As the longevity of our SLE populations increases, new challenges arise for which data are sparse to non-existent. This study leverages comprehensive information abstracted from three major SLE centers led by experts in the field who maintain robust longitudinal datasets. We provide the first evidence-based guidance with regard to the consequences of HCQ withdrawal in the elderly. In our retrospective study of patients age 55 or older with quiescent SLE, the withdrawal of HCQ did not significantly increase the rate of any lupus flares when compared to a matched group of patients continuing HCQ. While 60% of the flares in the HCQ withdrawal group were mild, all flares in the comparator group were moderate. No flares occurred in patients who discontinued HCQ due to retinal or life-threatening cardiac toxicity, providing reassurance to those patients who develop antimalarial toxicity.
In considering the balance between HCQ toxicity and its benefits, addressing flares in the elderly is warranted but data to date are very limited. One study comparing 190 patients followed for 6 years during pre-menopause (mean SLE onset 27 years) to 76 patients followed for 6 years during post-menopause (mean SLE onset 52 years), revealed a decrease in activity in both groups over time, unrelated to activity at onset, menopausal status or age at diagnosis [38]. The absence of severe flares in our study groups, both of which contained patients with long-standing SLE (mean duration of SLE 24.3 years and 17.8 years in the HCQ withdrawal and HCQ continuation groups, respectively), support these findings.
Disease quiescence, a common situation in older lupus patients, confers more confidence in considering withdrawal of HCQ. However, results of the paradigm-changing Canadian Hydroxychloroquine Withdrawal Study provide evidence for concern [1]. Two main distinctions between our aging patients at risk for maculopathy and those reported in the Canadian study limit its applicability and merit further investigation. The average duration of HCQ use was only 3.3 ± 1.6 years in the HCQ continuation group compared to 2.8 ± 1.7 years in the placebo/withdrawal group, and the average age of the SLE patients was 45 ± 13.9 years in HCQ continuation group compared to 44 ± 13.7 years in the placebo group. Thus, patients in the Canadian withdrawal study were younger and on HCQ for less time than the duration associated with the risk of ophthalmologic toxicity. Therefore, the conclusions of this seminal study may not be applicable to long-standing elderly lupus patients at increased risk for maculopathy.
Compounding the ocular risks of accumulated HCQ use, age-related changes within the retina may potentiate susceptibility to toxic damage in elderly patients. Assessment of toxicity is challenging in the elderly since the diffuse loss of fundus pigmentation with age makes bull’s-eye depigmentation harder to recognize [39]. The data suggest that elderly SLE patients indeed are those that ophthalmologists consider having “high-risk eyes” defined as HCQ use > 5 years, > 1000 g total HCQ consumption, and/or > 5 mg/kg/day HCQ daily dosing. Although no publications have identified a clear age cutoff, it has been recently suggested that patients older than 60 years old are at increased ocular risk [40]. HCQ is cleared by the kidneys and liver, and dysfunction in either organ can decrease the rate of drug removal resulting in higher blood levels [41, 42].
HCQ has been associated with multiple beneficial metabolic outcomes, including lipid-lowering effects [14, 15]. In our study, only one patient in the HCQ withdrawal group required initiation of a statin, but the specific effect of HCQ discontinuation on lipid levels could not be assessed as they were not systematically checked in these patients. Although slightly more patients in the HCQ withdrawal group were on a statin at the time of HCQ discontinuation, there was an overall low use of statins in patients in both groups, supporting previous findings that suggest low use of statins in lupus patients [43]. Statins are pleotropic agents, known to exert anti-inflammatory actions beyond their lipid-lowering effects. However, their role in SLE is controversial. Previous randomized studies have failed to demonstrate significant changes in subclinical measures of atherosclerosis of SLE disease activity [44, 45]. However, a potential benefit was observed in pubertal patients with high C-reactive protein on post-hoc analysis [46]. A longitudinal study of 21 patients (mean age 44.6) from the LUMINA cohort placed on statins within 5 years of SLE diagnosis demonstrated a significant decrease in lupus disease activity in patients who were started on statin therapy [47]. The immunomodulatory effects of statins in lupus may be due to decreased expression of cell adhesion molecules, pro-inflammatory cytokines, and inhibition of type I interferon production [48, 49]. The role of statins in older lupus patients and their benefit in those unable to use HCQ due to adverse effects or toxicity should be investigated.
Previous studies have suggested a reduction in the incidence of T2DM in patients with rheumatoid arthritis and SLE on HCQ [50, 51]. A population-based study in Taiwan with an average follow-up of 5.6 years showed a lower risk of T2DM in patients who had taken HCQ (at a cumulative dose of ≥ 129 g) compared to patients who had never been on HCQ (Hazard Ratio [HR] 0.26, 95% CI: 0.18, 0.37) [52]. Although previous studies have shown short-term effects of HCQ on insulin sensitivity and pancreatic beta cell function [53], it is unknown if the antidiabetogenic benefit remains after discontinuation of HCQ if the cumulative dose and/or duration of therapy has been substantial. In our study, only one patient developed T2DM during the study period (HCQ continuation group). However, a longer follow-up and a larger sample size may be required to detect significant differences in the incidence of T2DM between the HCQ continuation and withdrawal groups.
A protective effect of HCQ in thromboembolic events in patients with SLE was first described over three decades ago [9]; however, the doses of HCQ used by these patients were not reported and the patients described in this study were younger (mean age of 41.8 in all patients, compared to 45.5 for patients with thromboemboli) and with possibly a more active disease than patients in our study (proportion of patients on prednisone > 60%) [9]. A more recent cohort study of 232 patient with SLE also supports a protective effect of antimalarials against thrombosis (HR 0.28, 95% CI 0.08–0.90) [13], but these patients were young (mean age of 36.2 years) and the median time of HCQ use was < 5 years. Hence, the cumulative effect of HCQ and the potential for persistent antithrombotic benefits after discontinuation following long-term use remain to be elucidated. We did not identify any arterial thrombotic events, and only one venous thrombosis episode (postsurgical) occurred in the HCQ withdrawal group during the 12 months following drug discontinuation, which limits our ability to draw conclusions in regard to thrombotic outcomes in the setting of HCQ discontinuation.
Although several studies have addressed HCQ use with improved survival, most of them did not explicitly address the length of HCQ use [13]. Findings of a Latin American inception cohort study suggested a time-dependent survival benefit of HCQ, with decreased mortality in those patients who maintained HCQ use for 2 or more years [19]. In our study, all patients were on HCQ for at least 5 years before discontinuation and there were no deaths during the 12-month follow-up. However, accurate comparison of mortality rates between HCQ withdrawal or continuation groups will require longer observation times and a larger sample size.
Our study findings should be interpreted considering its limitations. First, this was a retrospective study, and most of the data were collected by chart review. The low number of cases of incident T2DM, deaths, and cardiovascular and thrombotic events, and relatively short follow-up period, decreased the statistical power to establish differences between the groups. Similarly, subgroup analyses were not feasible due to small sample size. There are also potential biases in the selection of matching cases. Patients discontinuing HCQ by their own preference may be relatively healthier than those continuing HCQ. However, all patients included in our study had low disease activity. On the other hand, the development of maculopathy or cardiotoxicity may be indicators of previous compliance to HCQ, whereas patients who, despite reported long-term treatment with this drug, have not developed toxicity may have not been as adherent to HCQ. Since HCQ blood levels prior to discontinuation were not available for either group, we are unable to confirm this hypothesis. Finally, it is important to note that our study findings can only be applied to older (postmenopausal) lupus patients with quiescent disease, after long-standing use of HCQ. Therefore, the data should not be extrapolated to younger patients who have to discontinue HCQ because of toxicity or to older patient who are active when HCQ toxicity is identified.
Among the strengths of our study are that patients were identified from well-established multi-ethnic cohorts of lupus patients with close follow-up and strict documentation of classification criteria, disease activity, and medication use. Flares were also blindly adjudicated to limit bias. Additionally, we provide data regarding flares in elderly lupus patients, a demographic with limited data.