Systematic Review of the Oxford Classification of IgA Nephropathy: Reproducibility and Prognostic Value

Abstract

Key Points

• The Oxford classification of IgA nephropathy defined five features scored subjectively in renal biopsies, identified by the initials MESTC.

• Two large studies with independent observers showed reproducibility was moderate for T, moderate or poor for M and S, and poor for E and C.

• In multivariate analyses including clinical features, T was related to 58% of outcomes, with no correlation of MESTC with 24% of outcomes.

Background

The Oxford classification of IgA nephropathy defined five prognostic features scored subjectively in renal biopsies: mesangial cellularity (M), endocapillary hypercellularity (E), segmental sclerosis (S), interstitial fibrosis/tubular atrophy (T), and (fibro)cellular crescents (C). Pathological scoring systems should be reproducible and have prognostic value independently of clinical features. Reproducibility of the classification was not previously investigated in a systematic review, and the most recent systematic reviews of prognostic value were in 2017.

Methods

This systematic review followed PRISMA 2020 guidelines. MEDLINE, PUBMED, and EMBASE databases were searched using the terms “IgA nephropathy” and “Oxford.” Eligible papers applied the classification and mentioned statistical analysis of interobserver reproducibility and/or included multivariate analysis of outcomes related to individual Oxford scores and clinical features, including treatment with corticosteroids or other immunosuppressive drugs.

Results

There were 99 suitable papers before September 23, 2022. Of 12 papers that mentioned reproducibility, only six reported statistics for MEST/MESTC scoring. Four of these were small studies and/or had observers at the same institution. These were considered less representative of application of the classification than two large studies with independent observers, in which agreement was moderate for T, either moderate or poor for M and S, and poor for E and C. In 92 papers with 125 multivariate analyses of various outcomes, the commonest Oxford element associated with outcomes was T (73 of 125, 58%), with no correlation of any element with outcomes in 30 analyses (24%). Treatment with immunosuppression was often related to scores, particularly C and E, without consistent relations between Oxford scores and outcomes in immunosuppressed patients.

Conclusions

This systematic review showed limitations of the Oxford classification in practice, particularly the moderate or poor reproducibility of scores. T was the Oxford score most often related to clinical outcomes, but even this was not consistently reliable as a prognostic indicator.

Introduction

In 2009, the Oxford classification of IgA nephropathy was published, as an internationally agreed, evidence-based, pathological classification, of help to clinicians.^1,2 This required subjective assessments of four renal biopsy features, mesangial cellularity (M), endocapillary hypercellularity (E), segmental sclerosis (S), and interstitial fibrosis/tubular atrophy (T). Another feature, glomerular crescents (C), was added in 2017.³ In the 2009 study, crescents were uncommon, not associated with outcome, and not initially included.^1,2 The classification was only intended to identify prognostic features, although it was recognized that such a classification may also in the future facilitate the identification of specific features that may predict response to immunosuppression or other specific treatments and refine recruitment to clinical trials by their capacity to stratify patients by their risk of progression.¹

The classification followed retrospective study of 265 patients. Renal biopsies were examined by pathologists using detailed score sheets.² Features finally included were derived from those that were reproducible after repeated comparisons, had least sampling error, were easiest to score, and were independently correlated with outcome. Agreement between the original pathologists for assessment of MEST is impossible to know because they did not specifically report MEST scores. Similarly, reproducibility of C scoring was not formally tested before this was introduced.^3,4

In brief, the scores are as follows^1,3: M1 means that more than half of glomeruli have more than three cells in a mesangial area. Otherwise, the score is M0. E1 means an increased number of cells within glomerular capillary lumina (otherwise, E0). S1 means sclerosis not affecting the whole tuft or presence of an adhesion (otherwise, S0), although the updated classification recommends addition of with/without podocyte hypertrophy/tip lesions to S1.³ T1 means that an estimated 26%–50% of the cortex contains tubular atrophy or interstitial fibrosis, and T2 means that more than 50% is affected (otherwise, T0). Crescents have extracapillary proliferation of more than two cell layers, cellular lesions with more than 50% occupied by cells, and fibrocellular with no more than 50% cells and less than 90% matrix. Fibrous crescents, with at least 90% matrix, are ignored. C0 means no crescents, C1 crescents in under 25% of glomeruli, and C2 crescents in at least 25% of glomeruli.

There were clinical restrictions.^1,2 Initial eGFR was at least 30 ml/min per 1.73 m²; initial proteinuria was over 0.5 g per 24 hours in adults and at least 0.5 g per 24 hours per 1.73 m² in children younger than 18 years; follow-up was at least 12 months; and secondary causes, such as Henoch Schoenlein purpura, and comorbid conditions, such as diabetes mellitus, were excluded. Biopsies contained at least eight glomeruli. Patients previously treated with immunosuppressive drugs, including corticosteroids, were included. Many patients received immunosuppression after biopsy, without set guidance for treatment.

Outcome was assessed by three measures: rate of decline of eGFR; survival from either 50% reduction in eGFR or ESKD, defined as eGFR less than 15 ml/min per 1.73 m², together called the combined event; and amount of proteinuria during follow-up. In multivariate analyses of outcome, rate of eGFR decline and survival from the combined event were assessed in two models including pathological features, one with initial eGFR, mean arterial pressure, and proteinuria and one with initial eGFR and follow-up averages of mean arterial pressure and proteinuria. Clinical features at biopsy were known to have prognostic significance, especially renal function, amount of proteinuria, and hypertension.⁵

The simplicity of MESTC definitions allowed the Oxford classification to be used extensively. There were three previous systematic reviews of the classification, none of which considered reproducibility of scoring. One concluded that M, S, T, and C lesions, but not E lesions, are strongly associated with progression to kidney failure.⁶ Another concluded that crescent formation represents an efficient prognostic factor associated with progression to kidney failure.⁷ The third concluded that IgA nephropathy patients with serious pathological changes (M1, S1, and T1/2) were more resistant to steroid than slight ones (M0, S0, and T0), and E1 is [sic] better response to steroid therapy than T1/2.⁸ This third review assessed scores as predictors of response to immunosuppression, not just prognostic indicators. Two other systematic reviews considered the use of the classification in Henoch Schoenlein nephritis (IgA vasculitis with nephritis).^9,10 This was excluded from the present review because the classification was neither validated nor recommended in this condition.^3,11

The present systematic review, following Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA) 2020 guidelines,¹² investigates aspects relevant to clinical usefulness. Is the classification reproducible? Does it have prognostic value, independently of clinical features, including treatment with immunosuppression? Inhibition of the renin-angiotensin system was recommended by Kidney Disease Improving Global Outcomes¹¹ for treatment of proteinuria and/or hypertension in IgA nephropathy and was not analyzed.

Methods

MEDLINE, PUBMED, and EMBASE databases were searched on September 22, 2022, using the terms “IgA nephropathy” and “Oxford.” The flow diagram (Figure 1) shows how papers were selected after exclusions, such as duplicates, conference abstracts, papers in which “Oxford” did not refer to the Oxford classification, and publications, such as reviews, without original findings. Available papers were searched to see if they mentioned analysis of reproducibility between observers and whether they included multivariate analysis of outcome in which at least one individual Oxford element was included with at least one of renal function, proteinuria, or BP. Studies that grouped MEST or MESTC scores as totals in multivariate analysis were not considered because grouping obscured the value of individual elements.

Figure 1.

PRISMA 2020 flow diagram for systematic review of aspects of the Oxford classification of IgA nephropathy. *Reports without multivariate analysis of outcome that included clinical features and at least one Oxford feature were excluded, unless they mentioned analysis of reproducibility between observers (n=7).

Analysis of reproducibility was usually by assessment of agreement using intraclass correlation coefficients (ICCs).¹³ For these, expressed by the kappa statistic, no agreement is 0, and complete agreement is 1. Values of kappa indicating poor, intermediate, and good agreement are arbitrary.¹⁴ Publications from the Oxford group accepted that a kappa score under 0.40 is a poor interrater reliability, 0.40–0.59 is moderate, 0.60–0.79 is substantial, and 0.80 or over is almost perfect/outstanding.^2,15 Another method was Cronbach alpha reliability estimate, where scores of over 0.7 indicate a strong interobserver reliability.¹⁶

Papers reporting multivariate analysis of outcome were studied for selection of patients, outcome measures, and number of observers. Analysis was usually by Cox proportional hazards regression but occasionally by logistic regression.¹⁷ All papers used the conventional value of P < 0.05 to indicate statistical significance. These papers were examined for mention of immunosuppressive treatment, its relation to scores, and whether treatment was a variate in analysis. The review was not registered.

Results

There were 99 eligible papers, 12 mentioning reproducibility,^2,15,18–27 92 with suitable multivariate analysis of outcome, and five with both.^{19–21,24,27} The papers with multivariate analysis appear in Supplemental Table 1.

Reproducibility of Oxford Scoring

Table 1 gives details of papers that mentioned interobserver reproducibility. The original Oxford papers^1,2 did not assess reproducibility of MESTC scores, but extrapolation suggested that agreement was best for T because relevant kappa values were highest, 0.79 and 0.78 for estimated percentages of cortex showing tubular atrophy and interstitial fibrosis, respectively.² There were disagreements between the Oxford pathologists when M was tested on 16 biopsies. For cases near the borderline, with an original mesangial score of 0.5–0.7, there was high interobserver variation for the cutoff of 50% of glomeruli showing hypercellularity.² This was the only MESTC element directly investigated, although just on a few biopsies and without reported statistics.

Table 1.

Interobserver reproducibility of Oxford scores in IgA nephropathy

Reference in Order Discussed in Text	No. of Observers	Location of Observers	No. of Renal Biopsies	ICC M	ICC E	ICC S	ICC T	ICC C	Notes
2	3, 4, or 5	Various countries	265	ND	ND	ND	ND	ND	ICC only on original dataset
18	2	Different countries	187	ND	ND	ND	ND	ND	As in ref. 2
20	2	Same institution	200	ND	ND	ND	ND	ND	As in ref. 2
22	4	Different institutions	100	NR	NR	NR	NR	NR	Seem as in ref. 2, not MESTC
26	5	Different institutions	Not clear: 411 in ref. 23?	NR	NR	NR	NR	NR	No results reported
27	5	Different institutions	411 in ref. 23	NR	NR	NR	NR	NR	Not MESTC, but as in ref. 2
19	5	Same institution	45	0.41	0.49	0.62	0.53	0.56
21	3	Same institution	410	0.66	0.62	0.68	0.75	ND	Means of two ICC scores
24	2	Same institution	412	0.77	0.74	0.86	0.86	0.84
25	4? 8? 4 pairs?	Four institutions	40	0.76	0.85	0.89	0.84	0.89	Cronbach alpha scores
23	5	Different institutions	411	0.45	0.37a	0.39a	0.45	ND	Means of ten ICC scores
15	Not said	55 institutions	1147	0.28a	0.19a	0.51	0.53	0.24a

ICC, intraclass correlation coefficients (kappa scores); M, mesangial cellularity; E, endothelial hypercellularity; S, segmental sclerosis/capsular adhesion; T, tubular atrophy/interstitial fibrosis; C, (fibro) cellular crescents; ND, not done; NR, not reported; ref., reference.

^aKappa values below 0.40.

Two other studies similarly assessed ICCs on features that were not MESTC^18,20 and so were not tests of reproducibility of the classification. Another paper was ambiguous, claiming to assess MESTC scores, although the kappa scores were apparently on features in the Oxford paper.²² Two papers mentioned ICCs for MESTC but gave either no results²⁶ or kappa values claimed to be for MESTC²⁷ although actually for features in the Oxford paper.²³

Six papers giving reproducibility of MEST or MESTC varied in numbers of biopsies and observers and in reported degree of independence between observers. Observers were at the same institutions in each of three papers, one with moderate to substantial kappa scores,¹⁹ one with substantial scores,²¹ and one with substantial to outstanding scores.²⁴ Another paper assessed scoring on 40 biopsies by observers in four institutions, but whether this was by four pairs or four or eight individually was unclear. Each of Cronbach alpha reliability estimates was strong.²⁵

Two studies with many biopsies (more than 400) and reportedly independent observers had lower reproducibility statistics than these four. One study circulated 411 biopsies to five pathologists at different institutions.²³ The average kappa statistics of ten interrater scores were 0.45 for M, 0.37 for E, 0.39 for S, and 0.45 for T (T1 and T2 were combined). Accordingly, agreement by Oxford standards^2,15 was moderate for M and T and poor for E and S. The other study,¹⁵ from the Oxford working group,^1,2 compared scores of pathologists in 55 centers that contributed 1147 biopsies with the VALIGA study,²⁸ with central review by one or two pathologists. Kappa values were 0.28 for M, 0.19 for E, 0.51 for S, 0.53 for T, and 0.24 for C. Agreement was, therefore, moderate for S and T and poor for M, E, and C.

Combining the findings of these two papers,^15,23 agreement of Oxford scoring appeared moderate for T, moderate or poor for M and S, and poor for E and C.

Relation of Oxford Scores to Outcome

Because reproducibility of Oxford scores appeared unreliable, how much weight to put on any study of the relation of scores to outcome was uncertain. Studies varied in types of patients and in adherence to Oxford guidance (Supplemental Table 1). Examples of differences from Oxford were different clinical features or different minimum numbers of glomeruli or scoring features not in Oxford. Most papers aggregated T1 and T2 and C1 and C2. Before C was introduced,³ many papers studied crescents, usually reporting absence or presence, corresponding to C0 and C1/2, respectively. For this review, these papers were regarded to have given MESTC scores.

Numbers of cases given scores ranged from 26 to 4151. The commonest stated numbers of observers were one and two in 21 and 22 papers, respectively. Three papers included review of the VALIGA biopsies,^28–30 apparently mostly scored by one observer.¹⁵ No study with a single scorer tested intraobserver agreement, meaning consistency of scoring of the same biopsies at different times.

The median length of follow-up ranged from 6 months to 22 years. Various outcome measures were used: ESKD alone; ESKD combined with something else; rate of eGFR decline; remission; and others (Supplemental Tables 2–4). The proportion of patients reaching end points, if given, ranged between 4% and 59%. Table 2 summarizes the findings of multivariate analyses of outcomes. For simplicity, considering the variety of studies and unreliability of scores, totals are given of individual Oxford features associated with outcomes. The commonest clinical feature related to outcomes was eGFR at biopsy. Immunosuppressive treatment is analyzed in the next section.

Table 2.

Oxford scores and clinical features related to clinical outcomes in multivariate analyses in IgA nephropathy

Clinical Outcome	No. of Papers	No. of Analyses	No. of Analyses with C	Clinical Feature Related To Outcome: eGFR or scr	Clinical Feature Related To Outcome: Proteinuria	Clinical Feature Related To Outcome: BP	Clinical Feature Related To Outcome: None of These	Clinical Feature Related To Outcome: Not Said	M	E	S	T	C	Total Oxford Scores Related To Outcome	None of MESTC Related To Outcome
ESKDa	27	34	22	23	16	4	4	3	3	3	7	24	1	38	6
ESKD with another end pointb	47	53	37	29	17	11	2	13	14	2	14	34	10	74	11
Rate of eGFR declinec	12	12	9	5	4	3	0	4	2	1	4	10	1	18	1
Remissiond	6	6	2	2	1	0	2	1	2	0	1	2	0	5	2
Otherse	19	20	14	4	6	2	3	6	5	1	3	3	1	13	10
Total	111f	125g	84	63	44	20	11	27	26	7	29	73	13	148	30

scr, serum creatinine concentration; M, mesangial cellularity; E, endothelial hypercellularity; S, segmental sclerosis/capsular adhesion; T, tubular atrophy/interstitial fibrosis; C, (fibro) cellular crescents.

^aDetails in Supplemental Table 2.

^bSuch as 50% decline in eGFR, details in Supplemental Table 3.

^cDetails in Supplemental Table 4.

^dRemission (including decline in proteinuria), details in Supplemental Table 4.

^eSuch as decline in eGFR or increase in cr to various extents or persistent proteinuria, details in Supplemental Table 4.

^fTwo outcomes in 15 of 92 papers and three outcomes in 2 of 92 papers.

^gTwo analyses in 16 of 92 papers, three analyses in 4 of 92 papers, and four analyses in 3 of 92 papers.

As a crude summary, there were 125 analyses, with 148 individual Oxford scores related to outcomes, 73 T, 29 S, 26 M, 7 E, and 13 C (although only 84 analyses can be regarded to have studied C). Thirty analyses showed no relation of any element to the various outcomes.

After 2017,³ only two papers in Supplemental Table 1 mentioned podocyte hypertrophy/tip lesions. One³¹ was small (12 affected biopsies) and short-term (outcome at 6 months). The other³² was even smaller (nine biopsies), without analysis of relation to outcome.

Relation between Oxford Scores and Immunosuppressive Treatment

Only two studies were prospective, controlled trials of immunosuppression. One, the TESTING study^33,34 stratified 503 eligible patients by various features including Oxford E, apparently scored by pathologists at each site. Some patients had already had immunosuppression. At a median of 3.5 years after randomization, when the methylprednisolone groups had reached significantly fewer end points than the placebo groups, there was no significant difference in the effects of methylprednisolone on the composite kidney failure outcome between E0 and E1.³⁴ Other Oxford scores were not analyzed.

The STOP-IgA nephropathy trial³⁵ included 162 eligible patients, none previously immunosuppressed, but only 70 renal biopsies were given Oxford scores retrospectively by one observer.³⁶ In multivariate analysis at 3 years after randomization, the only MEST score related to any outcome measure was T with ESKD. For C scores, the proportional hazards assumption was not met, and C was not included in Cox analyses.³⁶ At a median of 7.4 years, there was no difference in outcome between the immunosuppression and control groups. Oxford scores were not mentioned.³⁷

Only 23 of the other 89 papers reported whether patients had been treated with immunosuppression before biopsy, nine treated and 14 not (Supplemental Table 5). Immunosuppression was used after biopsy in no patients in five studies (six papers), in all patients in four studies, and between 0% and 100% in the rest, although 13 only said that some were treated and ten gave ambiguous figures. Use of immunosuppression after biopsy was often related to Oxford scores, especially C and E. A significant relation between at least one Oxford feature and use of immunosuppression was reported in 28 papers (Supplemental Table 5). In these, there were 47 individual Oxford scores, 21 C, 11 E, six M, five T, and four S, while four other papers showed no relation.

Without the two trials, Table 3 reports outcomes related to immunosuppression in the other 25 papers that treated between none and all patients after biopsy and included immunosuppression as a variate in outcome analyses (Supplemental Table 5). Seven did not say whether immunosuppression had a relation to outcome. In the other 18 papers, significantly improved outcomes with immunosuppression were found in ten, no improvement in seven, and worse outcome in one. There were no consistent relations between Oxford scores and outcomes in immunosuppressed patients.

Table 3.

Oxford scores related to clinical outcomes in multivariate analyses including treatment with immunosuppression in IgA nephropathy

Clinical Outcome After Immunosuppressiona	Clinical Outcome	No. of Papers	No. of Analyses	No. of Analyses with C	M	E	S	T	C	None of MESTC
Improved	ESKD	4	7b	3	0	0	2	6	1	0
	ESKD with another end point	2	2	2	0	1	0	1	1	1
	Rate of eGFR decline	1	1	1	0	0	0	1	0	0
	Other	3	3	3	0	0	0	0	1	2
Total: Improved		10	13b	9	0	1	2	8	3	3
Not improved	ESKD	2c	2	2	0	0	0	2	0	0
	ESKD with another end point	4	4	3	1	0	1	3	1	0
	Other	2c	2	2	0	0	0	0	0	2
Total: Not improved		8c	8	7	1	0	1	5	1	2
Worse: Total	ESKD	1	1	1	0	0	0	1	0	0
Not said	ESKD	2	2	2	0	0	1	1	0	1
	ESKD with another end point	2	2	1	0	0	0	1	0	1
	Other	3	4d	2	3	0	1	1	0	1
Total: Not said		7	8d	5	3	0	2	3	0	3
Overall total		26c	30b,d	22	4	1	5	17	4	8

M, mesangial cellularity; E, endothelial hypercellularity; S, segmental sclerosis/capsular adhesion; T, tubular atrophy/interstitial fibrosis; C, (fibro) cellular crescents.

^aDetails in Supplemental Table 5.

^bFour analyses in one paper.

^cTwo outcomes in one paper.

^dTwo analyses in one paper.

Discussion

The Oxford classification was an admirable idea to give uniformity of reporting of pathological features and clinical guidance, specifically prognosis. Following PRISMA 2020,¹² selection of papers for review was straightforward, with little opportunity for bias in extraction of details. Various sources of bias were found in papers, such as retrospective bias in selection of cases, treatment, and availability of biopsies. One bias was the widespread assumption that scoring was reliable, although its reproducibility was not formally examined before introduction.^1–4

The Oxford group recognized that even lower reproducibility could be expected in routine practice without the advantage of the iterative processes of the working group.¹ This was confirmed by two large studies^15,23 with independent observers and lower reliability statistics than the others,^19,21,24,25 which were considered less representative of everyday application of the classification (Table 1). In these two,^15,23 agreement was moderate for T, moderate or poor for M and S, and poor for E and C. The Oxford group gave its view: “We conclude that differences in the scoring of MEST-C criteria between local pathologists and a central reviewer have a significant impact on the prognostic value of the Oxford classification.”¹⁵ That paper discussed reasons for discrepancies, which have implications for practical application of the classification and interpretation of published findings.

Moderate or poor reproducibility of scores in IgA nephropathy reflects findings in other conditions. A study testing agreement between pathologists found kappa values of 0.40 or over for only 25 of 48 glomerular descriptors in the nephrotic syndrome.³⁸ A test of the Banff renal transplant pathology classification found kappa values of 0.40 or over for only two of 32 features.³⁹ A study of lupus nephritis showed poor agreement, a mean kappa of 0.26, between four pathologists applying the International Society of Nephrology/Renal Pathology Society 2003 classification.⁴⁰

Counts of components of MESTC related to outcomes were necessarily crude, but any possible loss of important details was counterbalanced by the unreliability of assessment of components and so the uncertainty that findings of any study were generally applicable (Table 2). There was no consistent finding that scores were independently related to outcomes. T was the commonest feature associated with most outcomes, as others found.¹⁵ The dominance of tubular changes, rather than glomerular, in relation to renal function was long known,⁴¹ as was the importance of chronic damage, assessed morphometrically in relation to prognosis in IgA nephropathy.⁴²

One finding was that reproducibility of scoring of C, crescents, was poor, despite detailed definitions.^2,4 One suggested reason was the difficulty of distinction of crescents from pseudocrescents associated with podocytopathic segmental sclerosing lesions.¹⁵ Incidentally, evidence about the prognostic value of podocytopathic features was severely limited.^31,32 C and E scores often correlated with use of immunosuppression after biopsy, although both were poorly reproducible, showing bias in the use of immunosuppression. Relatively few studies said whether immunosuppression was related to outcomes, but there was little evidence that C and E scores correlated with response to immunosuppression (Table 3). The two controlled trials, which had contradictory findings, gave scant indication about the value of scores^33,34,36 and were discussed without any mention of scores.⁴³ Four other trials in IgA nephropathy, NEFIGAN, PROTECT, APPLAUSE-IgA nephropathy, and DAPA-CKD, did not consider Oxford scores at all in stratification of patients or response to treatment.^44–47

Predictive models including Oxford scores have been constructed.^{4,28,30,48–50} The overriding significance of clinical factors is emphasized: “these small improvements (in prognostication, by adding Oxford scores) … reflect the already high discriminatory value of the model with clinical data only.”³⁰ Prediction models probably assume an unjustified accuracy of scoring. Kidney Disease Improving Global Outcomes explicitly agreed recommending treatment with immunosuppression for acute kidney injury resulting from a rapidly progressive glomerulonephritis IgA nephropathy course with extensive crescent formation, although it is important to realize that neither this calculator⁴⁹ nor the MEST-C score nor the presence or number of crescents can presently be used to determine the likely impact of any particular treatment regimen.¹¹

Improvement in reproducibility of Oxford scoring will be necessary before its true value in prognosis is established. Educational guidance to pathologists has been proposed^3,15 and should be investigated by multicenter studies.

Disclosures

All authors have nothing to disclose.

Funding

None.

Author Contributions

Conceptualization: Alexander J. Howie.

Data curation: Alexander J. Howie, Alexander D. Lalayiannis.

Formal analysis: Alexander J. Howie, Alexander D. Lalayiannis.

Writing – original draft: Alexander J. Howie.

Writing – review & editing: Alexander J. Howie, Alexander D. Lalayiannis.

Data Sharing Statement

All data used in this study are available in this article and/or supporting information. Previously published data were used for this study.

Supplemental Material

This article contains the following supplemental material online at http://links.lww.com/KN9/A374.

Supplemental Table 1. Details of 92 papers with Oxford classification of IgA nephropathy and multivariate analysis of outcome including clinical factors and at least one individual Oxford element.

Supplemental Table 2. Studies combining clinical factors with Oxford scores in multivariate analysis of outcome: end point: ESKD. Reference numbers are those in Supplemental Table 1. Summary of counts included.

Supplemental Table 3. Studies combining clinical factors with Oxford scores in multivariate analysis of outcome: end point: ESKD combined with 50% decline in eGFR (combined) or ESKD with another end point. Reference numbers are those in Supplemental Table 1. Summary of counts included.

Supplemental Table 4. Studies combining clinical factors with Oxford scores in multivariate analysis of outcome: outcome: something other than ESKD, alone or combined. Reference numbers are those in Supplemental Table 1. Summary of counts included.

Supplemental Table 5. References relevant to immunosuppression (n=89) other than controlled trials (n=3). Reference numbers are those in Supplemental Table 1.