FormalPara Key Summary Points

Elderly-onset rheumatoid arthritis (EORA) is linked to increased mortality risk, but the impact of various disease-modifying anti-rheumatic drugs on this risk remains unknown. The study investigates risk factors for all-cause mortality in EORA patients.

Data from 980 EORA patients diagnosed between January 2007 and June 2021 were analyzed, and significant mortality-associated risk factors were identified, including higher age, male sex, current smoker status and underlying malignancy.

Hydroxychloroquine treatment was found to be protective against mortality in patients with EORA, with a hazard ratio (HR) of 0.30 and a 95% confidence interval (CI) of 0.14–0.64 (p = 0.002).

Hydroxychloroquine treatment is associated with survival benefits in EORA in a dose-dependent manner, particularly in those receiving a monthly cumulative dose of 1374.5–5778.5 mg or more.

Introduction

Rheumatoid arthritis (RA) is a chronic autoimmune inflammatory disease that induces irreversible joint disabilities and increased comorbidities compared with the general population [1, 2]. Higher age, comorbidities and functional disability are independent risk factors when predicting 5-year mortality in RA patients [3]. In RA patients, comorbidities such as depression, lung disease (obstructive and restrictive), cardiovascular events and solid-organ malignancies were associated with increased mortality risk [1]. However, the clinical characteristics and the immune system of patients with elderly-onset RA (EORA) react very differently from those of patients with young-onset RA (YORA) and exhibit an increase in the age-related multi-morbidities that are triggered by the damage-associated molecular pattern [4,5,6]. Compared to patients with YORA, patients with EORA exhibited higher disease activity, which involved large joints and presented more radiographically detected damage [7]. Moreover, EORA patients have higher levels of systemic inflammatory biomarkers, less improvement with targeted therapy and received less intensive therapy with biologics and glucocorticoids than in YORA [8]. Furthermore, the persistent inflammation likely increased the mortality in RA patients compared with the general population [9, 10]. These aging-related risks might be related to immunosenescence and inflammageing that, in turn, confer great challenges when optimizing therapy for patients with EORA [6].

At least two-fold mortality rates in RA were determined in association with clinical severity before the biologic era [11]. With the development of biologic disease-modifying anti-rheumatic drugs (bDMARDs) and targeted synthetic DMARDs (tsDMARDs) in recent decades, the mortality risk of RA patients has decreased drastically [9, 10]. Treatment with hydroxychlroroquine has decreased the risk of cardiovascular events, diabetes mellitus and hyperlipidemia in RA patients [12,13,14].

After a 3-year treat-to-target strategy, EORA patients with chronic lung disease, previous malignancy and poor control of disease activity at baseline had serious adverse events compared to patients treated with bDMARDs, methotrexate or glucocorticoids [15]. However, higher age (≥ 65 years) was a strong risk factor for increase in all infections or serious infection events in RA patients treated with tofacitinib compared to treatment with TNF inhibitors [16]. Furthermore, knowledge of the risk factors and medication use in association with the EORA-associated mortality risk is limited. Attaining a balance between safety and effectiveness when treating EORA patients remains a challenging mission for the rheumatology community. In addition, with the aging of the rheumatic autoimmune population globally, it is imperative to identify contributory factors associated with mortality risks among EORA patients.

In this study, we aimed to identify predictors associated with EORA mortality and survival in the clinical setting.

Methods

Study Population

We conducted a study to determine the mortality risk and causes of death in patients with EORA, which is defined as RA onset after 60 years of age [17], from the data obtained from the electronic health records of Taichung Veterans General Hospital (TCVGH), Taiwan. The diagnosis of established RA was recorded in conformity with the ACR 1987 revised criteria and/or 2010 ACR/EULAR criteria [18, 19]. Patients with established RA and those who had received csDMARDs, bDMARDs and tsDMARDs from January 2007 to June 2021 were eligible for this study. The reimbursement of bDMARDs or tsDMARDs by Taiwan’s National Health Insurance (NHI) is approved in RA patients who inadequately respond to methotrexate-based triple therapy of csDMARDs and have a 28-Joint Disease Activity Score (DAS28) > 5.1 [20]. This study complies with the Declaration of Helsinki of 1964 and its later amendments, and the TCVGH’s Ethics Committee approved the study (CE18190A) and waived the requirement for informed consent as  patient data was  anonymized prior to analysis. The authors had permission to access and use the data.

Study Outcome

The outcome of this study was the all-cause mortality reported from electronic health records of TCVGH, Taiwan, and was defined by the death date recorded in the Taiwanese Ministry of Health and Welfare (TMHW). The occurrence of mortality and causes of death were automatically updated and maintained in the TMHW. The linkage between the electronic health records of TCVGH, Taiwan, and the TMWH was necessary. Adverse events and causes of death were ascertained and reported by the treating physicians.

Data Collection

We extracted information on covariates from the electronic health records of TCVGH, Taiwan, prior to the index date of June 30, 2021. The baseline demographics, lifestyle, RA disease activity determined by the DAS28-ESR, past and current medication use, history of concurrent comorbidities, change of disease status and therapeutic response (rheumatoid factor [RF] and/or anti-citrullinated protein antibody [ACPA], erythrocyte sedimentation rate [ESR], C-reactive protein [CRP] and DAS28) in patients with EORA were evaluated before and at quarterly intervals during treatment with csDMARDs, bDMARDs and tsDMARD.

The history and concurrent comorbidities were extracted from electronic health records of the TCVGH, Taiwan, and included ischemic cardiovascular disease (myocardial infarction, ischemic stroke, congestive heart failure, aortic aneurysm), diabetes mellitus, hyperlipidemia, malignancies (colon, skin, lung, breast and uterus for women, prostate for men) and lymphoproliferative disease, gastrointestinal diseases (diverticulitis, ulcers), infections (pneumonia, urinary tract infection, hepatitis), lung disease (asthma, chronic obstructive pulmonary disease [COPD] and interstitial lung disease [ILD]), osteoporosis and psychiatric disorders (depression). In addition, a history of joint surgery as recorded. Hepatitis B carrier, hepatitis C carrier and latent TB statuses were identified in accordance with the recommendations of Taiwan’s Centers for Disease Control (CDC) and Taiwan Rheumatology Association (TRA) [21, 22]. Patients with EORA were classified as having the comorbidities if the diagnosis was made between RA diagnosis and the last follow-up or death of the patient.

Demographics and the EORA Disease

The patient’s baseline demographics included age, sex, disease duration (or the date of RA diagnosis), body weight, body height, smoking status and alcohol consumption. Disease activity was evaluated by the Disease Activity Score using the 28-Joint Erythrocyte Sedimentation Rate (DAS28-ESR) and the 28-Joint C-reactive protein (DAS28-CRP) [20]. The first data of DAS28-ESR, RF level and/or ACPA level at baseline were recorded. The RF IgM levels were determined by nephelometry (Dade Behring Inc., Newark, DE, USA; positive if ≥ 14 IU/ml). The ACPA levels were measured using EliA CCP (Phadia, Nieuwegein, The Netherlands; positive if ≥ 10 U/ml).

RA Treatment

Past and current medications used for EORA were recorded and included glucocorticoids, non-steroidal anti-inflammatory drugs (NSAIDs), csDMARDs (hydroxychloroquine, methotrexate, sulfasalazine, cyclosporine, leflunomide), bDMARDs (adalimumab, certolizumab, etanercept, golimumab, abatacept, tocilizumab, rituximab) and tsDMARDs (baricitinib, tofacitinib). Patients with EORA were classified into csDMARD, bDMARD and tsDMARD categories based on the medication prescribed between RA diagnosis and the last follow-up or death of the patient.

Hydroxychloroquine Dose and Duration

The available hydroxychloroquine dose and duration of treatment based on prescription between RA diagnosis and the last follow-up or death of the patient were recorded. All participants were divided into quartiles according to their monthly cumulative dose of hydroxychloroquine: < 1374.5 mg: quartile 1; 1374.5–5778.5 mg: quartile 2 and 3; ≥ 5778.5 mg: quartile 4.

Statistical Analysis

The demographic data of the continuous parameters are expressed as the mean ± standard deviation (SD) and for the categorical variables as the number of patients. The chi-square test and Mann-Whitney U test were conducted to compare variables among patients with EORA in the survivor and non-survivor groups. Risk factors associated with EORA mortality were determined by the Cox proportional hazard regression. The EORA survival curves were calculated by the Kaplan-Meier method, and statistical significance among groups was analyzed by the log-rank test. The interaction between users/non-users of hydroxychloroquine and the presence of an underlying malignancy in the mortality risk of EORA was evaluated in this study.

All data analyses were performed using the Statistical Package for the Social Sciences (SPSS), version 23.0. Significance was set at p < 0.05.

Results

Baseline Demographic Data

A total of 980 EORA patients were enrolled (survivors 852 and non-survivors 128; Table 1), and data were extracted from electronic health records of the TCVGH, Taiwan. The study’s median follow-up duration was 13.8 (9.1–16.6) years. Of note, the demographics of EORA patients with male sex, higher age, current smoking, higher disease activity and concurrent treatment with cyclosporine, leflunomide and biologics were associated with a higher mortality rate. Those with pre-existing comorbidities in the EORA patient non-survivor group had more chronic kidney disease, hypertension, stroke and underlying malignancy than in the survivor group. In contrast, a higher rate of EORA with alcohol consumption was noted in the survivor group.

Table 1 Demographics of patients with EORA
Full size table

Predictors Associated with All-Cause Mortality

To identify independent factors associated with mortality among EORA patients, Cox regression analysis was performed (Table 2). We found that higher age (HR 1.10, 95% CI 1.07–1.12, p < 0.001), male sex (HR 1.92, 95% CI 1.22–3.00, p = 0.004), current smoking (HR 2.31, 95% CI 1.10–4.87, p = 0.027) and underlying malignancy (HR 1.89, 95% CI 1.20–2.97, p = 0.006) were significant risk factors for mortality. In contrast, 816 and 120 EORA patients in the survivor and non-survivor group received hydroxychloroquine, respectively. Hydroxychloroquine treatment appeared to be a protective factor against mortality in patients with EORA (HR 0.30, 95% CI 0.14–0.64, p = 0.002).

Table 2 Cox proportional hazard regression analysis of risk factors associated with mortality in patients with EORA
Full size table

Survival Analysis by the Use and Cumulative Dose of Hydroxychloroquine

The overall 10-year Kaplan-Meier survival probability of the hydroxychloroquine and non-hydroxychloroquine group (Fig. 1A) showed that hydroxychloroquine conferred a protective effect against mortality (p = 0.006). EORA patients with a monthly cumulative hydroxychloroquine dose < 1374.5 mg exhibited the lowest survival rate compared to patients with a dose of 1374.5–5778.5 and ≥ 5778.5 mg (Fig. 1B).

Fig. 1
figure 1

Kaplan-Meier analysis of A 10-year overall survival probability of hydroxychloroquine users and non-users, p = 0.006; B 5-year overall survival probability stratified by the monthly cumulative dose of hydroxychloroquine (≥ 5778.5, 5778.5–1374.5 and < 1374.5 mg) in patients with elderly-onset rheumatoid arthritis (≥ 5778.5 mg vs. 1374.5–5778.5 mg, p = 0.217; ≥ 5778.5 mg vs. < 1374.5 mg, p < 0.001; 1374.5–5778.5 mg vs. < 1374.5 mg, p < 0.001). Pairwise comparison by the log-rank test

Full size image

Interaction Between Mortality Risks of Malignancy and Hydroxychloroquine Treatment

Underlying malignancy conferred an increased mortality risk whereas hydroxychloroquine conferred a protective effect from death for EORA patients. The interaction between user/non-user of hydroxychloroquine and the presence of underlying malignancy for mortality risk of EORA was analyzed. For EORA patients, the presence of malignancy without hydroxychloroquine treatment exhibited the highest mortality risk compared with their counterparts (Fig. 2).

Fig. 2
figure 2

Stratification of the mortality risk by the presence of malignancy and hydroxychloroquine use in patients with elderly-onset rheumatoid arthritis. Error bars represent the 95% confidence intervals (CI) of the hazard ratios (HR)

Full size image

Discussion

Our study demonstrated that higher age, male sex, current smoking and underlying malignancy were significant risk factors of mortality. In contrast, hydroxychloroquine treatment was a protective factor against mortality in patients with EORA in a dose-dependent manner. Mortality risks for EORA patients with malignancy were further reduced with the concomitant hydroxychloroquine treatment. Our results shed light on a potential survival benefit of hydroxychloroquine treatment in older patients with RA.

Patients with RA are at higher risk for serious infection, osteoporosis, cardiovascular disease (CVD), respiratory disease and malignancy [6, 23,24,25]. Patients with EORA tend to have more geriatric syndromes, comorbidities, ILD, serious infection, cardiovascular events and higher malignancy rates compared to YORA [6, 26]. A 5-year longitudinal observational study demonstrated that risk factors of increased EORA mortality included male sex (OR 4.35), higher age (OR 1.17) and RF positivity (OR 11.93) [23]. Increased cardiovascular risk was reported among EORA patients in use of glucocorticoids (OR 1.5; 95% CI 1.1–2.1) or non-methotrexate immunosuppressants (azathioprine, cyclosporine or leflunomide; OR 1.8; 95% CI 1.1–3.0) [24]. A single-center study showed that EORA patients with comorbidities of hypertension (HR 12.02, 95% CI 1.08–133.54), ILD (HR 85.04, 95% CI 4.11–1759.19) and compressive fracture (HR 65.07, 95% CI 2.15–1971.10) are associated with increased mortality risk compared to YORA [25]. Our study demonstrated that an increased risk of EORA mortality was significantly associated with higher age, male sex, current smoking and underlying malignancy. Moreover, EORA with hydroxychloroquine treatment showed protection against mortality risk in a dose-dependent manner.

Hydroxychloroquine is widely used for treating autoimmune inflammatory rheumatic diseases because of its efficacy and safety. Hydroxychloroquine regulates innate and adaptive immunity by inhibition of multiple endo-lysosamal activities including autophagy, Toll-like receptor (TLR) activation and the calcium signaling pathway, to induce a synergic effect to decrease pro-inflammatory cytokine secretion [27].

RA patients with hydroxychloroquine treatment showed advantages including improved metabolic profile and decreased incident cardiovascular events [12, 14]. Moreover, RA patients under hydroxychloroquine treatment were not associated with development of heart failure [28]. Interestingly, the long-term use of hydroxychloroquine showed a reduction in diabetes risk up to 77% for RA patients [13]. Furthermore, an observational cohort study from Taiwan’s NHI Research Database demonstrated that hydroxychloroquine treatment in the newly diagnosed RA patients benefit a 36% reduction of incident chronic kidney disease compared with non-users [29]. Additionally, hydroxychloroquine therapy showed protective benefit for RA patients from non-alcoholic fatty liver disease [30]. At the same time, hydroxychloroquine treatment among RA patients revealed cardiovascular safety with long-term use [31, 32]. However, we had limited information for the impaction of hydroxychloroquine therapy to all-cause mortality in RA patients, especially in EORA. In our cohort, our result demonstrated that hydroxychloroquine treatment was a protective factor against mortality of EORA patients in a dose-dependent manner. In our speculations, hydroxychloroquine is associated with lower risk of all-cause mortality in EORA through decreased risks of cardiovascular events, hyperlipidemia, diabetes mellitus, chronic kidney disease and non-alcoholic fatty liver disease. Further research is required to confirm our findings.

Our data showed that underlying malignancy was a significant risk factor for mortality in EORA patients. Lung cancer is the highest incident malignancy, followed by gastric cancer, breast cancer and lymphoma, as reported by the Ninja cohort [33]. The incidence of malignancy from the CRANE cohort was higher in EORA than in YORA patients [34].

The standardized rates of overall malignancy excluding non-melanoma skin cancer varied from 0.56 to 0.87 per 100 person-years from five global registries. The IORRA cohort from Japan demonstrated that malignancy is the main cause of mortality among RA patients [35]. Moreover, the meta-analysis demonstrated an increased risk of lung cancer and lymphoma in RA patients [36]. A single medical center in Korea reported increased overall malignant incidence in RA with gender difference but without relevance to csDMARDs [37]. A real-world multi-database study showed no significant risk of specific cancer for RA patients with variant bDMARDs or tsDMARDs [38].

Recently, high-dose hydroxychloroquine was used more frequently in clinical studies for anti-cancer therapy and showed potential efficacy by inhibition of autophagic flux for tumor growth and modulation of tumor immune response [27]. Furthermore, overexpression of TLR9 of breast and prostate cancer is associated with cancer invasion and stimulation of TLR9/nuclear factor kappa B (NFκB) signaling pathway results in gastric cancer migration and proliferation [39, 40]. Hydroxychloroquine powerfully inhibits TLR9 through impaired endosomal acidification [41]. In general, hydroxychloroquine showed a potential mechanism in inhibition of tumor growth and modulation of anti-cancer immunity by enhancement of anti-tumor T-cell response, restoration tumor antigen sensing of TLR4-deficient dendritic cells, inhibition of chemotaxis and adhesion of tumor cells, triggering cell death by stimulation of mitochondrial apoptosis [27]. Although our data indicated that hydroxychloroquine treatment might mitigate cancer-related mortality risks, further studies are needed to prove the survival benefit of hydroxychloroquine treatment in EORA patients with malignancy.

This study has several limitations. First, our study design is retrospective observational. One of the primary constraints of our study is the incapacity to preclude the “sick stopper effect.” This phenomenon arises when EORA patients with higher mortality risk do not have the chance to receive HCQ treatment compared with low-risk patients. Consequently, these individuals are susceptible to biases and may have influenced our model’s estimations. Moreover, patient’s adherence to HCQ and HCQ blood levels were not available in this study. However, we provided novel findings of potential survival benefit of hydroxychloroquine for EORA patients. Second, the number of EORA patients in this study is limited, and the sample could be underpowered to detect significant predictors of mortality in this group of patients. Third, our study population included only Taiwanese individuals, and the result may not be extrapolatable to the other ethnic groups. Future study of mortality risk and hydroxychloroquine therapy in patients with EORA is needed to confirm our findings.

Conclusions

Hydroxychloroquine treatment might confer survival benefits through a dose-response relationship for patients with EORA, especially in patients with malignancies. Further studies are needed to ascertain the underlying mechanism of this association.