RATIONAL AND BACKGROUND: Systemic Lupus Erythematosus (SLE) is a chronic, multisystemic, autoimmune, systemic rheumatic disease of clinical and biologic heterogeneity. Given the individual variability in SLE manifestations, there is no single treatment paradigm. A tailored, multidisciplinary strategy is required which needs to be adjusted to patients’ individual clinical manifestations. Anifrolumab, a human monoclonal antibody that binds to subunit 1 of the type 1 interferon receptor (IFNAR1), was developed based on the evidence supporting the role of type 1 interferon pathway in SLE. Anifrolumab was approved via a centralised procedure in the European Union (EU) on 14 February 2022. It is indicated as an add-on therapy for the treatment of adult patients with moderate to severe, active autoantibody SLE, despite standard therapy (EMEA/H/C/004975/0000). As part of the original marketing authorisation application to the European Medicines Agency, AstraZeneca included a proposal to conduct a non-interventional multi-country post-authorisation safety study to characterise the risk of malignancies and serious infections among real-world users of anifrolumab.
RESEARCH QUESTION AND OBJECTIVES: The main research question is to evaluate the risk of malignancies and serious infections among moderate/severe SLE patients who receive anifrolumab compared with a comparable population of moderate/severe SLE patients on standard of care (SOC) who do not initiate anifrolumab.
– Primary objectives
The following objectives pertain to the malignancy outcomes:
1) To estimate the incidence of new malignancies (as a composite outcome) in moderate/severe SLE patients initiating anifrolumab and in comparable moderate/severe SLE patients who do not initiate anifrolumab (exposed to SLE SOC).
2) To compare hazard rates of new malignancies (as a composite outcome) in moderate/severe SLE patients initiating anifrolumab versus comparable moderate/severe SLE patients who do not initiate anifrolumab (exposed to SLE SOC).
The following objectives pertain to the serious infection outcomes:
3) To estimate the incidence of the first occurrence of a serious infection (as a composite outcome) in moderate/severe SLE patients initiating anifrolumab and in comparable moderate/severe SLE patients who do not initiate anifrolumab (exposed to SLE SOC).
4) To compare hazard rates of the first occurrence of a serious infection (as a composite outcome) in moderate/severe SLE patients initiating anifrolumab versus comparable moderate/severe SLE patients who do not initiate anifrolumab (exposed to SLE SOC).
– Secondary objectives
The following objective pertain to the malignancy and serious infection outcome cohorts:
5) To describe the demographic and clinical characteristics of patients in each study cohort (malignancy cohort and serious infection cohort) at index date, by exposure status (exposed to anifrolumab vs. exposed to SLE SOC).
The following objectives pertain to the malignancy outcomes:
6) To estimate the incidence of new pre-specified malignancy sub-types (separately) in moderate/severe SLE patients initiating anifrolumab and in comparable moderate/severe SLE patients who do not initiate anifrolumab (exposed to SLE SOC).
7) To compare hazard rates of new pre-specified malignancy sub-types (separately) in moderate/severe SLE patients initiating anifrolumab versus comparable moderate/severe SLE patients who do not initiate anifrolumab (exposed to SLE SOC).
The following objectives pertain to the serious infection outcomes:
8) To estimate the incidence of serious infection components – infections leading to hospitalisation, infections requiring treatment with intravenous antimicrobials, or infections related to death – in moderate/severe SLE patients initiating anifrolumab and in comparable moderate/severe SLE patients who do not initiate anifrolumab (exposed to SLE SOC).
9) To estimate the incidence of the first occurrence of opportunistic serious infections, other serious infections, pneumonia (overall), fatal and non-fatal pneumonia (separately) in moderate/severe SLE patients initiating anifrolumab and in comparable moderate/severe SLE patients who do not initiate anifrolumab (exposed to SLE SOC).
10) To compare hazard rates of the first occurrence of opportunistic serious infections, other serious infections, pneumonia (overall), fatal and non-fatal pneumonia (separately) in moderate/severe SLE patients initiating anifrolumab versus comparable moderate/severe SLE patients who do not initiate anifrolumab (exposed to SLE SOC), when feasible (i.e., if sample size allows).
Exploratory objectives
The following objectives pertain to the serious infection outcomes:
11) To estimate the incidence of recurrent infections leading to hospitalisation, in moderate/severe SLE patients initiating anifrolumab and in comparable moderate/severe SLE patients who do not initiate anifrolumab (exposed to SLE SOC).
12) To compare the hazard rates of recurrent infections leading to hospitalisation in moderate/severe SLE patients initiating anifrolumab and in comparable moderate/severe SLE patients who do not initiate anifrolumab (exposed to SLE SOC), when feasible.
STUDY DESIGN: This long-term safety study is a cohort study based on secondary use of data from Denmark, France, Germany, and Spain. The study will start after anifrolumab market launch dates in each of these countries. For malignancies, the study period will end on the 31st of May 2029 and for serious infections it will end on the 31st of May 2025.
This study is based on a prevalent new-user design, i.e., adult patients with a SLE diagnosis, previous exposure to SLE SOC (indicated for moderate to severe SLE), uncontrolled SLE and initiating anifrolumab (exposed to anifrolumab) will be compared with a comparable group of patients who do not initiate anifrolumab (exposed to SLE SOC). Time-based exposure sets will be created to identify comparative study cohorts. Exposure sets will be defined within a time interval (+/- 45 days, [wider time intervals may be considered if matches cannot be obtained using this time window]) of anifrolumab initiation where all comparator members within each set match the anifrolumab exposed patient according to time since first SLE SOC prescription. Additionally, to increase comparability, comparators will be selected taking into account SLE disease activity (uncontrolled SLE despite SOC), disease severity and matched on propensity scores (PS).
Two cohorts will be constructed for the outcomes of interest (a malignancies cohort – for evaluating new malignancies as a composite outcome and for malignancies’ sub-types; and a serious infections cohort – for evaluating serious infections outcomes), considering outcome-specific inclusion and exclusion criteria. In the main analysis, on-treatment definition of exposure will be considered: patients will be considered at risk while exposed to the first anifrolumab treatment (or the SOC treatment for the exposed to SLE SOC group). For the analysis of malignancy outcomes, follow-up will start after a 12-months latency period after the index date and an additional 12-months period at risk (off-drug) will be considered to capture outcomes that might occur after drug cessation.
EXPOSURE VARIABLES: Patients exposed to anifrolumab will be defined by the initiation of anifrolumab (anatomical therapeutic chemical code L04AA51).
SLE SOC patients will be defined by the use of drugs indicated for moderate to severe SLE (in addition to antimalarials and/or low dose corticosteroids): medium to high dose corticosteroids, immunosuppressants (synthetic or biologics [except anifrolumab]), plasmapheresis or intravenous immunoglobulins.
OUTCOME VARIABLES
– Primary outcomes
Malignancies will be defined as the first coded diagnosis for haematological malignancies and solid tumours available in the data sources.
Serious infection will be defined as an infection leading to hospitalisation, use of intravenous antimicrobials or an infection-related death, operationalised as:
Infections leading to hospitalisation: a) infection diagnosis as part of a hospitalisation episode or b) infection diagnosis in primary or secondary care settings up to 7 days before hospitalisation.
• Prescription/administration of intravenous antimicrobials.
• Infection-related death: recorded diagnosis of infection in primary or secondary care settings with record of death within the subsequent month.
– Secondary outcomes
Specific types of malignancies: haematologic, solid and skin malignancies.
Serious infection components: infection leading to hospitalisation, infection requiring treatment with IV antimicrobials and infection-related death.
Serious infection types grouped as opportunistic serious infections, other serious infections, pneumonia (overall), fatal and non-fatal pneumonia (separately)
– Exploratory outcomes
All episodes of infections leading to hospitalisation.
OTHER VARIABLES: The following variables will be used to describe the cohorts’ characteristics and to control for confounding: demographics, lifestyle characteristics, SLE disease history (including measures of disease severity and activity), medical history and comorbidities, healthcare resource utilisation and other risk factors specific for malignancies and/or serious infections.
DATA SOURCES: This study will use secondary data from multiple countries. The final list of data sources is based on a feasibility assessment conducted between June 2022 and January 2023. All data elements for this study will be collected from information routinely recorded in the regional and national data sources of:
– Denmark (Danish National Registries)
– France (National Health Data System)
– Germany (Statutory Health Insurance Claims)
– Spain (Information System for the Development of Primary Care Research)
STUDY SIZE: Study size estimations were carried out for the primary outcomes (serious infections and malignancies) at the meta-analysis level varying four parameters – hazard ratio, standard deviation (SD) of hazard ratio, background estimate of the primary outcome, and matching ratios of exposed:non-exposed patients. Simulations were run to calculate expected power to detect a pre-specified hazard ratio (HR) given an overall sample size; and to get exact sample size estimates to achieve 80% power to detect the pre-specified HR. Three specifications of HR threshold were considered, 1.5, 1.8, and 2.0, along with two pre-specified SD (0.1 and 0.2) of HR to get an effect-distribution in lognormal space within which the true HR lies. The background estimate of incidence of serious infections was varied from 9.6 to 39.8 per 1000 person-years, while the background estimate of incidence of malignancies incidence was varied from 5.26 to 6.31 per 1000 person-years. Matching ratios of exposed to non-exposed patients were specified as 1:1 and 1:3. Achieved power was computed for sample sizes ranging from a total of 500 anifrolumab exposed patients across all study countries to 5000 anifrolumab exposed across all study countries.
Approximately 1,312 anifrolumab exposed patients across all study countries would be necessary to achieve 80% power to detect a true HR of 1.5 (geometric mean of 0.405 in lognormal space) from an effect distribution with 0.2 SD (geometric coefficient of variation of 0.132 in lognormal space) for serious infections outcome. This assumes a matching ratio of 1:3, and background IRs of serious infection of 39.8 per 1000 person years.
Approximately 3,195 anifrolumab exposed patients across all study countries would be necessary to achieve 80% power to detect a true HR of 1.5 (geometric mean of 0.405 in lognormal space) from an effect distribution with 0.2 SD (geometric coefficient of variation of 0.132 in lognormal space) for malignancy outcome. This assumes a matching ratio of 1:3, and background IRs of malignancies of 6.31 per 1000 person years.
STATISTICAL ANALYSES: The data analysis for all the study objectives will be performed separately for each data source and using appropriate study sub-cohorts. In this study, PS adjustment will be performed separately in each data source to control for confounding using a prevalent new user design. First, time-based exposure sets will be defined based on duration of SLE SOC use: for each exposed patient, the time-based exposure set will include all unexposed patients with a similar duration of SLE SOC use as the exposed when they become exposed to anifrolumab. A single time-conditional propensity score model will be fit using all the exposure sets. This model will then be used to compute a time-conditional propensity score for all patients (exposed and unexposed) in each time-based exposure set. Each exposed patient will then be matched with up to three unexposed patients from his time-based exposure set by selecting those patients with the closest time-conditional propensity score. Covariate balance after matching will be assessed by calculating the absolute standardised differences for each variable between the study sub-cohorts.