Long-Term Outcomes in IgA Nephropathy

Visual Abstract Background IgA nephropathy can progress to kidney failure, and risk assessment soon after diagnosis has advantages both for clinical management and the development of new therapeutics. We present relationships among proteinuria, eGFR slope, and lifetime risks for kidney failure. Methods The IgA nephropathy cohort (2299 adults and 140 children) of the UK National Registry of Rare Kidney Diseases (RaDaR) was analyzed. Patients enrolled had a biopsy-proven diagnosis of IgA nephropathy plus proteinuria >0.5 g/d or eGFR <60 ml/min per 1.73 m2. Incident and prevalent populations and a population representative of a typical phase 3 clinical trial cohort were studied. Analyses of kidney survival were conducted using Kaplan–Meier and Cox regression. eGFR slope was estimated using linear mixed models with random intercept and slope. Results The median (Q1, Q3) follow-up was 5.9 (3.0, 10.5) years; 50% of patients reached kidney failure or died in the study period. The median (95% confidence interval [CI]) kidney survival was 11.4 (10.5 to 12.5) years; the mean age at kidney failure/death was 48 years, and most patients progressed to kidney failure within 10–15 years. On the basis of eGFR and age at diagnosis, almost all patients were at risk of progression to kidney failure within their expected lifetime unless a rate of eGFR loss ≤1 ml/min per 1.73 m2 per year was maintained. Time-averaged proteinuria was significantly associated with worse kidney survival and more rapid eGFR loss in incident, prevalent, and clinical trial populations. Thirty percent of patients with time-averaged proteinuria of 0.44 to <0.88 g/g and approximately 20% of patients with time-averaged proteinuria <0.44 g/g developed kidney failure within 10 years. In the clinical trial population, each 10% decrease in time-averaged proteinuria from baseline was associated with a hazard ratio (95% CI) for kidney failure/death of 0.89 (0.87 to 0.92). Conclusions Outcomes in this large IgA nephropathy cohort are generally poor with few patients expected to avoid kidney failure in their lifetime. Significantly, patients traditionally regarded as being low risk, with proteinuria <0.88 g/g (<100 mg/mmol), had high rates of kidney failure within 10 years.


Introduction
IgA nephropathy is the most common form of primary glomerulonephritis and a major cause of CKD and kidney failure worldwide. 1,2 Most patients are diagnosed before age 40 years. [3][4][5] Life expectancy in countries where IgA nephropathy is commonly encountered varies from 70 to 85 years. 6 Current perceptions of risk of progression are based on outcome data typically spanning 10 or 15 years, 5,7 a relatively short period that represents less than half the remaining lifespan of the typical patient with IgA nephropathy.
To facilitate investment in new therapies, a focus of recent research has been to identify surrogate end points that predict long-term clinical outcomes. 8,9 Reduction in proteinuria has recently been accepted by regulatory authorities as a reasonably likely surrogate end point for IgA nephropathy. 10 The design of ongoing phase 2 and 3 randomized controlled trials (RCTs) focuses on generating data over a 1-to 2-year period for proteinuria and rate of eGFR loss. After the inception of this approach, more than 15 phase 2 or 3 RCTs for new therapeutic approaches for IgA nephropathy are in progress in 2022.
Outstanding questions are first, the extent to which such short-term proteinuria and eGFR data predict the long-term rate of eGFR loss and risk of kidney failure in IgA nephropathy, and second, the degree of proteinuria reduction that would be associated with slowing eGFR decline so that kidney failure is not reached during a patient's lifetime. According to the Kidney Disease Improving Global Outcomes (KDIGO) 2021 Clinical Practice Guideline for the Management of Glomerular Diseases, 11 a reduction of proteinuria to ,1 g/d is considered as a reasonable treatment target in patients with IgA nephropathy. However, long-term outcomes for patients with proteinuria ,1 g/d need to be better understood.
To address these issues, we studied clinical data from patients with IgA nephropathy enrolled into the UK National Registry of Rare Kidney Diseases (RaDaR), a national registry with validated end points and no withdrawals from follow-up. We analyzed the characteristics and outcomes of patients with IgA nephropathy and delineated the burden of disease progression by age at diagnosis to gain insight into the magnitude of proteinuria change and rate of eGFR loss required of new treatments to prevent kidney failure in a patient's lifetime. We assessed the relationships among key parameters including baseline and time-averaged proteinuria, rate of eGFR loss, and kidney survival. Finally, we assessed the value of short-term changes in proteinuria and eGFR slope, typically measured in IgA nephropathy RCTs, for predicting longer-term clinical outcomes. Through these analyses, we aim to improve understanding of current outcomes and begin discussion on how lifetime kidney failure risk should be evaluated and management strategies developed such that no patient with IgA nephropathy reaches kidney failure.

Data Source
This retrospective cohort study used data from the RaDaR IgA nephropathy cohort, for which enrollment began in 2013. Patients eligible for enrollment must have a biopsyproven diagnosis of IgA nephropathy plus proteinuria .0.5 g/d or eGFR ,60 ml/min per 1.73 m 2 at any time in the history of their disease. All forms of secondary IgA nephropathy are excluded. Further description of the data source is included in the Supplemental Methods.

Study Population and Definitions
Summaries of the eligibility criteria, patient disposition, and study attrition are provided in Supplemental Figure 1.
Urinary protein-creatinine ratio (UPCR) measurements were only available at diagnosis for 23% of adult patients in the full-analysis population (Table 1). Four adult subpopulations were, therefore, developed to facilitate focused analysis of proteinuria versus eGFR slope and clinical outcomes. Population 1 is a representative incident population examining time-averaged proteinuria over follow-up without requirement for a baseline UPCR at diagnosis. Populations 2 and 3 were selected to be representative of incident and prevalent cases, respectively, and allow assessment of baseline and time-averaged proteinuria. Population 4 is a prevalent population selected using criteria representative of a typical phase 3 RCT. Supplemental Figure 1 and Supplemental Table 1 provide detailed summaries of the four subpopulations. Patients could be in multiple subpopulations (Supplemental Figure 2); the only restriction was that population 2 was a subset of population 1.
Diagnosis was the earliest of either biopsy date or primary kidney diagnosis date recorded in RaDaR. Kidney failure was defined as the first occurrence of either long-term KRT, a confirmed 12 eGFR ,15 ml/min per 1.73 m 2 , or CKD stage 5 recorded in RaDaR. Kidney survival was defined as the absence of either kidney failure or death. eGFR was calculated using the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) 2021 formula 13 in adults and the bedside Schwartz equation 14 for patients younger than 18 years.
Annualized eGFR slopes were calculated using linear regression to fit a straight line through patients' mean eGFR values for each 3-month period of follow-up. Slopes were calculated for total follow-up and 6-30 months after baseline.
UPCR values were recorded and presented in g/g with urinary albumin-creatinine ratio (UACR) values converted to UPCR using UPCR5UACR/0.7. [15][16][17] For comparison with other studies, a UPCR of 0.88 g/g (100 mg/mmol) may be considered comparable with protein excretion of 1 g/d. 18 Time-averaged proteinuria was defined as the timeweighted averages for UPCR (Supplemental Methods). 5 Life expectancy was based on World Bank year of birth and sex estimates.

Statistical Analyses
Continuous variables were reported as mean (SD) and median (interquartile range [IQR]) and categorical variables as frequencies and percentages. Kidney survival times were calculated from baseline to first kidney failure event, death from any cause, or end of follow-up for those with no event. Kaplan-Meier plots display kidney survival estimates; logrank tests were used for comparison where appropriate. Analyses of kidney survival were conducted using Cox regression. eGFR slope was estimated using linear mixed models with random intercept and slope. eGFR slope and kidney survival analyses were adjusted for age, sex, ethnicity, baseline CKD stage, baseline UPCR category, time-averaged proteinuria category, and other factors found to have significant associations with outcome during univariable analysis. Linear mixed models were used to analyze the association between eGFR slope and log (% change) in proteinuria, calculated from baseline value to both 6-12-month and 6-24-month time-averaged proteinuria. Data were analyzed using SAS 9.4. A two-sided P value of ,0.05 was considered statistically significant with no correction for multiple comparisons.

Characteristics, Outcomes, and Lifetime Kidney Failure Risk
The full-analysis population included 2439 patients: 2299 adults and 140 children. Kaplan-Meier survival analysis of pediatric versus adult patients demonstrated that patients diagnosed at younger than 18 years had significantly longer median kidney survival than adults (log-rank P , 0.001) ( Figure 1A). In the adult cohort, baseline CKD stage was the strongest predictor of 10-year kidney survival and eGFR slope in unadjusted and multivariable analyses (Supplemental Table 2). Asian ethnicity was also associated with higher risk of disease progression (  Kaplan-Meier survival analysis by decade of age at diagnosis showed that most patients progressed to kidney failure within 10-15 years in all age groups ( Figure 1E). On the basis of eGFR and age at diagnosis, almost all patients were at risk of kidney failure within their expected lifetime unless a rate of eGFR loss #1 ml/min per 1.73 m 2 per year was maintained from diagnosis ( Figure 1F). An annual eGFR decline of 3 ml/min per 1.73 m 2 , if sustained, would result in 100% of patients younger than 40 years at diagnosis reaching kidney failure within their expected lifetime ( Figure 1G). Strikingly, a decline of as little as 1 ml/min per 1.73 m 2 per year would still result in approximately 40% of adult patients younger than 50 years at diagnosis reaching kidney failure within their expected lifetime.

Relationship between Proteinuria and Kidney Failure
Baseline characteristics and clinical outcomes for all subpopulations are presented in Supplemental Table 3.
Time-averaged proteinuria from diagnosis over total follow-up (median [Q1, Q3], 4.5 [2.5, 6.8] years) (Supplemental Table 3) was analyzed as a categorical variable in population 1, with higher time-averaged proteinuria significantly associated with worse kidney survival and more rapid eGFR loss ( Table 2; Supplemental Table 4). Similar results were seen with time-averaged proteinuria over 0-24 months. Accompanying Kaplan-Meier survival analyses calculated from both total and 0-24-month timeaveraged proteinuria showed that patients with time-averaged proteinuria .0.88 g/g (.100 mg/mmol) were likely to progress to kidney failure or death more quickly than patients with time-averaged proteinuria ,0.88 g/g (Figure 2).
Considering time-averaged proteinuria over total follow-up, this analysis demonstrates that 30% of patients with time-averaged proteinuria of 0.44 to ,0.88 g/g (50 to ,100 mg/mmol) and approximately 20% of patients with time-averaged proteinuria ,0.44 g/g developed kidney failure within 10 years (Table 2). Time-averaged proteinuria calculated over 0-24 months showed similar results. Kaplan-Meier analysis demonstrated that kidney survival probability decreased by almost half over the 15-year follow-up among patients with time-averaged proteinuria of 0.44 to ,0.88 g/g (Figure 2). KDIGO guidelines describe reduction in proteinuria to ,1 g/ d (approximately equivalent to UPCR ,0.88 g/g) as a reasonable treatment target in patients who remain at high risk of progressive CKD. 11 However, this analysis indicated poor long-term outcomes for many patients at time-averaged proteinuria levels typically perceived as low risk. Incident (population 2) and prevalent (population 3) populations were used to assess the relationship between time-averaged proteinuria and clinical outcomes among patients commonly perceived as high or low risk on the basis of baseline UPCR values ($0.88 versus ,0.88 g/g). Among patients in population 3 with baseline UPCR $0.88 g/g, the probability of kidney survival at 5, 10, and 15 years was lower with higher time-averaged proteinuria ( Table 3). The same association of poorer kidney survival with higher time-averaged proteinuria was evident among patients with baseline UPCR ,0.88 g/g, highlighting the strong predictive value of proteinuria, even in patients currently perceived to be at low risk of progression. Similar results were observed among newly diagnosed patients in population 2 (Supplemental Table 5).

Clinical Outcomes in the RCT-Representative Population 4
Population 4 was selected to be representative of a typical phase 3 RCT: age 18 years or older, baseline UPCR $0.88 g/g (considered comparable with protein excretion $1 g/d), 18 and eGFR $30 ml/min per 1.73 m 2 , thereby excluding patients commonly perceived not to be high risk. This population was used to determine how changes in proteinuria may be associated with near-and long-term rate of eGFR loss and kidney survival. In population 4, comparisons of kidney survival and rate of eGFR loss were made between time-averaged proteinuria over 6-24 months (representing a full RCT duration) and 6-12 months (representing an interim analysis).
In both the 6-12-and 6-24-month analyses, higher timeaveraged proteinuria was associated with higher risk of kidney failure and greater mean eGFR loss (Table 4). Greater separation of outcomes for time-averaged proteinuria over 6-24 months suggested that there was a value in longer follow-up for estimating long-term outcomes, although time-averaged proteinuria over 6-12 months produced similar results, allowing reliable prediction of longer-term outcomes. Additional outcomes for population 4 are presented in Supplemental Table 6.
Kaplan-Meier survival analysis demonstrates that median kidney survival was significantly lower at timeaveraged proteinuria $1.76 g/g using either the 6-12-or 6-24-month time-averaged proteinuria windows compared with lower time-averaged proteinuria (log-rank P , 0.001) (Figure 3). A decrease from baseline to 6-12month time-averaged proteinuria was associated with a   Figure 4A). The same pattern was observed in the total rate of eGFR loss when proteinuria was measured over 6-24 months ( Figure 4B). Each 10% decrease in proteinuria from baseline was estimated to be associated with a hazard ratio (95% CI) for kidney failure/death of 0.89 (0.87 to 0.92) after adjusting for age, sex, baseline eGFR, and time from diagnosis to baseline ( Figure 4C). Five-year kidney survival rates were lower with greater eGFR loss (6-30 months) ( Figure 4D

Discussion
We assessed a large cohort of patients with biopsyproven IgA nephropathy enrolled into RaDaR; 77% were White and 94% were adult, with a mean (SD) age at diagnosis (among adults) of 42 (14) years (Table 1). Using representative incident and prevalent populations, we provide insight into the burden of IgA nephropathy, lifetime kidney failure risk, and value of short-term proteinuria and eGFR slope assessments for predicting long-term outcomes.
Outcomes for both adult and pediatric cohorts were poor, with a median (95% CI) kidney survival of 10.8 (10.0 to 12.0) years and mean (SD) age at kidney failure of only 49 (14) years for adults and 21.6 (15.9 to not estimable) years and 27 (10) years, respectively, for those diagnosed while children (Table 1). When assessing adult incident (population 1) and prevalent (population 3) populations, we show that higher time-averaged proteinuria was associated with a greater likelihood of progressing to kidney failure more quickly. These findings in our predominantly White cohort are aligned with findings in other White cohorts 7,19 and an Asian cohort. 5 Also consistent with published literature, 1,10 Asian ethnicity was found to be a risk factor for progression (Supplemental Tables 2 and 7). Phase 2 and 3 RCTs in IgA nephropathy currently use proteinuria as a surrogate end point and typically recruit patients with baseline proteinuria .1 g/d (approximately equivalent to 0.88 g/g), which is perceived as high risk in accordance with KDIGO. 11 However, focusing on patients with proteinuria .1 g/d means that long-term outcomes for those with lower proteinuria measurements remain largely unexplored. We show that in an incident population (population 1), approximately 20% of patients with timeaveraged proteinuria ,0.44 g/g and 30% with timeaveraged proteinuria 0.44 to ,0.88 g/g progressed to kidney failure within 10 years of diagnosis. A similar pattern was observed in a prevalent population (population 3) among patients with baseline UPCR ,0.88 g/g, with higher time-averaged proteinuria associated with significantly poorer outcomes. These results demonstrate that IgA nephropathy cannot be considered a benign condition, even when proteinuria is ,1 g/d. There was substantial improvement associated with time-averaged proteinuria ,0.44 g/g on rate of eGFR loss and lifetime kidney failure risk in patients commonly perceived as being at low risk of progression (Table 3; Supplemental Table 5). This brings into question the appropriateness of using the KDIGO threshold of ,1 g/d as a treatment target in IgA nephropathy. 11 This is the first time that the substantial risk of kidney failure among patients currently perceived as low risk has been demonstrated in a predominantly White population. In a Chinese cohort, Le et al. 5 demonstrated that proteinuria of 0.5 g/d could be a more appropriate risk threshold than 1 g/d.
Age at diagnosis steadily increased over the past 30 years ( Figure 1B). This likely reflects the increasing threshold for kidney biopsy in the United Kingdom, driven by the perception that people with suspected IgA nephropathy and proteinuria ,1 g/d are at low risk for progression and, irrespective of the biopsy findings, would not be eligible for therapy beyond renin-angiotensin-system inhibition, BP control, and lifestyle management. As our results show, this approach fails to identify patients with IgA nephropathy who are at significant risk of kidney failure, where early intervention could affect their lifetime kidney failure risk. In   particular, disease-modifying therapies that specifically target the immune system are more likely to be effective early in the natural history of IgA nephropathy, before the kidneys accumulate significant irreversible fibrosis.
A striking observation in our cohort was that almost all patients with IgA nephropathy were expected to progress to kidney failure within their lifetime, regardless of age or eGFR at diagnosis. We show that eGFR decline of    21 We must, therefore, consider what can be done to reach this target rate of eGFR loss (e.g., earlier diagnosis, intervention at lower levels of proteinuria, new drugs, and new combinations) and how these approaches can be assessed in future RCTs. Short-term assessment of proteinuria in our RCT-representative population (population 4) predicted longer-term outcomes with a relatively high degree of certainty, supporting the validity of shortterm proteinuria changes in RCTs as a surrogate end point. However, we also demonstrate that a more than 80% reduction in time-averaged proteinuria would be needed to achieve the target rate of eGFR decline in that population.
The results of this study must be considered in the context of several limitations inherent in registry analyses. The cohort was representative of patients managed in the United Kingdom and of the genetic and ethnic makeup of the UK population, and its generalizability to populations in other countries is unknown. Recruitment into RaDaR was initiated in 2013; survivor bias may have affected assessment of patients diagnosed earlier than 2013, while ascertainment bias may have reduced the chance of recruiting individuals presenting before 2013 whose clinical course was benign. Availability of patient medication and BP data was a limiting factor in this study. Future analysis of these crucial parameters will be essential to strengthen recommendations regarding the modification of existing patient care approaches.
To achieve the aspiration that no patient with IgA nephropathy should progress to kidney failure, the current approach to patient care needs to be re-evaluated. This will require a lower threshold for biopsy allowing earlier diagnosis and initiation of treatment before extensive, irreversible damage has occurred. We must also consider the use of combination therapies to maximize effectiveness. A lower proteinuria target for assessing treatment response and inclusion in clinical trials of patients presenting with proteinuria levels below 0.88 g/g (100 mg/mmol) should also be considered.

Supplemental Material
This article contains the following supplemental material online at http://links.lww.com/CJN/B758. Supplemental Methods. Supplemental Figure 1. Study population disposition. Supplemental Figure 2. Venn diagram displaying the degree of overlap between proteinuria subpopulations.
Supplemental Table 3. Demographic and clinical characteristics at diagnosis and clinical outcomes during follow-up (populations 1-4).
Supplemental Table 4. Clinical outcomes of proteinuria analysis in population 1.
Supplemental Table 5. Clinical outcomes of proteinuria analysis in populations 2 and 3.