investigators upon request to the corresponding author.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
WHAT IS ALREADY KNOWN ON THIS TOPIC
Several disease-related characteristics and comorbid diseases differentiate difficult to treat (D2T) patients from the rest of the rheumatoid arthritis population, while cross-sectional analysis revealed differences within the D2T group.
WHAT THIS STUDY ADDS
Our analysis of a prospectively followed cohort indicated that D2T represents a heterogeneous group in terms of long-term functional and disease activity evolution. Presence of mental-health/pain-related illnesses as well as metabolic diseases significantly contribute to adverse outcomes.
HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY
Together with a better control of inflammatory burden, a special focus in the above comorbid diseases could possibly improve the outcome of these patients with major unmet-needs.
Introduction
The clinical application of novel targeted therapies has significantly changed the outcome of patients with rheumatoid arthritis (RA). Nevertheless, data from well-organised registries and cohort studies have shown that significant hurdles of achieving remission—the treatment target—still exist.1 Α group of patients with RA to whom targeted therapies of different modes of action fail to adequately control disease activity have been characterised as ‘difficult to treat’ RA (D2T).2–7
Studies assessing factors that predict the evolution to D2T have shown that this represents a rather heterogeneous group concerning the characteristics of the patients at RA diagnosis or when starting a targeted therapy.4–6 8 9 Most of the studies characterised this population cross-sectionally, while long-term prospective data are missing.4–6 8–11 Additionally, although ‘by definition’ D2T patients are expected to have adverse outcome this has not been studied. In the present study we sought to address whether D2T patients constitute a homogeneous group of patients regarding the long-term outcome of their functional status and disease activity evolution. Finally, we sought to investigate the potential impact of factors contributing to less favourable outcomes of D2T patients.
Methods
Setting and patient inclusion
For the present study we analysed patients registered in the University of Crete Rheumatology Clinic Registry (UCRCR), a single centre prospective cohort established in year 2004. Adults ≥18 years old with inflammatory arthritis are included in UCRCR at the time of initiation of the first biological or targeted synthetic disease-modifying antirheumatic drug (b/ts-DMARD) and they are prospectively followed irrespectively of treatment switches for as long as they receive biological/targeted therapy. Only patients fulfilling the 2010 American College of Rheumatology (ACR)/EULAR criteria for an RA diagnosis were included in the analysis.
Study design
As this is an observational cohort study, all treatment decisions (b/ts-DMARD initiation and selection of type/mode of action, co-medication, dosage adjustments/switches) are made by the attending rheumatologists based on clinical assessments, national guidelines and patients’ preferences. According to national guidelines, patients are considered candidates for biological treatment if they have active disease (defined as Disease Activity Index 28 (DAS28)>3.2) and have failed previous treatment with at least one conventional synthetic DMARD (cs-DMARDs), most commonly methotrexate. Although, based on national guidelines for the management of RA, treatment adaptations according to disease activity levels are recommended, a strict treat-to-target approach is not mandatory in clinical practice. Treatment decisions are mostly based on physician’s assessment for disease activity, taking into consideration disease activity, comorbidities and patient’s preferences.
Data sources
According to the UCRCR protocol, data on demographics, disease characteristics, comorbidities, disease activity, function and quality of life are collected at the first b/ts-DMARD initiation and every 3–6 months for the first 2 years and yearly thereafter.12 Treatment discontinuations and any adverse events are registered prospectively by the attending physician. In cases of loss of follow-up of a patient for more than one and a half years, the patient is reported as ‘lost to follow-up’ at the date of the last recorded follow-up visit. In the present study, we included all patients fulfilling the 2010 ACR/EULAR criteria for RA, who started any b/ts-DMARD between 10 January 2004 and 14 December 2021, had available data for DAS28 joints using the erythrocyte sedimentation rate (ESR) at baseline and had a total follow-up time of at least 1 year. Data were extracted for analysis on 24 December 2022. Patients were followed until discontinuation of all b/ts-DMARD therapies, death, loss of follow-up, or date of data extraction. At inclusion in the registry all patients provided a written informed consent according to the Declaration of Helsinki.
Definitions
Patients were classified as D2T according to the EULAR definition7 (online supplemental figure 1—flowchart of the study). In the D2T cohort we included only patients who have either failed two or more b/ts-DMARD classes, or were currently receiving a second b/ts-DMARD class therapy, but DAS28-ESR during the last year of follow-up was ≥3.2 in all measurements. In the latter case, where patients had not stopped the second b/ts-DMARD class, despite the moderate or high disease activity, the criterion regarding the problematic management perceived by the clinician or the patient was judged by the score on the global Visual Analogue Scale (VAS) of the patient and/or VAS of the physician during the last follow-up year (either or both of the values should be ≥50/100).
Supplemental material
Based on the comorbidities and treatments documented by treating rheumatologists, the Rheumatic Disease Comorbidity Index (RDCI) was calculated.13 14 In case of missing information, we also had access to the national prescription platform, in which all our patients are recorded. Obesity diagnosis was based on body mass index with a cut-off >30 kg/m2.
In order to assess whether patterns of rather common comorbid diseases may affect the outcome of D2T RA, we decided to adopt data from large scale epidemiological studies, instead of developing ‘custom-made’ multimorbidity grouping. For that reason, we adopted the proposed multimorbid clusters originally developed by England et al.15 We used the exact same terminology and definitions as the authors proposed and which they studied in two large epidemiological studies. Briefly, in order to define multimorbidity they focused on 44 chronic conditions, selected based on their prevalence and importance in the general and RA population, informed by prior studies, including systematic reviews of multimorbidity.16–19 To minimise misclassification of these conditions they required at least two diagnoses for these chronic conditions to be considered present. Applying machine learning methodology, they developed patterns of multimorbidities both in RA and in non-RA population. The validity of the proposed analysis was shown by the significant consistency of multimorbidity patterns across analyses and across the data sets they applied. The diseases per cluster are presented in (online supplemental table 1). According to methodology suggested, for a patient to be classified in each multimorbidity pattern he/she should have been diagnosed with ≥2 chronic illnesses that constitute each cluster.
Finally, as regards the adverse events (AEs), two main categories were created; one containing the adverse events that were classified as moderate, serious, life-threatening or lethal (namely at least moderate AEs) and one containing serious, life-threatening and lethal adverse events (namely at least serious AEs). The incidence was calculated using the number of adverse events divided by the follow-up period. The first calculated incidence regarded the total follow-up period since b/ts-DMARD initiation and the other the follow-up period after the patient was classified as D2T.
Outcomes
Validated patients’ reported outcomes and composite disease activity indices were used for the assessment of disease activity and function: DAS28-ESR and modified Health Assessment Questionnaire (mHAQ) for RA.
Statistical analysis
Comparisons of selected demographics and disease related characteristics at b/ts-DMARD initiation between D2T and non-D2T patients were performed using univariate analysis (Pearson’s χ2 test/Fisher’s exact test in cases of categorical data, independent samples t-test in cases of normally distributed data or Mann-Whitney test in cases of non-normally distributed data). A multiple logistic regression model was used to assess the odds of a patient becoming D2T. Independent variables used were gender (male/female), age at b/ts-DMARD initiation (years), seropositivity (yes/no), presence of fibromyalgia (yes/no), presence of osteoarthritis (yes/no), DAS28-ESR at b/ts-DMARD initiation, DAS28-ESR difference between at b/ts-DMARD initiation and 6 months of treatment >1.2 units (yes/no), obesity (yes/no), depression (yes/no), RDCI index >3 (yes/no) and the time period that the patient started the first b/ts-DMARD and was enrolled in the study (2004–2008, 2009–2014 and 2015–2021). Clustering of longitudinal data of mHAQ and DAS28-ESR scores of D2T patients (n=251) were analysed using latent class trajectory analysis. The number of latent classes as well as the type of function (linear, polynomial, etc) were assessed based on the values of Akaike information criterion and the Bayesian information criterion (BIC). To this end, the trajectory plot of mHAQ scores over time were fitted using a linear model of four latent classes (namely HAQ traj-groups). Trajectory plots of DAS28 scores were fitted using second order polynomial function of four latent classes (namely DAS traj-groups). Differences in patient characteristics between HAQ trajectory groups and DAS28 trajectory groups were assessed using univariate analysis (Pearson’s χ2 test in cases of categorical data, one-way analysis of variance in cases of numerical data). In order to assess the impact of selected clusters of comorbid diseases on the long-term outcomes of HAQ and DAS28 scores of D2T patients (n=251), linear mixed models were performed. These multimorbidity clusters were identified in a large observational study and all of them are over-represented within patients with RA (see above in ‘Definitions’). In order to minimise classification bias, cluster membership (yes/no) was assessed if patients were diagnosed with ≥2 chronic illnesses that comprised each cluster.15 The dependent variables used in the linear mixed models were mHAQ/DAS28-ESR scores over time. Independent variables were cluster membership (yes/no), time (months) and the interaction term of cluster membership x time. The level of statistical significance was set to 95%. An intention-to-treat analysis was carried throughout. All analyses were performed using the statistical software IBM SPSS V.25, and Stata V.17. Latent class trajectory analyses were performed using the traj module in Stata.20
Results
Patient characteristics at baseline and predictors for D2T
We analysed 1264 patients with RA (81.2% women), prospectively followed-up for a total of 6045 person-years. The flowchart of the study can be found in online supplemental figure 1. Based on the EULAR definition for D2T, 251 (19.9%) were characterised as D2T at a median time of 37 months (min, max; IQR (6, 215; 43 months) after starting the first b/ts-DMARD. Selected demographics and RA-related parameters of D2T and non-D2T at baseline of b/ts-DMARD initiation are presented in table 1. D2T patients were significantly different compared with non-D2T patients in terms of gender, age at first b/ts-DMARD initiation and RF positivity; they had a shorted disease duration before receiving the first b/ts-DMARD, had also higher DAS28-ESR and mHAQ scores and had received more non-biological drugs. Moreover, patients who evolved to D2T had a higher burden of comorbidities, since dyslipidaemia, obesity, hypothyroidism, osteoarthritis, fibromyalgia and depression were more prevalent in this group (online supplemental table 2). Finally, D2T patients had more extra-articular manifestations and particularly sicca and peripheral neuropathy (online supplemental table 3). Multivariate logistic regression indicated that younger age at first b/ts-DMARD initiation, presence of fibromyalgia, osteoarthritis, higher DAS28-ESR scores at baseline and failure to improve DAS28>1.2 at the first 6 months of first targeted therapy increased the odds of a patient for evolving to D2T. Additionally, patients enrolled after 2015 had significantly lower odds of becoming D2T compared with patients enrolled during 2004–2008 (table 2, p<0.05 for all).
Demographics and disease-related characteristics b/ts-DMARD initiation in all patients and comparison between D2T and non-D2T patients
Logistic regression predicting the odds of patient being D2T (yes/no)
Trajectories of long-term functional outcome
Aiming to assess the long-term outcome and to depict heterogeneity within D2T patients, we analysed the trajectories of functional evolution (mHAQ scores) of D2T patients (n=251) during a total of 60 months of follow-up. Latent class trajectory analysis revealed the existence of four distinct functional HAQ trajectories (namely HAQ traj-groups 1–4) (figure 1). About one-fifth (18.2%) of D2T patients (HAQ traj-group 1) had a more favourable functional evolution with mean mHAQ scores approximately of 0.41 on b/ts-DMARD initiation which remained relatively stable during the total period of follow-up. Interestingly, HAQ traj-group 3 (39.9% of the D2T group) although had a moderately compromised function at b-DMARD or ts-DMARD initiation (mean mHAQ score 1.21) showed a gradual improvement even after being characterised as D2T resulting in a milder functional disability (mean mHAQ score 0.87 at 60 months). On the contrary, two groups of adverse functional status evolution were identified. Patients in HAQ traj-group 2 (31.9%), although they started from a mildly limited functional level, showed a stable gradual deterioration of functionality, resulting in a moderate compromised status (mean mHAQ score 1.10 at 60 months). Finally, patients in HAQ traj-group 4 (~10% of D2T patients) had stable and significant functional limitations from the onset and throughout the follow-up period of b/ts-DMARD therapy with mHAQ scores >1.5.
mHAQ latent-class trajectory analysis plots. Median time-to-characterisation as difficult to treat rheumatoid arthritis of each trajectory is given on the plots. mHAQ, modified Health Assessment Questionnaire.
Since mHAQ has been associated with several ‘strong’ outcomes and in order to support the clinical validity of the above-described HAQ-trajectories, we compared the incidence of adverse events and hospitalisations between the different HAQ traj-groups. Indeed, it was shown that patients in HAQ traj-group 1 had the lowest incidence of events (both serious and moderate) as well as hospitalisations during the whole follow-up period and also specifically after being characterised as D2T (table 3).
Univariate comparisons between HAQ traj-groups
Trajectories of disease activity evolution
Although by definition D2T patients have inadequate disease activity control, it has not been reported whether ‘patterns’ of disease activity exist within this group. For that reason, we applied latent class trajectory analysis on the evolution of DAS28-ESR scores of D2T patients (n=251) over time assuming a second order polynomial fit. Comparable to HAQ trajectories, four distinct DAS28-ESR trajectories were identified during the follow-up period (figure 2). A minority of the patients (8.3%) had a stable moderate disease activity while 38% showed a gradual improvement from high to moderate disease activity. The remaining two groups (total 53.6% of the patients) had distinct but constantly high disease activity during the total follow-up period.
DAS28-ESR latent class trajectory analysis plot. DAS28-ESR, Disease Activity Index 28-erythrocyte sedimentation rate.
Disease-related characteristics and comorbidities comparison between D2T long-term subgroups
We next assessed potential differences of RA-related characteristics and comorbidities between the functionally distinct groups. Patients in mHAQ traj-group 1, had lower disease burden and functional limitation at the start of targeted therapies (table 3). Moreover, they had a more favourable profile in terms of comorbidities’ burden, examined either separately (like hypertension, diabetes, obesity, osteoarthritis and fibromyalgia) or as a group (RDCI index score). Interestingly enough, patients in mHAQ traj-group 3 (which had a more favourable functional evolution), had lower comorbidities’ burden, resembling HAQ traj-group 1, but shared similarities with HAQ-traj groups 2 and 4 (which showed the most adverse functional evolution) in terms of disease-related characteristics and demographics.
Τhe above analysis was repeated between the different long-term disease activity groups (DAS traj groups) (table 4). Results indicated that younger age, lower initial DAS28 scores, lower initial mHAQ scores and higher rates of improvement in DAS28-ESR at 6 months were associated with more favourable long-term disease activity course (DAS traj-group 1). Interestingly, although a lower burden of comorbidities assessed by RDCI was identified in DAS28 traj-group 1, only fibromyalgia was significantly less common in patients of DAS28 traj-group 1 compared with other groups (table 4).
Univariate comparisons between DAS28 traj-groups
Multimorbidity patterns’ contribution in the outcome of D2T RA
Since certain comorbidities and groups of them (referred as multimorbidity clusters) seem to have a special interest for RA, we assessed the impact of certain multimorbidity patterns in D2T patients’ long-term outcomes (n=251 D2T patients included, methodology is analysed in ‘Definitions’ section). Briefly, we applied patterns of ‘multimorbidity clusters’ identified in a large observational study in which all of them are over-represented within patients with RA online supplemental table 1.15 Our results indicated that although ‘cardiopulmonary and vascular diseases’ did not significantly affect HAQ evolution, the presence of ‘mental-health and pain-related illnesses’ as well as ‘metabolic diseases’ had a significantly contribution in HAQ deterioration (p<0.0001 for both, figure 3, plots a and d). Interestingly, the same patterns of multimorbidity adversely affected DAS28-ESR evolution; patients with ‘mental-health and pain-related illnesses’ as well as ‘metabolic diseases’ had significantly worse DAS28 scores evolution (p<0.0001 and p=0.018, respectively, figure 4, plots a and d).
Linear mixed models plots of predicted mHAQ values over time using time and disease clusters membership as predictors. (A) Mental health and pain related (B) cardiopulmonary (C) vascular (D) metabolic. mHAQ, modified Health Assessment Questionnaire.
Linear mixed models plots of predicted DAS28-ESR values over time using time and disease clusters membership as predictors. (A) Mental health and pain related (B) cardiopulmonary (C) vascular (D) metabolic. DAS28-ESR, Disease Activity Index 28-erythrocyte sedimentation rate.
Since the background of the patients starting b-DMARD or ts-DMARDs has changed over the last years and, as shown above, the period of inclusion in UCRCR was a predictor for D2T characterisation, we assessed whether the period of inclusion affected the long-term outcomes. Thus, we repeated the latent class trajectory analysis of mHAQ and DAS28 scores as well as the mixed models plots of significantly contributing comorbidity clusters, for patients included after 2009 (online supplemental figures 2 and 3). On comparison, we did not find any major differences between the total cohort analysis and the one performed in patients recruited after 2009. Interestingly, trajectory analysis for HAQ and DAS28 evolution was comparable between the total cohort and this subgroup (online supplemental figure 2) as well as mixed models for the contribution of selected comorbidities clusters in long-term outcomes (online supplemental figure 3).
Discussion
Representing a group of patients with unmet needs, D2T RA has been in the epicentre of RA clinical investigation during the last years. Most of the studies investigated for predictors of D2T at RA diagnosis or at b/ts-DMARD initiation and reported on the patients’ characteristics in cross-sectional studies.4–6 8 9 11 Data assessing RA-related long-term outcome and for factors contributing to its evolution in patients with D2T RA are not available. Herein, we aimed to assess the long-term functional outcome of D2T and to characterise the factors related with its evolution. It was shown that D2T patients are clustered in four distinct functional groups with different baseline characteristics and different outcomes concerning hospitalisations and moderate/serious events. Interestingly, certain clusters of comorbidities were identified as having a significant impact in functional and disease activity evolution of D2T RA.
Several studies have assessed for factors which may predict the evolution to D2T either early at RA diagnosis or when starting a targeted therapy.4–6 8 9 In our cohort, disease’s and comorbidities’ burden, failure to improve disease activity during the first 6 months were predictors of developing D2T RA. Interestingly, patients enrolled after 2015 had significantly lower odds of becoming D2T compared with patients enrolled during the early phase of b-DMARDs use (2004–2008), supporting data that patients’ characteristics starting targeted therapies during the last 20 years have changed.21 22 Additionally, whether treatment strategies evolution, like the implementation of the ‘treat-to-target’ (T2T) approach or the choice of sequential agents after first b-DMARD failure, may contribute to D2T are interesting questions. Nevertheless, we could not assess the importance of T2T since this information is not available in UCRCR and generally not documented in most of the registries and cohort studies analysis. Concerning the sequential treatments employed, in an unadjusted analysis for the first three lines of treatments, we found that although the rate of D2T was comparable among tumor necrosis factor inhibitors (TNFis) and non-TNFis in the first line treatment, concerning the second and third targeted therapy choice, a higher proportion of patients treated with non-TNFis were classified eventually as D2T (p=0.035 and p=0.014, respectively) (data not shown). Nevertheless, this was a preliminary not adjusted analysis of a low number of patients which should be interpreted with caution.
Patients with D2T RA have been considered as a rather homogeneous group of significant inflammatory burden and mostly adverse prognosis. Heterogeneity within the group concerning RA-related outcome has not been reported. In the present analysis we assessed the functional status evolution based on mHAQ, an index accepted as a strong outcome in RA clinical investigation. It is a patient reported outcome significantly associated with both disease activity and damage and it is a reliable prognostic factor for mortality and hospitalisations in early and established disease in earlier and in most recent studies.23–27 In this study, we report that D2T patients are a heterogeneous group in reference to functional evolution during a 60 months follow-up. Applying latent class trajectory analysis, we identified four distinct groups of mHAQ long-term evolution. Interestingly, a total of 58% of the patients had a more ‘favorable’ functional outcome. Thus, 18.2% (HAQ traj-group 1) had a relatively stable functional status (mean HAQ scores approximately 0.41), while 39.9% (HAQ traj-group 3) showed a gradually improvement even after being characterised as D2T (mean mHAQ score from 1.21 to 0.87) (figure 1). Several groups have analysed the functional evolution of patients with RA in different disease states (early, established, moderated disease activity) and have reported that patients with RA have distinct but mostly stable groups of functional status. Thus, comparable to our results, Norton et al have reported four different HAQ trajectories of patients with early RA (‘Early Rheumatoid Arthritis Study’ and the ‘Norfolk Arthritis Register’),28 while a seven HAQ trajectory model fitted better in biological-naïve patients receiving cs-DMARDs while on medium disease activity.29 In both studies trajectories of HAQ were rather constant in contrast to our data showing that almost 70% of the patients may change functional status, 32% showing a decline while 40% had an improvement. The present analysis is the first report assessing the long-term functional evolution of patients with D2T RA, categorising them in distinct functional subgroups. Moreover, it gives insights into the functional evolution of established and aggressive RA, since most of the studies report on the mean HAQ scores of the total study groups without identifying distinct subgroups.
As commented above, mHAQ has been a reliable predictor of significant outcomes in patients with RA.24 26 In order to assess the clinical validity of the functional grouping of the cohort, we compared the incidence of events and hospitalisations between the different HAQ traj-groups. Interestingly, we found that the group with the most favourable functional trajectory (traj-group 1) had the lowest incidence of events (both serious and moderate in grading) and hospitalisations during the total follow-up period and for the period after becoming D2T as well. These data further support the clinical utility of the herein described grouping of the patients and the heterogeneity within the D2T population. In order to better characterise the functional evolution of the traj-group 1, we compared patients’ functional status between HAQ traj-group 1 and the non-D2T patients having well controlled disease between 12 and 24 months of follow-up (DAS28<3.2), and it was shown that HAQ scores between theses subgroups were comparable (data not shown). Nevertheless, our data supporting a favourable long-term functional outcome should be interpreted cautiously due to the several limitations of our analysis. Among these limitations, one should consider the limited follow-up time of our cohort, the definition of function, the small number of patients analysed, the apparent lack of appropriate control and the single-centre design of our study.
Our analysis showed that the two distinct groups with the most positive functional outcome (HAQ traj-groups 1 and 3) had comparable and more favourable characteristics (lower DAS28-ESR scores, lower burden of comorbidities assessed by RDCI) at baseline of targeted therapy initiation, as compared with mHAQ traj-groups 2 and 4, which had continuous deterioration or severe and stable compromised functional status, respectively (table 3). It was known that functional status of patients with RA is significantly affected by the presence of comorbidities both in early28 30 31 and established disease.14 32 33 Herein we report for the first time the strong association of disease-related characteristics (inflammatory burden) and the presence of comorbidities with adverse functional evolution of patients with D2T RA. Interestingly, although the period of inclusion in UCRCR as commented above predicted the evolution to D2T (tables 1 and 2), it did not affect trajectory evolution (online supplemental figure 2).
Apart from separate comorbidities, the importance of the combined impact of multiple chronic conditions, or multimorbidity, on patients with RA’s well-being has been recognised.34 35 Since we found that comorbidities contribute to worse functional outcome, we sought to assess the importance of certain groups of comorbid conditions in the functional status progression of our cohort. England et al have recently reported on multimorbidity patterns in patients with RA.15 The selected diseases per cluster which was applied are presented in online supplemental table 1. Interestingly, although both traj-groups 1 and 3 had lower prevalence of ‘metabolic’ cluster of comorbidities as compared with traj-groups 2 and 4 in univariate analysis, only traj-group 1 had lower prevalence of ‘mental health and pain’ cluster of comorbidities, a finding which emphasises the importance of this ‘phenotype’ as related with a more benign group within D2T patients (table 3).
We further performed linear mixed models in order to better assess the impact of the selected multimorbidity clusters in long-term mHAQ evolution. We found that although cardiopulmonary and vascular diseases did not affect mHAQ evolution, the presence of ‘mental-health and pain-related illnesses’ as well as ‘metabolic diseases’ had a significant contribution in mHAQ deterioration (p<0.0001 for both, figure 3). The idea that different comorbidities differentially affect RA several outcomes has been described several years ago by analysing the National Databank for Rheumatic Diseases,14 although the importance of multimorbidity was not specifically assessed. Stouten et al have recently reported the effect of different types of comorbidities in RA outcome, from a post hoc evaluation of the pragmatic randomised controlled CareRA trial in early RA.31 They showed that among different comorbidities only the presence of hypertension or depression had an impact on HAQ after 2 years. We consider that our approach to segregate diseases sharing common pathophysiological mechanisms or affecting the same organ as described by England et al15 better addresses the question of the effect of multimorbidity patterns in RA outcome, especially in this cohort with adverse prognosis. Our data showing the important contribution of ‘mental-health and pain-related illnesses’ as well as ‘metabolic diseases’ in both functional deterioration and adverse disease activity evolution, underscore the clinical significance of effectively targeting these comorbidities in order to improve D2T outcome.
The major strength of the present study is the large cohort of unselected patients receiving biologics in ‘real-world’ clinical setting. Patients were prospectively monitored by the same protocol, by a group of experienced rheumatologists with track-record in clinical research. Nevertheless, due to the observational design of the study, some covariates were missing and some patients had incomplete follow-up records or were lost to follow-up. We tried to reconcile this by removing patients in whom missing data precluded their categorisation as D2T or not. In order to assess which factors had a significant impact in the long-term outcomes we first used latent class trajectory analysis, a method which is able to account for missing values in repeated measures, followed by univariate analysis of selected factors between various trajectory groups. In order to select the best-fitting model and obtain the optimum number of groups (latent classes), we used the BIC and the ‘elbow criterion’. Linear, second and third degree polynomial fits were tested. Nevertheless, since there is no clear-cut criterion for choosing the best model, we triangulated our findings using linear mixed models with dependent variables both DAS28 scores and HAQ scores over time.36 Our results indicated that disease clusters which were identified as significantly different between trajectory-groups, were also significant predictors of DAS28 or HAQ scores.
In conclusion, our analysis revealed a heterogeneity in functional evolution and disease activity progression within a D2T RA cohort. A group of patients with lower burden of ‘mental-health and pain-related illnesses’ as well as ‘metabolic diseases’ showed the most favourable outcome, underscoring the importance of targeting these common comorbid diseases together with RA inflammatory burden in order to improve the outcome of these patients.
Data availability statement
Data are available upon reasonable request. The data and analytical methods that support the findings of this study are available to qualified investigators upon request to the corresponding author.
Ethics statements
Patient consent for publication
Ethics approval
Ethical approval was obtained from the Institutional Review Board of the University Hospital of Heraklion, Crete (decision-number: 1476/20-03-2012). All participants provided written informed consents.
Acknowledgments
The authors wish to acknowledge all consultant and trainee rheumatologists, specialist nurses and contributing patients of the Department of Rheumatology, Clinical Immunology and Allergy of the University Hospital of Heraklion, Crete, Greece, for the acquisition, maintenance and provision of clinical data.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
Twitter @george_bertsias
Contributors IDF, PS, GB: conceptualisation, methodology, formal analysis, manuscript drafting, approval of final version. AB: formal analysis, software, manuscript drafting. AR: investigation, resources, manuscript drafting. NA, SP, EK: investigation, resources, manuscript proofreading and approval. PS: guarantor.
Funding This study has been funded by ‘Pancretan Health Organization’. The funder had no role in study design, collection, analysis and interpretation of the data, in the writing of the manuscript and in the decision to submit the manuscript for publication.
Competing interests None declared.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.