Kidney Failure Events, Cardiovascular Disease Events, and All-Cause Mortality in Patients with IgA Nephropathy in a Real-World Database

Visual Abstract


Introduction
IgA nephropathy (IgAN) is an immune complex-mediated inflammatory disease, which, although rare at the population level, is the most common primary glomerular disease in young non-Hispanic White adults in the United States. 1 Because of glomerular injury, patients develop proteinuria, hematuria, and progressive loss of GFR, ultimately leading to kidney failure (KF). 2,3he Kidney Disease Improving Global Outcomes (KDIGO) guidelines recognize proteinuria $0.75-1 g/d as indicative of a high risk of progression and recommend reduction to ,1 g/d as a therapeutic goal. 4In addition, more advanced CKD stages (eGFR ,60 ml/min per 1.73 m 2 ) are associated with both cardiovascular and allcause mortality. 5Dynamic assessment of patient risk is recommended to guide treatment decisions, and improved cardiovascular screening in patients with increased proteinuria and high risk of disease progression is critical. 4ncreasing proteinuria is associated with an increase in cardiovascular risk.6][7] However, there is limited research examining this relationship in IgAN.][10][11][12][13][14] However, a recent observational cohort study, from a Canadian registry, reported a crude incidence of CVD events (any major CVD event or revascularization procedure) of 12.2 per 1000 person-years, translating to an absolute 10-year risk of 7.4%. 15The standardized incidence ratio for CVD events compared with the general population was 1.38.Proteinuria and lower eGFR were both associated with an increased incidence of CVD events although these associations were not specific to IgAN. 15 In addition, data from a registry-based analysis of patients with biopsyproven IgAN suggest that almost all patients with IgAN were expected to progress to KF within their lifetime, regardless of age, 16 with the implication that all patients are subsequently exposed to more general CKD progression-related adverse cardiovascular outcome risk.Despite this, there remains a need to assess more comprehensively the association of proteinuria reduction and slowing of eGFR decline with CVD events, specifically in IgAN.
In a meta-analysis of 13 IgAN randomized trials, change in proteinuria was found to be associated with KF events. 17,18The authors reported a strong and consistent relationship between the level/duration of proteinuria and the decline or loss of kidney function. 17,180][21] In our initial investigation into the burden associated with IgAN, we identified that healthcare resource utilization and costs were higher for patients with high-risk proteinuria ($1 g/d; observed in 54% of the overall and 55% of the adult population at baseline, respectively) and worsening kidney function. 22he aim of this study was to evaluate the prognostic effects of proteinuria on the outcome of CVD/all-cause mortality and KF/all-cause mortality using a US real-world data source.In addition, we evaluated the relationship between progression to KF and occurrence of CVD/mortality events.In addition, we completed exploratory analyses of CVD and nephrotic syndrome event rates by proteinuria and post-KF status and the incremental costs associated with nephrotic syndrome events, CVD events, and CKD progression.

Study Design and Data Source
This was an observational, retrospective cohort study using Optum's deidentified Market Clarity Data and proprietary Natural Language Processing (NLP) data. 23The Optum deidentified Market Clarity Dataset deterministically links electronic health record data from providers across the care continuum with historical, linked administrative claim data, pharmacy claims, physician claims, and facility claims (with clinical information) and is inclusive of medications prescribed and administered.The Optum NLP system was developed using vocabulary from the Unified Medical Language System that includes multiple medical dictionaries such as the Logical Observation Identifiers Names and Codes, the Systemized Nomenclature of Medicine-Clinical Terms, and RxNorm, a listing of generic and branded drugs (among others).NLP concepts are identified and created based on broad topics such as medications, signs, disease, and symptoms (SDS) terms, measurements, and observations.The data are harvested from the notes fields within the electronic medical records provided to Optum from more than 50 large health care systems throughout the United States.The data used for the development of each NLP concept are deidentified, and accuracy is verified through a series of quality assurance steps before release for use.Each NLP concept included in the data is associated with a unique subject record and a date of observation, allowing longitudinal tracking of concepts over time.
The Optum Market Clarity Dataset is fully Health Insurance Portability and Accountability Act compliant and contains statistician-certified, deidentified data.Institutional Review Board approval was not required for this study.

Study Population
Because of the limited data available on the presence of a kidney biopsy, the study population included adult (age $18 years) patients with at least two SDS NLP term entries for "IgA nephropathy," "immunoglobulin A nephropathy," "berger's disease," "berger's nephropathy," "IgA glomerulonephritis," or "immunoglobulin A glomerulonephritis," within 180 and $30 days apart within the identification period.Patients with negation terms (e.g., "deny," "failed," "ignore," "n/a," "negative," "question," "reject," "rule out," "uncertain," and "unspecified") in relation to the IgAN SDS term were excluded (Supplemental Figure 2).Patients were required to have $6 months of preindex activity (baseline period).The index date was the first IgAN NLP term within the identification period.Patients with evidence of coronavirus disease 2019 before or after index were excluded (Figure 1).Patients were excluded from the proteinuria-CVD/mortality cohort if they had evidence of KF (on the basis of diagnosis codes, eGFR ,15 ml/min per 1.73 m 2 , procedure codes for dialysis or kidney transplant) or CVD at baseline (CVD event during baseline period or evidence of myocardial infarction [MI], congestive heart failure, or stroke by the Charlson comorbidity index [CCI]), or if they did not have a proteinuria measurement during the baseline period.For the proteinuria-KF/mortality cohort, patients were excluded if they had evidence of KF at baseline or if they did not have a proteinuria measurement during the baseline period.For the KF-CVD/mortality cohort, patients were excluded if they had evidence of CVD at baseline (CVD event during baseline period or evidence of MI, congestive heart failure, or stroke by the CCI).

Exposures and Outcomes
Baseline demographics and clinical characteristics included age, sex, region, race/ethnicity (collected by database), insurance type, eGFR, CKD stage, KF, proteinuria, and CCI.
To assess the effect of proteinuria, adjusted hazard ratios (HRs) with 95% confidence intervals (CIs) for CVD/allcause mortality and KF/all-cause mortality events were reported.Adjusted HRs with 95% CIs were also reported for the association between KF status and CVD/all-cause mortality events.
Proteinuria level was stratified by the KDIGO threshold for high-risk proteinuria (,1 versus $1 g/d).To support proteinuria-based analyses, all proteinuria data were presented as g/d with urinary protein-creatinine ratio values in g/g converted to g/d.KF status (pre versus post) was determined based on any of the following: diagnosis codes, eGFR ,15 ml/min per 1.73 m 2 , and procedure codes for dialysis or kidney transplant, at any time during follow-up.
CVD events were defined as patients with $1 hospital admission with a primary diagnosis of MI, unstable angina, ischemic stroke, transient ischemic attack, or congestive heart failure or $1 inpatient or outpatient revascularization procedure (percutaneous coronary intervention, coronary artery bypass graft).Baseline CVD events included those occurring in the 6 months before the index date.
KF events were defined using the same components of the definition for KF status.The first occurrence of the components was considered an event; therefore, patients could only have one occurrence of each component.
All-cause mortality was defined as patients with a death date, calculated as the last day of the death month.

Statistical Analyses
Baseline demographic and clinical characteristics were analyzed descriptively, including counts and percentages for categorical variables and means with SD and medians with first and third quartiles (Q1-Q3) for continuous variables.
Kaplan-Meier analysis was conducted to assess the association between baseline proteinuria and CVD/mortality or KF/mortality events during follow-up and between baseline KF status and CVD/mortality events during follow-up.
Multivariate Cox proportional hazards models with time-dependent predictors were used to assess the association between proteinuria and CVD/mortality or KF/mortality events, and between KF status and CVD/ mortality events.The time-dependent methodology allowed for all time for a given patient to be used in the models: time with proteinuria ,1 and/or $1 g/d; time with pre-KF and/or post-KF status.Patients with only baseline proteinuria variables were included in the modeling analysis, and their baseline values were used to categorize proteinuria during follow-up.
For the proteinuria and CVD/mortality analysis, censoring occurred at the end of database activity, end of the study period, or transition to post-KF.For the proteinuria and KF/mortality analysis, censoring occurred at the end of database activity or end of the study period.In the KF status and CVD/mortality analysis, censoring occurred at the end of database activity or end of the study period.Full details on the model establishment and variable section are included in the Supplemental Material.
A Cox proportional hazards model adjusting for age, insurance status, sex, diabetes or metastatic solid tumor (CCI), and use of diuretics at baseline was conducted to determine the association between elevated proteinuria and CVD/mortality events in the follow-up period.To determine the association between elevated proteinuria and KF/ mortality events in the follow-up period, a Cox proportional hazards model was conducted adjusting for age, insurance status, age/insurance interaction, sex, having a CVD event in the baseline period, baseline CKD stage, liver disease or peptic ulcer disease (CCI), and use of the following medications at baseline: calcineurin inhibitors, mycophenolate, potassium binders, mineralocorticoid receptor antagonists, and renin-angiotensin system inhibitors.For the KF status and CVD/mortality event analysis, a separate Cox proportional hazards model was conducted adjusting for age, insurance status, age/insurance interaction, sex, race/ethnicity, Charlson comorbidities: cancer, liver disease, peripheral vascular disease, diabetes, chronic pulmonary disease, and rheumatologic disease, and use of the following medications at baseline: diuretics, b blockers, calcium channel blockers, and renin-angiotensin system inhibitors.
Details on the methods for the exploratory event rate and cost analyses are included in the Supplemental Material.
The data analysis for this study was generated using SAS software.Copyright 2023 SAS Institute Inc. SAS and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS Institute Inc., Cary, NC.

KF Status Was Associated With CVD/All-Cause Mortality Events
In the KF-CVD/mortality cohort, 632 of 5784 patients with baseline pre-KF status experienced a CVD/all-cause mortality event over a median (Q1-Q3) follow-up of 44.4 (22.9-73.0)months, while 221 of 1029 patients with baseline post-KF status experienced a CVD/all-cause mortality event over a median (Q1-Q3) follow-up of 33.5 (15.3-60.1)months.The Kaplan-Meier analysis resulted in a significantly higher risk of CVD/mortality for patients with baseline post-KF status (P , 0.001; Figure 4A).In the unadjusted Cox proportional hazards model, post-KF status was associated with elevated risk for CVD/mortality events (HR [95% CI]: 3.85 [3.35 to 4.43]; P , 0.001).Post-KF status was also associated with elevated risk for CVD/all-cause mortality events (HR [95% CI]: 3.28 [2.82 to 3.81]; P , 0.001) after adjustment (Figure 4B).

Discussion
To our knowledge, this is the first real-world study analyzing the effect of proteinuria on the risk of CVD/ mortality events in patients with IgAN in the United States.We demonstrated that elevated proteinuria $1 g/d and progression to KF in patients with IgAN were associated with a significantly higher risk of patients experiencing a CVD/mortality event.Our analysis demonstrated consistent results of the prognostic effects of proteinuria through different approaches: the Kaplan-Meier  analysis by the baseline proteinuria, the univariate Cox model by time-dependent proteinuria, and the covariateadjusted Cox model by time-dependent proteinuria.We note that time-dependent proteinuria dynamically reflects disease progression and has been validated as the best metric to account for the prognostic effects of proteinuria over the majority of the follow-up time for IgAN, better than baseline proteinuria and time-averaged proteinuria. 24These results highlight the potential for reduction in CVD/mortality risk with effective reduction of proteinuria and prevention of KF in the management of IgAN.
We also showed that pre-KF elevated proteinuria was associated with a significantly elevated risk of progression to KF (defined by diagnosis code, eGFR, dialysis, or transplant status) and/or mortality events.][21] It should be noted that even among patients with baseline proteinuria ,1 g/d in our cohort, 25% of patients progressed to KF, showing similarity with a long-term national registry in the United Kingdom. 15These results suggest that a substantial proportion of patients progress to KF despite having proteinuria ,1 g/d, the KDIGO recommended threshold for assessing the risk of progression in IgAN. 4 From our exploratory analyses, CKD stage progression and the occurrence of CVD events were associated with substantial incremental costs although we noted a high level of variability in these results (Supplemental Tables 1-2).The rate of CVD events per 100 person-years was high in patients with elevated baseline proteinuria ($1 g/d) and for patients with KF at baseline (Supplemental Table 3).Combined, these results align with previous research on the economic burden associated with IgAN 22 and provide additional insight into the role CVD-related events may play.Our results must be considered in the context of potential limitations associated with using secondary data sources.For example, this study was limited to patients in the Market Clarity Data and may not be representative of the broader US population.In addition, missing data or errors are inherent in retrospective analyses and errors in detection of IgAN-related terms in patient records may introduce bias.Patients with IgAN-related terms may have more severe disease, the IgAN diagnoses in our cohort were not all biopsy-proven, and MEST-C scores were not available in the dataset.In addition, given limitations in the data, we were unable to identify whether patients were diagnosed with secondary IgAN or IgA vasculitis, which may bias results.It is possible that we have not captured events that occur post-KF because patients transitioned from commercial to traditional Medicare coverage, potentially leading to an underestimation of event rates in this group of patients.The full spectrum of services may not be captured within the dataset, and claims containing enrollment information may have been missed.Within the electronic health record dataset, patients who may have received services in facilities outside of Optum's data would not be captured.Finally, the dataset does not include information on the cause of death; therefore, we were unable to separate non-CVD and non-KF death from our event outcomes.
A clinically meaningful proportion of patients with IgAN experienced CVD events, and our results showed that elevated proteinuria and progression to KF were associated with a statistically higher risk of CVD/ comorbidity index; CVD, cardiovascular disease; KF, kidney failure; Q, quartile.IgAN Outcomes in a US Real-World Cohort, Lerma et al.

Table 1 .
Baseline patient demographics and clinical characteristics Figure 2. Elevated proteinuria was associated with CVD/all-cause mortality events.Kaplan-Meier analysis (A) and Cox proportional model (B) results for proteinuria and CVD/mortality events.Log-rank P value from Kaplan-Meier analysis and HR (95% CI) in Kaplan-Meier plot from univariate Cox proportional hazards model.Forest plot generated from the multivariable Cox proportional hazards model.Arrow at the end of the CI bar for metastatic solid tumor indicates the upper end of the CI extends beyond the x axis.*N (%) at baseline.**Reference group5commercial and all other insurance.CI, confidence interval; HR, hazard ratio; Q1, first quartile; Q3, third quartile.
Figure 3. Elevated proteinuria was associated with KF/All-cause mortality events.Kaplan-Meier analysis (A) and Cox proportional model (B) results for proteinuria and KF/mortality events.Log-rank P value from Kaplan-Meier analysis and HR (95% CI) in Kaplan-Meier plot from univariate Cox proportional hazards model.Forest plot generated from the multivariable Cox proportional hazards model.*N(%) at baseline.Reference groups5**commercial and all other insurance; ***stage 1 and unknown.CNIs, calcineurin inhibitors; MCRA, mineralocorticoid receptor antagonists; RAS, renin-angiotensin system.Figure 4. KF status was associated with CVD/all-cause mortality events.Kaplan-Meier analysis (A) and Cox proportional model (B) results for KF status and CVD/mortality events.Log-rank P value from Kaplan-Meier analysis and HR (95% CI) in Kaplan-Meier plot from univariate Cox proportional hazards model.Forest plot generated from the multivariable Cox proportional hazards model.*N (%) at baseline.Reference groups5**commercial and all other insurance; ***non-Hispanic White.