Studying glomerular disease epidemiology: tackling challenges and paving a path forward

Background: Glomerular diseases are a group of rare immune-mediated kidney diseases that affect the glomeruli, or filtering units, of the kidney. Major knowledge gaps remain in our understanding of glomerular disease epidemiology. Efforts to describe glomerular disease distributions based on geographic, demographic, and temporal factors (descriptive epidemiology) are limited by the absence of population-level disease registries in most jurisdictions. The extent to which glomerular disease subtype independently associates with clinical outcomes (analytic epidemiology), especially once kidney disease has progressed to end-stage kidney failure, remains largely unknown. Further, much of what is known regarding glomerular disease epidemiology is derived from the experiences of highly-selected patient populations enrolled in clinical trials or attending academic medical centres. Larger-scale, population-level, studies of glomerular disease epidemiology would help to close knowledge gaps regarding the distribution and determinants of glomerular disease and, in doing so, would inform clinical care, public health policy, and clinical trial design. Hypotheses: Two major hypotheses are explored in this thesis: 1. Significant geographic and temporal variation in glomerular disease frequencies exist, that are not solely explained by racial-ethnic variation, thus supporting a role for socioeconomic and environmental factors in the development of clinically manifest glomerular disease; 2. Glomerular disease subtype independently associates with clinical outcomes even after glomerular disease has advanced to end-stage kidney failure, challenging the prevailing paradigm to group all glomerular disease subtypes together in research and public health reporting of clinical outcomes in patients with end-stage kidney failure. Aims: The overall aim of this research was to close knowledge gaps in glomerular disease epidemiology by identifying geographic and temporal variation in glomerular disease frequency distributions and by determining associations between glomerular disease subtype and clinical outcomes (mortality, cardiovascular events) in patients with end-stage kidney failure. Methods: For the first two manuscripts (Chapters 3 and 4), I analysed two large-scale pathology datasets created by my collaborator, Dr. Charles Jeannette: a) the International Kidney Biopsy Survey (IKBS) that includes kidney biopsy diagnoses and associated patient demographics from 29 international kidney pathology laboratories, which I used to study geographic variation in glomerular disease frequencies within and across racial-ethnic groups; b) the Glomerular Disease Collaborative Network (GDCN), a registry of all kidney biopsies referred to the University of North Carolina since 1986, which I used to study temporal trends in glomerular disease frequencies within and across demographic groups over the last three decades. For the next two manuscripts (Chapters 5 and 6), I analysed data – including physician-reported cause of kidney failure – from virtually all U.S. patients with treated end-stage kidney failure who are enrolled, by federal mandate, in the United States Renal Data System (USRDS). In the first of these two manuscripts, I determined associations between glomerular disease subtype and mortality; in the second, I determined associations between glomerular disease subtype and cardiovascular events. Advanced statistical methods included multivariable regression to handle confounding, proportional sub-distribution hazard models to handle competing events, and multiple imputation to handle missing data. Results: Major findings from these manuscripts include: a) significant differences in glomerular disease frequencies across continents, even among patients with similar racial-ethnic backgrounds; b) significant temporal trends in the relative frequencies of many biopsy-proven glomerular diseases, including stabilization in the 21st century of the rapid increase in focal segmental glomerulosclerosis observed at the end of the 20th century, and a dramatic increase in diabetic glomerulosclerosis over time, to become the second most frequent biopsy-proven glomerular disease diagnosis in the modern era; c) significant differences in the hazards of mortality and cardiovascular events across glomerular disease subtypes, even after accounting for between-group differences in case-mix. Conclusions: In addition to answering specific research questions regarding glomerular disease epidemiology, this research exemplifies the strengths and feasibility of population-level, internationally collaborative, approaches to studying glomerular diseases. Findings from these studies can shape public health policy (e.g. promotion of healthy lifestyle approaches to curb the high frequency of diabetic glomerulosclerosis in contemporary U.S. populations), future research design (e.g. recognising the importance of glomerular disease subtype as a prognostic indicator in studies involving patients with end-stage kidney failure), and clinical care (e.g. formulating differential diagnoses based on patient demographics, or counselling U.S. patients regarding their absolute and relative risks of mortality and cardiovascular events following dialysis initiation).