Hydroxychloroquine in Systemic Lupus Erythematosus

B OBJECTIVES B1 To evaluate in SLE patients who lower but do not discontinue HCQ factors associated with flare (significant increase in disease activity, augmentation of therapy or hospitalization for SLE).

B2 To determine in SLE patients who discontinue HCQ factors associated with flare (significant increase in disease activity, therapy augmentation or hospitalization for SLE).

B3 To evaluate temporal relationship between HCQ use and retinal toxicity in SLE patients and how this may be associated with demographic and clinical factors.

B4 To assess SLE patient preferences for HCQ therapy and how preferences are associated with demographic and clinical factors.

B5 To evaluate temporal relationship between antimalarial use and other forms of toxicity in SLE patients and how this may be associated with demographic and clinical factors.

Team members have prospectively followed cohorts > 3700 adult SLE patients (90% HCQ exposed) with collected and harmonized data on demographics and clinical factors. Team members (Bernatsky Abrahamowicz Fortin Hazlewood) are founding members of CAN-AIM CIHR-funded Drug Safety and Effectiveness Network team.

Members have history of working with patients as research partners (Bartlett Li Bernatsky Fortin Hazlewood). Investigators' ties to stakeholders including patient research partners and Lupus Canada facilitates knowledge translation.

METHODS C1 Cohorts: Investigators will use existing data on adult SLE patients enrolled into study cohorts in Montreal Quebec Halifax Winnipeg Calgary and Toronto.

Investigators are members of national (Canadian Network for Improved Outcomes in SLE CaNIOS) and/or international (Systemic Lupus International Collaborating Clinics SLICC) research networks. CaNIOS and SLICC have worked decades on harmonized approach to prospectively-collected. Cohorts enroll unselected patients when they present to clinic. At annual assessments data are collected on drugs disease activity organ damage etc C2.1 Patient selection for Objective 1: At baseline and follow up visits records are made of patient HCQ doses. Starting with first visit with HCQ exposure investigators will determine how many patients are recorded as receiving lower dose or discontinuing HCQ at follow-up visit. Patients identified as having lowered HCQ dose will form sub-cohort for time-to-event (survival) analyses. Investigators will evaluate only first-ever lowering of HCQ. Analysis will evaluate patients from time zero (annual visit at which lower HCQ dose was recorded) to assess combined outcome of earliest of a) increase in SLEDAI-2K of at least 4 points and/or b) hospitalization for SLE and/or c) augmented SLE therapy defined as increase in HCQ or new start or increase in corticosteroids or other immunosuppressant. In addition to accepted minimal clinically significant SLEDAI-2K change investigators include drug changes and SLE-related hospitalizations.

Patients studied in Objective 1 will be right censored if HCQ is discontinued completely; these will enter into analyses for Objective 2. Otherwise patients will be followed until outcome of interest or end of study.

C2.3 Subject selection for Objective 2 Analyses will be similarly conducted but time zero will begin when individual discontinues HCQ. Objective 1 and 2 analyses conducted separately; patients who have HCQ dose lowered but then discontinued will be in analyses for Objective 1 up to point of discontinuation then patients will be right censored for Objective 1. Subjects are able to contribute data (from the point of HCQ discontinuation) to analyses for Objective 2. Patient can contribute to analyses for one or both objectives and contributing to Objective 3. If not right-censored patients will be followed until outcome of interest or end of study.

C2.3 Subject selection for Objective 3 Outcome is retinal toxicity assessed in HCQ-exposed patients from time of first visit with HCQ exposure up to time of retinal toxicity documentation or end of study for patients without outcome. Exposure to CQ (relatively uncommon drug exposure) is also risk factor for retinal toxicity but essentially patients exposed to CQ will have prior exposure to HCQ.

C3 Demographic and clinical factors C3.1 Variables assessed include sex age at SLE onset and race/ethnicity. Analyses adjusted for education as education can alter adherence.

C3.2 Clinical factors for multivariable analyses include smoking BMI baseline disease activity malignancy data and renal damage. Baseline refers to time zero. Renal damage is captured with SLICC Damage Index measuring accumulated damage. Baseline disease activity measured by internationally used SLEDAI-2K reliable validated widely used global score index.Investigators will model time-varying drug exposures with step-up approach initially with baseline use ie current, recent, ever or total past dose. Investigators use sophisticated and novel weighted cumulative dose modelling to flexibly represent exposures accumulated by individual subjects up to given time-point corresponding to each risk set for time-to-event analyses. In multivariable analyses approach will allow investigators to explore how risk of retinal toxicity may depend on details of past HCQ exposure including not only total cumulative dose but temporal pattern of doses at various time in past with differential weights for early versus later exposures.

C4 Outcome assessment: SLEDAI-2K and drugs are recorded. SLICC Damage Index has item for retinal damage and updated yearly.

C5 Model development: Investigators will perform multivariable time-to-event analyses in patients who lower or discontinue HCQ. Data for outcomes are available longitudinally at annual visit with actual dates only available for addition/increase in corticosteroid or SLE hospitalization. For other outcomes investigators will deal with interval-censored data familiar issue in pharmacoepidemiology.

Models include baseline demographics education BMI SLE duration SLEDAI-2K renal damage and medications. Simplistic survival analyses often assume linear effects of continuous covariates and impose conventional proportional hazards (PH) assumptions; investigators know that prognostic factors may have time-dependent and/or nonlinear effects that violate assumptions. Investigators will account for this. Methods account for possible changes in hazard ratio over time (violation of the PH) and allow estimation of relative risks (hazard ratios) and actual probabilities of event of interest occurring.

Similar approach will be used for Objective 3 in time-to-event analyses of retinal toxicity in patients exposed to HCQ.

Competing risks occur when subjects can experience one or more events that may compete with outcome of interest and hinder observation of event of interest or modify chances for event to occur. In sensitivity analyses investigators use competing risks models to simultaneously estimate associations between HCQ exposures and hazards of competing event.

D POWER/SAMPLE CALCULATIONS D1 Power calculations for multivariate models Combined cohorts include over 3300 patients have exposed to HCQ. Patients have had lowering of dose 45% or discontinuation of HCQ during their follow-up. Power calculations are based on available data suggesting that 45%-60% of SLE patients may flare over ten years of follow-up.

For baseline SLEDAI-2K assuming 35% of patients score less than 3 investigators would have 90% power to detect unadjusted HR of 1.30 and adjusted HR of 1.37 for variable. Assuming 20% of patients have renal damage investigators will have 90% power to detect an unadjusted HR of 1.37 and an adjusted HR of 1.45 for variable.

E KNOWLEDGE TRANSLATION activities are informed by history of partnerships between investigators and knowledge-users directed by Dr. Linda Li Canada Research Chair in Patient-Oriented KT and Director of Knowledge Translation Research Program at UBC's Arthritis Research Centre of Canada.

E1 Integrated KT Proposal developed in consultation patients from Lupus Canada patients Canadian Alliance of Patients with Arthritis (CAPA) and Singer Family Fund for Lupus Research. Wendy Singer (providing expertise health information dissemination and provides insight of SLE patient who has been affected by HCQ-induced retinal toxicity) will assist with educational tool developments. Co-PI Dr. Bartlett has over decade of work with patient partners through OMERACT (Outcome Measures in Rheumatology) and other organizations to help disseminate findings patient communities.

E2 Dissemination via MyLupusGuide: Data on HCQ from current study will be incorporated into MyLupusGuide web-based patient-centred health-management tool developed by Dr. Fortin in partnership with Jack Digital Productions CaNIOS Lupus Canada. Development validation/dissemination of current version funded by CIHR operating and dissemination grants.

E3 Clinicians and professional groups: Dr. Bernatsky is member of CRA Therapeutics Committee which advises CRA on policies related to drug use for rheumatic diseases. Group monitors information about adverse events related to drugs used in rheumatology including HCQ. Investigators submit abstracts to scientific meetings (American College of Rheumatology, European League Against Rheumatism, CRA meeting). Investigators will seek feedback on feasibility/approaches to integrate MyLupusGuide as gold standard for patient education tool concerning HCQ.

E4 Informing future guidelines: Team member Dr. S. Keeling has already produced Canadian SLE care guidelines (funded by the CIHR and endorsed by the CRA and CaNIOS) . In 2019, again mandated by these groups, Dr. Keeling will spearhead guidelines for SLE therapies.

E5 Health agency stakeholders: Investigators have connections Drug Safety and Effectiveness Network (DSEN) Health Canada's Marketed Health Product Directorate and Canadian Agency for Drugs and Technologies in Health (CADTH). Drs. Abrahamowicz and Bernatsky are co-leads DSEN-funded team.