Age of Onset and Disease Course in Biopsy-Proven Minimal Change Disease: An Analysis From the Cure Glomerulopathy Network

Abstract

Rationale and Objective:

Adolescent- and adult-onset disease minimal change disease (MCD) may have a clinical course distinct from childhood-onset disease. We characterized the course of children and adults with MCD in the Cure Glomerulopathy Network (CureGN) and assessed predictors of rituximab response.

Study Design:

Prospective, multicenter, observational study.

Study Participants:

CureGN participants with proven MCD on biopsy.

Exposures:

Age at disease onset. Initiation of RAAS blockade and immunosuppression including rituximab during the study period.

Outcomes:

Relapse and remission, change in eGFR and kidney failure.

Analytical Approach:

Remission and relapse probabilities are estimated using Kaplan-Meier curves and gap time recurrent event models. Linear regression models were used for the outcome of change in estimated glomerular filtration rate (eGFR). Cox proportional hazards models were used to estimate the association between rituximab administration and remission.

Results:

304 childhood (≤12 years old), 49 adolescent (13-17 years old), and 201 adult onset (≥18 years) participants were included with 2.7-3.2 years of follow-up after enrollment. Children had longer time to biopsy (238 days vs. 23 in adolescents, and 36 in adults, p<0.001) and were more likely to have received therapy prior to biopsy. Children were more likely to be treated with immunosuppression but not RAAS blockade. The rate of relapse was higher in childhood-onset versus adult-onset participants (hazard ratio (HR) 1.69 (95% confidence interval (CI) 1.29-2.21)). The probability of remission was also higher in childhood-onset disease (HR 1.33 (95%CI 1.02-1.72)). eGFR loss in all groups was minimal. Children were more likely to remit after rituximab than adolescents and adults (adjusted HR 2.1, p=0.003). Across all groups, glucocorticoid-sensitivity was associated with a greater likelihood of achieving complete remission after rituximab (adjusted HR 2.62, p=0.002).

Limitations:

CureGN was limited to biopsy-proven disease. Comparisons of childhood to non-childhood cases of MCD may be subject to selection bias, given that childhood cases who are biopsied may be limited to those patients who are least responsive to initial therapy.

Conclusions:

Among patients with MCD who underwent kidney biopsy, there were differences in the course (relapse and remission) of childhood-onset compared to adolescent and adult-onset disease, as well as rituximab response.

Grapical Abstract

Plain Language Summary:

Minimal change disease is a biopsy diagnosis for nephrotic syndrome. MCD is diagnosed in childhood, adolescence, or adulthood. Patients and clinicians often have questions about what to expect in the disease course and how to plan therapies. We analyzed a group of patients followed longitudinally as part of the Cure Glomerulopathy Network (CureGN) and describe differences in disease (relapse and remission) based on the age of onset. We also analyzed rituximab response. We found that those with childhood-onset disease had a higher rate of relapse but also have a higher probability of reaching remission when compared to adolescent or adult-onset disease. Children and all steroid-responsive patients are more likely to achieve remission after rituximab.

Introduction

Minimal change disease (MCD) is classically defined by biopsy findings of normal glomeruli on light microscopy accompanied by diffuse podocyte foot process effacement on electron microscopy but the underlying pathophysiology is not yet well understood.¹ While many with childhood onset nephrotic syndrome are diagnosed and treated clinically without biopsy, MCD is the biopsy pathologic correlate for steroid sensitive nephrotic syndrome.² A biopsy provides clarity regarding diagnosis and pathology details such as global glomerulosclerosis and interstitial fibrosis that help with prognostic assessment.³. In contrast, adults with idiopathic nephrotic syndrome typically undergo biopsy prior to immunosuppressive therapy and MCD is identified in approximately 5-15% of biopsied adult.^4,5 There is some evidence that adolescent- and adult-onset disease may have a distinct course that has not been differentiated clinically and pathophysiologically from childhood-onset disease.⁶ In our study, we align a cohort of children, adolescents, and adults with biopsy proven minimal change disease and describe their subsequent disease course (Figure 1).

Figure 1.

Visual summary figure of study design and outcomes.

Until the Cure Glomerulopathy (CureGN) Study⁷, we lacked adequate sample sizeswith MCD to determine whether there are distinct entities. Few studies have prospectively gathered clinical, treatment, and outcomes data to compare MCD across age groups.⁸ Adult treatment approaches have been mostly borrowed from studies on childhood-onset MCD but response and long term relapsing patterns have not been well documented.⁹ Studies suggest adults may have slower response to glucocorticoids or are more likely to be non-responsive.^8,10-12 Proteinuria remission is indicative of favorable long-term GFR preservation in adult- and childhood-onset disease.¹³ Many prior studies in children include patients without biopsy proven disease. There are no direct comparisons between different age groups describing clinical characteristics, hypertension, or renal insufficiency associated with biopsy proven MCD.^14,15,16 Differentiating the clinical characteristics based on age of onset can provide insight that may support future research and management of MCD.

We describe differences in childhood-onset, adolescent-onset, and adult-onset biopsy-proven MCD using CureGN data. To our knowledge, this is the largest prospective, multi-center longitudinal study of its kind. We report features at disease onset and therapy response. Longitudinal analyses provide insight into the clinical course of this complex relapsing/remitting disease and long-term outcomes (eGFR and kidney failure). In addition, given the growing use of rituximab as a glucocorticoid-sparing agent in MCD^17,18,19-29, we tested the hypothesis that patients who are glucocorticoid-sensitive have a better response to rituximab therapy compared to those who are not glucocorticoid sensitive. While our analyses cannot provide biological evidence of disease differences, the descriptive comparisons between age groups and predictors of successive rituximab therapy set a stage for future studies. For example, before clinical trials can be carried out, it is important to understand whether adults, adolescents, and children should be expected to have similar outcomes. Additionally, evidence from this observational study could serve as a basis to design a randomized clinical trial to test the efficacy of rituximab as a steroid sparing agent.

Methods

Participant cohort

CureGN is an on-going, prospective, multicenter, observational cohort study of individuals with biopsy-proven glomerular diseases.⁷ CureGN excludes patients with diabetes mellitus, systemic lupus erythematosus, HIV-1 infection, active malignancy, hepatitis B or C, kidney failure, and kidney transplant. Participants were enrolled from 70 academic medical centers in the United States, Canada, Italy, and Poland within five years of the diagnostic kidney biopsy and could be enrolled at any stage of disease activity (relapse or remission). This study includes participants enrolled between December 2014 and December 2020 with a pathologic diagnosis of MCD (referred to here as the MCD cohort). Demographics, clinical characteristics, and treatment were collected at enrollment and prospectively. Limited clinical information and treatment history from the time of disease onset and biopsy were collected retrospectively. Participants self-identified race and ethnicity from provided categories. CIinical data were collected approximately every four months after enrollment at in-person or remote study visits. Data collected included non-immunosuppressive treatments, immunosuppressive treatment, clinical laboratory results, and other co-existing conditions. Institutional Review Boards at each center approved the study protocol. All adult participants and legal guardians of minor participants provided written informed consent and, where applicable, minors provided assent.

A subset of the cohort who had not received rituximab (“rituximab-naïve”) prior to enrollment but went on to receive rituximab therapy during the follow up period were identified for sub analysis. For this analysis, only participants who received rituximab after enrollment were included.

Definitions

Outcomes

Urine protein to creatinine ratio (UPCR) was calculated from a 24-hour urine collection, first morning void, or random spot urine using methods previously described.⁷ Minimal change disease activity was defined with the following criteria. Relapse was defined as UPCR ≥3 g/g on a spot collection, ≥3 g/d protein by timed collection, urine dipstick protein >3+, or study team reported relapse event by Case Report Form (CRF). Complete remission was defined as normal urine protein, including urinalysis dipstick protein negative or trace, UPCR <0.3 g/g, <300 mg protein/d, or reported remission on the CRF. Relapse and remission were considered recurring events, with the first designation (post-enrollment) determined based on above definitions, and subsequent relapse and remission were defined as occurring after complete remission or relapse, respectively. To be considered at risk for subsequent relapse (remission), participants must have returned to remission (relapse) based on the above definition. As data were unavailable from the time of disease onset or biopsy, only post-enrollment events are modeled, with event-specific follow-up time resetting to zero when the participant becomes at risk for the next event, and overall follow-up time censored at the last available study visit with available disease activity data, End Stage Kidney Disease (ESKD) or death.

eGFR was calculated using Chronic Kidney Disease Epidemiology Collaboration formula for participants aged ≥26 years,³⁰ and the bedside Schwartz formula for participants age <18 years.³¹ For ages 18 to 26, eGFR values from both equations were averaged.³² To account for decreased precision of estimating formulas at low serum creatinine values and reduce any undue influence of high eGFR values that may reflect transient episodes of hyperfiltration, eGFR was winsorized at 120 mL/min per 1.73m^2, in multivariable analyses.³³ Kidney failure (after enrollment) was defined by transplantation, chronic dialysis, or 2 or more consecutive eGFR measurements < 15 separated by 1 to 6 months. Kidney failure and death were not considered competing events given the low incidence post-enrollment in this cohort.

Exposures

The exposure used for the primary analysis of the MCD cohort was age at disease onset. Childhood-onset was defined as onset at 12 years of age or before, adolescent-onset was defined as onset between 13-17 years of age, and adult-onset was defined as onset at ≥18 years of age. Trained study coordinators enter in date of disease onset based on patient response and chart review (for proteinuria onset, nephrotic syndrome onset or physician diagnosis). Disease onset is defined as “Date of onset of first signs/symptoms (e.g., edema, gross hematuriaetc.)”.

For the rituximab-naive cohort, the primary exposure was treatment response classification in the MCD cohort. At enrollment, local investigators classified participants with MCD as Infrequently Relapsing, Frequently Relapsing/Glucocorticoid-Dependent, Glucocorticoid-Resistant, Multi-Drug Resistant, or Untreated/Unknown (Table S1). Glucocorticoid-resistant and multi-drug resistant were combined and referred to as “steroid-resistant” in multivariable analyses. Many participants did not have disease classification at enrollment (41% adolescents and 48% adults were unknown or untreated). Adjudication of disease classification at the time of rituximab initiation for the rituximab-naïve cohort was completed by study authors (D.P.C., D.S.G., K.T., T.S.) using the previous definitions (Table S1) as this analysis was dependent on the disease classification.

Statistical Methods

Continuous measures are described as mean and standard deviation (SD) or median and interquartile range (IQR), and categorical data are reported as frequencies and percentages. Differences among the MCD onset age groups were assessed using Wilcoxon rank sum tests for continuous variables and chi square or Fisher exact tests for categorical data. Kaplan–Meier curves were used to descriptively display the probability of relapse and remission for each event from the start of the most recent at-risk period and censored at the last study visit with available data, separately for children, adolescents, and adults.

Prentice, Williams, and Peterson (PWP) gap time recurrent event models were fitted to assess the probability of relapse and remission. These models were stratified by event (relapse or remission) number, allowing for a comparison between probabilities of each relapse or remission event between children, adolescents, and adults through testing of interactions between covariates and event number. Due to infrequent occurrence, events beyond the third relapse or remission were collapsed. A mixed effects linear regression model with random intercepts and slopes was used to assess differences in the change in eGFR over time between groups. Models were adjusted for sex, race/ethnicity, steroid use prior to biopsy, and years between disease onset and biopsy.

In a sub-analysis, a Cox proportional hazards model was fitted to assess differences by disease classification (glucocorticoid resistant vs. glucocorticoid sensitive) and rituximab response in the rituximab-naive cohort. Response was defined as time to first complete remission after initiation of rituximab; models were adjusted for MCD onset age groups, sex, race/ethnicity, glucocorticoid use prior to kidney biopsy, and urine protein/creatinine (UPCR) value at the time of rituximab initiation.

Results

Participant Characteristics of CureGN Minimal Change Disease Cohort

There were 554 subjects with biopsy-proven MCD in the MCD cohort: 304 childhood-onset disease, 49 adolescent-onset disease, and 201 adult-onset disease (Table 1). Median ages at disease onset were 4, 15, and 43 years for childhood-, adolescent-, and adult-onset disease, respectively. Median follow-up time was longer in the childhood-onset group (3.2 years) than the adolescent-onset and adult-onset groups (2.7 years for both, p=0.008). There were more females in the adult group (60%) than the childhood- and adolescent-onset groups (39% and 43%, respectively, p<0.001). There were no differences in ethnicity distribution, with the majority identifying as White (range 63%-69%, p=0.27). Participants with adult-onset disease were more likely to have private insurance (71%) than childhood- or adolescent-onset disease (38% and 49%, respectively, p<0.001).

Table 1:

Baseline characteristics of 554 participants with minimal change disease, including those with childhood-onset disease (age ≤12), adolescent-onset disease (age 13-17) and adult-onset disease (age ≥18 years).

	Child at Disease Onset (n=304)	Adolescent at Disease Onset (n=49)	Adult at Disease Onset (n=201)	P-value
Female, n(%)	118 (39%)	21 (43%)	120 (60%)	<0.001
Age at Disease Onset (years)	4.0 (2.0-6.0)	15.0 (14.0-16.0)	43.0 (28.0-57.0)	<0.001
Age at Biopsy (years)	5.0 (3.0-9.0)	15.0 (14.0-16.0)	44.0 (30.0-58.0)	<0.001
Incident Cohort, n(%)	101 (33%)	21 (43%)	65 (32%)	0.4
Race, n(%)a				0.40
White/Caucasian	193 (63%)	30 (61%)	138 (69%)
Black/African American	58 (19%)	12 (24%)	27 (13%)
Asian	21 (7%)	5 (10%)	23 (11%)
Other	20 (7%)	1 (2%)	5 (2%)
Ethnicity: Hispanic/Latino n(%)	38 (13%)	4 (8%)	19 (10%)	0.4
Insurance, n(%)				<0.001
Private Insurance	117 (38%)	24 (49%)	143 (71%)
Public Insurance (US)	120 (39%)	14 (29%)	30 (15%)
Public Insurance (non-US)	37 (12%)	7 (14%)	16 (8%)
Other Insurance	15 (5%)	1 (2%)	7 (3%)
No Insurance	3 (1%)	1 (2%)	3 (1%)
Overweight/Obese, n (%)	166 (55%)	23 (47)%	133 (66%)	0.02
Country, n (%)				0.002
United States	264 (87%)	41 (84%)	173 (86%)
Canada	15 (5%)	4 (8%)	20 (10%)
Italy	21 (7%)	4 (8%)	1 (1%)
Poland	4 (1%)	0 (0%)	7 (3%)
Median Follow up time (years, IQR)	3.2 (2.1,4.4)	2.7 (1.8,4.1)	2.7 (1.6,3.9)	0.008

^aRace Unknown, total n = 21

N(%) or mean (standard deviation) or median (IQR).

Differences at time of biopsy between groups

Date of disease onset was based on self-report by participants and confirmed by coordinators and/or physician with medical chart review (when available). Children had longer time to biopsy after disease onset (238 days) than those with adolescent-onset (23 days) or adult-onset disease (36 days, p<0.001, Table 2). UPCR at biopsy increased with age at onset (3.0 g/g vs. 5.6 g/g vs. 6.9 g/g) for childhood-, adolescent- and adult- onset disease, respectively (p<0.001), and eGFR decreased with age at onset (139 vs. 95 vs. 88 mL/min/1.73m², respectively, p<0.001). Children had higher serum albumin at biopsy than those with adolescent- and adult-onset disease (2.6 vs. 1.9 and 2.2 g/dL, respectively, p=0.002). Hypertension was more common among adults (47%) than children (33%) or adolescents (14%, p<0.002).

Table 2.

Clinical characteristics and therapy

	Child at Disease Onset (n=304)	Adolescent at Disease Onset (n=49)	Adult at Disease Onset (n=201)	P-value
Days between onset and biopsy	238 (85,663)	23 (11,62)	36 (16,93)	<0.001
UPCR at Biopsy *	3.0 (0.4-8.9)	5.6 (1.0-8.3)	6.9 (3.6-10.0)	<0.001
UPCR at Disease Onset *	7.5 (2.3-12.2)	4.6 (1.0-10.7)	6.5 (2.8-10.0)	0.2
eGFR (mL/min/1.73m^2) at Biopsy**	139 (110-173)	95 (78-114)	88 (63-108)	<0.001
eGFR (mL/min/1.73m^2) at Disease onset**	127 (101-165)	100 (67-114)	89 (64-109)	<0.001
Serum albumin (g/dL) at Biopsy***	2.6 (1.8-3.3)	1.9 (1.6-2.4)	2.2 (1.7-2.8)	0.002
Serum albumin (g/dL) at Disease Onset***	1.8 (1.3-2.4)	1.9 (1.7-2.3)	2.2 (1.7-2.8)	0.002
Hypertension, n(%)	99 (33%)	7 (14%)	94 (47%)	0.001
Steroid Response Classification at Enrollment				<0.001
Infrequently Relapsing	24 (8%)	6 (12%)	30 (15%)
FRNS/SDNS	173 (57%)	9 (18%)	56 (29%)
Steroid Resistant	50 (16%)	11 (22%)	22 (11%)
Multi-drug resistant	6 (2%)	0 (0%)	8 (4%)
Untreated	14 (5%)	6 (12%)	15 (8%)
Unknown	38 (12%)	18 (36%)	65 (33%)
Immunosuppression prior to biopsy
Any Immunosuppression	245 (81%)	16 (33%)	51 (26%)	<0.001
Corticosteroids	231 (76%)	16 (33%)	45 (23%)	<0.001
CNIa	44 (15%)	1 (2%)	4 (2%)	<0.001
Rituximaba	8 (3%)	0 (0%)	5 (5%)	0.5
Mycophenolatea	20 (7%)	1 (2%)	2 (1%)	0.007
RAAS Blockade prior to biopsy	23 (8%)	4 (8%)	39 (20%)	<0.001

^*missing at biopsy: 115 (38%) pediatric, 13 (26%) adolescent, 69 (36%) adult, missing at disease onset: 193 (66%) pediatric 28 (56%) adolescent and 87 (46%) adult-onset participants

^**missing at biopsy: 74 (24%) pediatric, 6 (12%) adolescent, 50 (26%) adult, missing at disease onset: 186 (61%) pediatric, 23 (46%) Adolescent-, 89 (47%) adult-onset participants

^***missing at biopsy: 90 (30%) pediatric, 8 (16%) adolescent, 3 (31%) adult, missing at disease onset: 189 (62%) pediatric 24 (48%), 87 (46%) adult-onset participants

FRNS/SDNS – frequently relapsing nephrotic syndrome/steroid dependent nephrotic syndrome

^aFisher’s exact test.

The most common disease classification at enrollment in the childhood-onset group was frequent relapsing/glucocorticoid-dependent (57%); among adolescents and adults, the most common classification was “Untreated/Unknown” (36% and 33%, respectively). Children were more likely to receive immunosuppressive medication pre- biopsy compared to adolescents and adults (81% vs. 33% and 26%, respectively, p<0.001). In children, pre-biopsy therapy was mainly prednisone (76%), but calcineurin inhibitors were also more commonly used than in adolescents or adults (15% vs. 2% and 2%, respectively, p<0.001) (Table 2). Prior to biopsy, 20% of adults were treated with renin angiotensin-aldosterone system (RAAS) blockade, compared to 8% of adolescents and children (p<0.001). Since the details of many patients at disease onset were unknown, we started clinical analysis at time of enrollment when most data were available. After enrollment, we had evidence to determine relapse or remission (corroborated by proteinuria or medication changes) in 93% of relapse events and 95% of remission events.

At enrollment, more children were relapsing (36% vs. 20% for the adolescent and adult groups, p<0.001, Table S2). Overall, children had higher remission and relapse rates (61.2 remission events/100-person-years, 47.1 relapse events/100-person-years) than adolescent-onset (56.0 remission events/100-person-years and 33.3 relapse events/100-person-years, p<0.001) and adult-onset (41.7 remission events/ 100-person-years, 22.7 relapse events/100-person-years) participants. Children remitted faster (median 53 days) than adolescents (62 days) and adults (73 days), p=0.003. Children also relapsed faster (159 days) than adolescents (171 days) and adults (246 days), p = 0.008.

Probability of Relapse and Remission

To understand the nature of this relapsing/remitting disease and to compare the groups, we used post-enrollment data as it is most robust. Event-specific follow-up time was reset to zero when the participant was at risk for the next event. Of those in remission at enrollment, 50% relapsed by 1.3 years (children) and 2.7 years (adolescents). Only 45% of adults ever relapsed during a median observation period of 2.7 years (range 1.6 to 3.9 years). The probability of first relapse at one year was highest in children and lower in both adolescent- and adult-onset disease (44%, 30%, and 25%, respectively, p<0.001, Figure 2A). After remission, the probability of a second relapse at one year was higher among childhood-onset and adolescent-onset disease, and slightly lower in adult-onset disease (52%, 50%, and 40%, respectively, p=0.08, Figure 2B). After three or more remission occurrences, at one year there was a 67%, 37%, and 26% probability of relapse in childhood-, adolescent-, and adult-onset disease, respectively (p<0.001, Figure 2C). Among those who reached the third or subsequent remission, half relapsed again by 0.6, 1.7, and 2.5 years, respectively. Median time to relapse decreased from 1.9 to 1.0 to 0.7 years for the first, second, and third or more relapse event, respectively (p<0.001). Participants were only included in the middle and right panels if they experienced at least one or two events, respectively, as shown by the number at risk in each group along the x-axis.After adjustment for sex, race/ethnicity, glucocorticoid use prior to biopsy, and years from onset to enrollment, those with childhood-onset disease had 69% (hazard ratio [HR]=1.69, 95% CI=1.29-2.21) higher risk of relapse compared to those with adult-onset disease. Risk of relapse was similar between adolescent-onset and adult-onset disease age groups (Table S3a). There was no significant interaction between age at onset and relapse number (first, second, etc.) (p=0.62). The same analysis was done on a subset of patients (n=166) with incident disease (defined as patients with biopsy within 6 months of enrollment). These results (non-significant) show similar direction and magnitude of effect within patients with new or recent onset disease (Table S3b).

Figure 2:

Among CureGN participants, time from complete remission to first initial relapse and subsequent relapses (index date is first remission after enrollment): A) Time to first initial relapse comparing childhood, adolescent and adult-onset disease, B) Time to subsequent second relapse comparing childhood-onset, adolescent-onset, and adult onset disease. C) Time to third or more subsequent relapses. Gray horizontal line shows the number of years from remission when 50% of the population experienced relapse.

Among those in relapse at enrollment, the six-month probability of remission was highest in the childhood-onset (63%) compared to the adolescent-onset (52%) and the adult-onset group (54%, p=0.009, Figure 3A). Subsequently, after the second relapse, the probability of remission by 6 months was 80% for childhood-onset compared to 69% for adolescent-onset, and 61% for adult-onset disease (p=0.003, Figure 3B). The probability of a third or later remission event was similar across the three groups (range 64% to 83%, p=0.17, Figure 3C). After adjustment for sex, race, ethnicity, glucocorticoid use prior to biopsy, and years from onset to enrollment, participants with childhood-onset disease had a 1.33-fold greater chance of achieving remission compared to those with adult-onset disease (HR=1.33, 95% CI 1.02-1.73, p=0.03), while there was no difference between adolescent- versus adult-onset disease (p=0.40) (Table S4a). We performed additional analyses limited to patients classified as steroid sensitive or non-steroid sensitive (excluding unknown or untreated patients) and found similar risks of relapse in children compared to adults or adolescents. An analysis of incident patients (n=166) as above found similar results (non-significant) shown in Table S4b.

Figure 3:

Time from relapse to complete remission is faster in childhood-onset disease compared to adult-onset and adolescent-onset for initial and subsequent second remission and similar for all age groups for subsequent remissions (index date is first relapse after enrollment). A) Time to initial complete remission comparing childhood-, adolescent-, and adult-onset disease. B) Time to subseqent second remission comparing childhood-onset, adolescent-onset and adult-onset disease. C) Time to third or subsequent remissions. Gray horizontal line shows the number of years from relapse when 50% of the population achieved remission.

Kidney Function Outcomes

There were no substantial differences in the frequency of kidney failure (4% in childhood- vs. 6% in adolescent- and 3% in adult-onset disease) or rate of kidney failure (1.40, 2.22 and 1.34 events per 100 patient years for childhood, adolescent, and adult onset respectively) and there were very few death events (1 in childhood onset disease and 5 in adult onset disease) (Table S5). Children and adolescents had a higher eGFR at the time of biopsy (139 and 95, respectively) than the 88mL/min/1.73m² in adults, p<0.001. Unadjusted mean slopes of eGFR were −0.27 (p=0.38), 0.02 (p=0.93), and −1.11 (p=0.001) mL/min/1.73m² per year for those with childhood-onset, adolescent-onset, and adult-onset disease, respectively. Differences in the rate of eGFR change did not differ statistically between age-at-onset groups before and after adjustment (p=0.172 and p=0.128, respectively) and are shown in Figure 4. After adjustment for sex, race/ethnicity, and pre-biopsy immunosuppression use, these estimates of average yearly rate of change remained consistent, and the overall average rate of change in eGFR was −0.56 mL/min/1.73m² (p=0.03, Table S6).

Figure 4:

The means and 95% confidence intervals of the eGFR predicted slopes over follow-up duration in childhood-onset, adolescent-onset and adult-onset MCD are shown (index date is date of disease onset). The intercepts differ substantially, but the slopes are not significantly different. Unadjusted slope (Std Err) - childhood-onset: −0.27 (0.23), adolescent-onset: 0.02 (0.73), adult-onset: −1.11 (0.34), adjusted slope (Std Err) - childhood-onset −0.20 (0.23), adolescent-onset: 0.06 (0.74), adult-onset: −1.13 (0.35). eGFR (mL/min/1.73 m²) was calculated using the Schwartz formula for Pediatric patients <18 years, the CKD-EPI formula for Adult patients >26 years, and the average of pediatric and adult formulas for Adolescent patients 18 to 26 years. Post-ESKD values were excluded.

Rituximab Response

Most participants did not receive rituximab prior to study entry (n=462, 83%). Among those, 111 participants [59 childhood-onset (25%), 8 adolescent-onset (19%), 44 adult-onset (24%)] received rituximab during follow-up and this cohort was analyzed to understand the rituximab response. The cohort included in the rituximab analysis is described in Table S7. Among the childhood onset group, only 7% were Hispanic/Latino compared to 13% in the broader cohort. Overall, there was a higher proportion of infrequently relapsing and steroid resistant patients in this group. At the time of rituximab initiation, most patients were classified as frequently relapsing or steroid-dependent (n=46, 41%); the rest were classified as infrequently relapsing (n=27, 24%) or glucocorticoid-resistant (n=28, 25%). Ten participants could not be classified with available data and were excluded from analysis. Subjects were further grouped as glucocorticoid-sensitive if they were frequently relapsing, glucocorticoid-dependent, or infrequently relapsing, vs. glucocorticoid-resistant or multidrug resistant. Remission probability one year after rituximab was higher in the glucocorticoid-sensitive group compared to those who were resistant, (83% vs. 31%, p=<0.001, Figure 5). Glucocorticoid-sensitive patients were more likely to enter complete remission after rituximab initiation (adjusted HR 2.62, p=0.0015) than those who were not (Table S8). Those with childhood-onset disease were more likely to enter remission than either adolescents or adults after rituximab initiation (adjusted HR 2.1, p = 0.0028). The median time of follow-up in all groups prior to rituximab was less than 1 year and therefore we were not powered to compare before and after rituximab therapy. We analyzed whether groups were different in terms of steroid response classification and found no major differences.

Figure 5.

Time from rituximab initiation to complete remission is faster among those who were glucocorticoid-sensitive compared to those who were not glucocorticoid sensitive (index date is date of rituximab therapy).

Discussion

CureGN provides an opportunity to investigate differences in the clinical course of childhood-onset, adolescent-onset, and adult-onset biopsy-proven MCD. Our focus was not to compare clinically diagnosed and treated nephrotic syndrome but rather a group with MCD confirmed by biopsy. There are limited comparably large epidemiological studies in nephrotic syndrome. Since childhood onset nephrotic syndrome is often treated clinically rather than based on pathologic diagnosis, biopsied pediatric patients may represent a complex and distinct group. We reviewed prior studies of pediatric steroid sensitive nephrotic syndrome and noted they have similar make up in terms of sex and age at disease onset though the proportion of frequently relapsing disease was slightly higher in our biopsied cohort.³⁴ Despite similarities, it should be noted that this study highlights findings from a cohort of biopsied patients and may not represent the entire spectrum of minimal change disease.

MCD is a chronic relapsing and remitting disease but predicting either of these events presents a clinical challenge. We found that those with childhood-onset disease had more remission and relapse events overall compared to adults and adolescents. Risk of relapse was higher, and time to relapse shorter, in children than adults or adolescents. This suggests that children with biopsy proven MCD have a distinct course with regard to relapse. Those with childhood-onset disease achieved remission at a higher rate by 6 months. This may reflect the differences in therapies used in children vs. adults or adolescents or may be related to divergent mechanisms of disease. Additional research will be needed to clarify.

We also evaluated differences in longitudinal changes in kidney function based on age of onset. The slope of eGFR over time was stable in all groups, with no significant differences. We analyzed by the number of immunosuppressive medications used and found no differences, though additional studies could analyze the duration of specific medications. Finally, ESKD and all-cause mortality across all ages was low (~5%).

Rituximab is increasingly used as a glucocorticoid-sparing therapy. There is evidence that B cell targeted therapy is effective in a subset of patients with glucocorticoid-sensitive disease.^35-38 T cells have long been implicated in the immunopathogenesis of MCD, and the effect of anti-CD20 therapy and T cells is still being studied.³⁹ Overall, data are limited on which patients benefit from rituximab therapy. We explored the role of clinical phenotype and rituximab response in a subgroup of patients from our larger cohort. We found that 83% of those who are frequently relapsing, glucocorticoid-dependent, or infrequently relapsing had a robust response to rituximab therapy, compared to only 31% of those who were classified as glucocorticoid-resistant. Those with childhood-onset disease and glucocorticoid-sensitivity were twice as likely to enter remission after rituximab therapy. Even though we lack validated, clinically available biomarkers to select for rituximab responsiveness, the clinical phenotype based on response or resistance to other immunosuppression can guide clinical decisions on using rituximab, especially for patients who would most benefit from glucocorticoid-sparing agents. Additional research is needed to evaluate long term safety and prognostic biomarkers.

Even in such a large cohort of biopsy proven disease, certain considerations should be mentioned. It is not surprising that those with childhood-onset disease had longer time to biopsy (238 days vs. 36 days in adult-onset disease), as most cases of childhood onset nephrotic syndrome are treated presumptively as MCD and only biopsied for complications or poor treatment response. Given the selection bias inherently present in the CureGN cohort, our results should be extrapolated with caution to children with glucocorticoid-sensitive idiopathic nephrotic syndrome who never undergo kidney biopsy. Within our cohort, sensitivity analyses in steroid sensitive vs. non-steroid sensitive groups showed similar results. Clinical characteristics varied based on age at onset and may be confounded by differences in pre-biopsy treatment approaches across the groups. As CureGN is a multi-center study, we acknowledge that treatment approaches are not standardized across centers. Adults and adolescents had more proteinuria at time of biopsy, which may reflect pre-biopsy treatment effects in children. There were higher rates of hypertension and lower eGFR at disease onset in the adult-onset group, possibly due to age-related changes, e.g. arterionephrosclerosis rather than disease specific differences. While childhood-onset disease was more frequently treated with glucocorticoids pre-biopsy, they were less likely to be treated with RAAS blockade. This may reflect clinical practice patterns, prevalence of hypertension, or assessment of risk benefit considerations related to acute kidney injury and may warrant further investigation.

In addition to the limitations of a biopsied cohort, our study focuses on patients who either lived near or were referred to a glomerular disease center and subsequently enrolled in CureGN. We found that among children and adolescents, 91% obtained local care (their kidney biopsy was performed at the enrolling CureGN site), but non-local adult patients were commonly referred to a glomerular disease center. The mixture of local and referred adults required further investigation. We performed statistical tests (Table S9) to understand how adult participants differed in being “local” (first biopsy at enrolling CureGN site) or “referred” (biopsy outside of the enrolling CureGN site). Those more likely to be referred included females, older age at disease onset, prevalent patients (first biopsy >6 months), non-Hispanics, those overweight or obese, and those with a history of hypertension. Compared to local patients, the referred patients contributed diversity in sex, age, race, and comorbidities. The combination of local and referred patients increases the generalizability of the broad spectrum of MCD.

The design of CureGN allowed for recruiting both incident and prevalent cases, and despite efforts to retrospectively obtain clinical data from the time of biopsy and disease onset, there were missing data in all groups. As expected, data after enrollment was more robust. Data missingness at time of biopsy or enrollment is reported in the tables 1 and 2. The adolescent group was relatively small compared to the other groups. Self-reported race and ethnicity used in models has limitations as our analyses do not fully explore the social implications of race in MCD.

To our knowledge, our study represents the largest cohort of MCD across the lifespan, allowing comparisons among childhood-onset, adolescent-onset, and adult-onset MCD in a well-defined study such as CureGN. The clinical trajectory based on age groups was previously only compared in smaller studies.^12,40 We show that childhood-onset MCD is distinct in its course of relapse and remission and established therapeutic approaches. Understanding the similarities and differences is key to development of translational and mechanistic studies to elucidate common and distinct underlying pathophysiology so that targeted therapies can be developed. Increasingly there are studies identifying molecular markers such as antibodies, B cells and urinary markers which may be important in understanding the risk of relapse and remission.^41-43 Future investigations correlating pathologic findings and molecular markers can shed light on the phenotypic differences reported here.

Supplementary Material

1

Table S1. Disease classification for the joint processes of glucocorticoid treatment and the frequency and timing of relapse or remission.

Table S2. Summary table of relapse and remission events for childhood-, adolescent- and adult-onset glomerular disease.

Table S3a. Recurrent event model of time from remission to subsequent relapse in children vs. adults and in adolescents vs. adults.

Table S3b. Analysis limited to subset of “incident” patients.

Table S4a. Recurrent event model of time from relapse to subsequent remission in children vs. adults and in adolescents vs. adults.

Table S4b. Analysis limited to subset of “incident” patients.

Table S5. Mean number of years from enrollment to End Stage Kidney Disease (ESKD) or death; number (%) of ESKD events after enrollment based on age at disease onset.

Table S6. Mixed effects linear regression model with covariate predictors of estimated glomerular filtration rate (eGFR).

Table S7. Patient characteristics of subset of rituximab naive cohort.

Table S8. Time to first remission after rituximab initiation among rituximab naive patients at enrollment.

Table S9. Table of differences between ‘local’ and ‘referred’ CureGN adult participants.

Members of the CureGN Consortium:

At the CureGN Participating Clinical Centers (PCC) network through Columbia University: Wooin Ahn, Gerald Appel, Paul Appelbaum, Revekka Babayev, Andrew Bomback, Brenda Chan, Vivette Denise D'Agati, Samitri Dogra, Hilda Fernandez, Ali Gharavi**, William Hines, Syed Ali Husain, Namrata Jain, Krzysztof Kiryluk, Fangming Lin, Maddalena Marasa^#, Glen Markowitz, Hila Milo Rasouly, Sumit Mohan, Nicola Mongera, Jordan Nestor, Thomas Nickolas, Jai Radhakrishnan, Maya Rao, Simone Sanna-Cherchi, Shayan Shirazian, Michael Barry Stokes, Natalie Uy, Anthony Valeri, Natalie Vena (Columbia University, New York, NY); Bartosz Foroncewicz, Barbara Moszczuk, Agnieszka Perkowska-Ptasińska (University of Warsaw, Warszawa, Poland); Gian Marco Ghiggeri*, Francesca Lugani (Gaslini Children’s Hospital, Genoa, Italy). At the CureGN PCC network through the Pediatric Nephrology Research Consortium: Josephine Ambruzs, Helen Liapis (Arkana Laboratories, Little Rock, AR); Rossana Baracco, Amrish Jain* (Children’s Hospital of Michigan, Detroit, MI); Isa Ashoor, Diego Aviles* (Children’s Hospital of New Orleans/LSU Health, New Orleans, LA); Sun-Young Ahn* (Children’s National Medical Center, Washington DC); Prasad Devarajan, Elif Erkan*, Donna Claes, Hillarey Stone (Cincinnati Children’s Hospital Cincinnati, OH); Sherene Mason* (Connecticut Children’s Medical Center, Hartford, CT); Liliana Gomez-Mendez* (East Carolina University Brody School of Medicine, Greenville, NC); Chia-shi Wang, Hong (Julie) Yin (Emory University, Atlanta, GA); Yi Cai*, Goebel Jens, Julia Steinke (Helen DeVos Children’s Hospital, Grand Rapids, MI); Donald Weaver* (Levine Children’s Hospital/Atrium Health, Charlotte, NC); Jerome Lane* (Lurie Children’s Hospital, Chicago IL); Carl Cramer* (Mayo Clinic, Rochester, MN); Cindy Pan, Neil Paloian, Rajasree Sreedharan* (Medical College of Wisconsin, Milwaukee, WI); Corinna Bowers^#, Mary Dreher^#; John Mahan, Samantha Sharpe^#, William Smoyer** (Nationwide Children’s Hospital, Columbus, OH); Amira Al-Uzri*, Sandra Iragorri (Oregon Health and Science University, Portland, OR); Myda Khalid* (Riley Children’s Hospital, Indianapolis, IN); Craig Belsha* (Cardinal Glennon Children’s Medical Center/St. Louis University, St. Louis, MO); Joseph Alge*, Michael Braun, AC Gomez, Scott Wenderfer* (Texas Children’s Hospital, Houston, TX); Tetyana Vasylyeva* (Texas Tech Health Sciences Center, Amarillo, TX); Daniel Feig* (Children’s of Alabama, University of Alabama, Birmingham, AL); Gabriel Cara Fuentes, Melisha Hannah* (University of Colorado Children’s Hospital, Colorado, Aurora, CO); Carla Nester* (University of Iowa Children’s Hospital, Iowa City, IA); Jon Klein** (University of Louisville, Louisville, KY); Chryso Katsoufis, Wacharee Seeherunvong* (Holtz Medical Center, University of Miami, Miami, FL); Michelle Rheault* (University of Minnesota Children’s Hospital, Minneapolis, MN); Craig Wong* (University of New Mexico Health Sciences Center, Albuquerque, NM); Nisha Mathews* (University of Oklahoma Health Sciences Center, Oklahoma City, OK); John Barcia*, Agnes Swiatecka-Urban (University of Virginia, Charlottesville, VA); Sharon Bartosh* (University of Wisconsin, Madison, WI); Vikas Dharnidharka*, Joseph Gaut (Washington University in St. Louis, St. Louis, MO). At the CureGN PCC network through the University of North Carolina: Louis-Philippe Laurin*, Virginie Royal (Hôpital Maisonneuve-Rosemont, Montreal, Canada); Anand Achanti, Milos Budisavljevic*, Sally Self (Medical University of South Carolina, Charleston, SC); Cybele Ghossein, Yonatan Peleg, Shikha Wadhwani* (Northwestern University, Chicago, IL); Salem Almaani, Tibor Nadasdy, Samir, Parikh, Brad Rovin* (Ohio State University, Columbus, OH); Anthony Chang (University of Chicago, Chicago, IL); Huma Fatima, Bruce Julian, Jan Novak, Matthew Renfrow, Dana Rizk* (University of Alabama at Birmingham, Birmingham, AL); Vimal Derebail, Ronald Falk**, Keisha Gibson, Dorey Glenn, Susan Hogan, Koyal Jain, J. Charles Jennette, Caroline Poulton^#, Manish Kanti Saha (University of North Carolina Kidney Center, Chapel Hill, NC); Agnes Fogo, Neil Sanghani* (Vanderbilt University, Nashville, TN); Selvaraj Muthusamy (Virginia Commonwealth University, Richmond, VA). At the CureGN PCC network through the University of Pennsylvania: Jeffrey Schelling* (MetroHealth Medical Center/Case Western Reserve University, Cleveland, OH); Jean Hou (Cedars-Sinai Health System, Los Angeles, CA); Kevin Lemley*, Warren Mika, Pierre Russo (Children’s Hospital of LA, Los Angeles, LA); Michelle Denburg, Amy Kogon, Kevin Meyers*, Madhura Pradhan (Children’s Hospital of Philadelphia, Philadelphia, PA); Raed Bou Matar*, John O'Toole*, John Sedor* (Cleveland Clinic, Cleveland, OH); Christine Sethna*, Suzanne Vento^# (Cohen Children’s Medical Center, New Hyde Park, NY); Mohamed Atta, Serena Bagnasco, Alicia Neu, (Johns Hopkins University, Baltimore, MD); Sharon Adler*, Tiane Dai, Ram Dukkipati (Lundquist Institute at Harbor-UCLA Medical Center, Torrance, CA); Fernando Fervenza*, Sanjeev Sethi (Mayo Clinic, Rochester, MN); Frederick Kaskel, Kaye Brathwaite, (Montefiore Medical Center, The Bronx, New York, NY); Joseph Weisstuch, Ming Wu, Olga Zhdanova (New York University, New York, NY); Jurgen Heymann, Meryl Waldman, Cheryl Winkler (NIDDK, Bethesda, MD); Katherine Tuttle* (Spokane Providence Medical Center, Spokane, WA); Jill Krissberg, Richard Lafayette*, Kamal Fahmeedah, Elizabeth Talley (Stanford University, Palo Alto, CA); Michelle Hladunewich* (Sunnybrook Health Sciences Centre, Toronto, Canada); Carmen Avila-Casado, Daniel Cattran*, Reich Heather, Philip Boll (University Health Network, Toronto, Canada); Yelena Drexler, Alessia Fornoni* (University of Miami, Miami, FL); Patrick Gipson*, Jeffrey Hodgin, Andrea Oliverio (University of Michigan, Ann Arbor, MI); Jon Hogan, Lawrence Holzman**, Matthew Palmer, Gaia Coppock (University of Pennsylvania, Philadelphia, PA); Blaise Abromovitz*, Michael Mortiz* (University of Pittsburgh School of Medicine, Pittsburgh, PA); Charles Alpers, J. Ashley Jefferson* (University of Washington, Seattle, WA); Elizabeth Brown, Kamal Sambandam*, Bethany Roehm (UT Southwestern, Dallas, TX). At the Data Coordinating Center: Bruce Robinson** (Arbor Research Collaborative for Health, Ann Arbor, MI); Cynthia Nast (Cedar Sinai Medical Center, Los Angeles, CA); Laura Barisoni (Duke University, Durham, NC); Matthias Kretzler, Laura Mariani** (University of Michigan, Ann Arbor, MI). Steering Committee Chair: Lisa M. Guay-Woodford, Children’s National Hospital, Washington DC. **CureGN Principal Investigators; *CureGN Site Principal Investigators; ^#CureGN Lead Coordinators.