Subclassification of focal segmental glomerulosclerosis in IgA nephropathy: is it of clinical value? Evidence from the Oxford Classification cohort.
SS Bellur1, F Lepeytre2, O Vorobyeva1, S Troyanov2, HT Cook3, ISD Roberts1 on behalf of the International IgA Nephropathy Working Group.
1John Radcliffe Hospital, Oxford, UK
2Hôpital du Sacré-Coeur de Montréal, University of Montreal, Montreal, Canada
3Imperial College, London, UK
Address for correspondence:
Prof Ian SD Roberts, Department of Cellular Pathology, John Radcliffe Hospital, Headley Way, Headington, Oxford OX3 9DU. Tel: +44 1865 222889, Fax: +44 1865 220519
Keywords:
IgA Nephropathy, podocytopathy, segmental sclerosis
Abstract
Focal segmental glomerulosclerosis (FSGS) is a common finding in IgA nephropathy (IgAN). We assessed FSGS lesions in the Oxford Classification patient cohort and correlated histology with clinical presentation and outcome. In 137 individuals with segmental glomerulosclerosis or adhesion (S1), subclassification of lesions identified 38% of patients with podocyte hypertrophy, 10% hyalinosis, 9% resorption droplets within podocytes, 7% tip lesions, 3% perihilar sclerosis and 2% endocapillary foam cells. Reproducibility was good or excellent for tip lesions, hyalinosis and perihilar sclerosis, moderate for podocyte hypertrophy and poor for resorption droplets, adhesion only and endocapillary foam cells. Podocyte hypertrophy and tip lesions were strongly associated with greater initial proteinuria. During follow-up of patients without immunosuppression, those with these features had more rapid renal function decline and worse survival from a combined event compared to S1 patients without such features and those without FSGS. Also in individuals with podocyte hypertrophy or tip lesions, immunosuppressive therapy was associated with a better renal survival. In IgAN, the presence of podocyte hypertrophy or tip lesions, markers of podocyte injury, are reproducible. These podocytopathic features are strongly associated with proteinuria and, in untreated patients, carry a worse prognosis. Our findings support reporting these features alongside the S score of the Oxford Classification.
Introduction
The first pictorial illustration of focal segmental glomerulosclerosis (FSGS) by Fahr was published in the atlas of human pathology as early as 1925[1]. Since its recognition as a clinical entity distinct from minimal change disease in 1979[2], it has caught the attention of nephrologists and pathologists alike in a continued attempt to obtain a clear understanding of this disease process. Segmental glomerulosclerosis is characterised by segmental increase in the glomerular matrix with obliteration of capillary lumens and is a common morphological feature of biopsies, either as the only manifestation of glomerular injury or in combination with other pathology. It is a common finding in IgA nephropathy (IgAN), with at least one glomerulus showing segmental sclerosis in 76% of biopsies in the Oxford Classification of IgAN cohort[3] and 73% of biopsies in the only study to date that has attempted to subclassify segmental sclerosing lesions in IgAN[4].
FSGS is an adverse prognostic factor in IgAN. In the Oxford Classification study, the presence of segmental sclerosis or adhesion (S1) in at least one glomerulus correlated with a higher level of proteinuria at the time of biopsy, in addition to more rapid loss of renal function and worse renal survival, independent of other histological features, GFR, mean arterial blood pressure (MAP) and initial proteinuria, and follow-up MAP and proteinuria[5]. In a retrospective series of 1147 patients from 13 European countries, the VALIGA study confirmed the S score to be an independent predictor of outcome in IgAN[6].
Segmental sclerosis may result from organisation and fibrosis of segmental necrotising or endocapillary proliferative lesions. However, in the Oxford cohort, there was no significant correlation between the extent of segmental sclerosis and either necrosis or endocapillary hypercellularity. Alternatively, segmental sclerosis in IgAN may reflect adaptive haemodynamic changes or podocyte injury (podocytopathy) analogous to primary FSGS[7]. The underlying cause of the sclerosis might be associated with different histological features within the segmental sclerosing lesions. In the Oxford Classification, segmental sclerosis was not subclassified and, for the purpose of analysis, sclerosis and adhesions were merged in order to improve reproducibility of diagnosis. Following the First Oxford Conference on IgA nephropathy in June 2014, in which the classification of IgAN was reconsidered, one priority was to determine whether subclassification of the S score in IgAN is reproducible and of clinical value.
In this study, we defined the spectrum of histological features in IgAN-associated segmental glomerulosclerosis by re-evaluating the original Oxford Classification cohort, and correlated histology with clinical variables at the time of biopsy and at follow-up. Since the severity of the proteinuria is the most important factor influencing the decision to give immunosuppression in IgAN, we focused on which histological features within the S score were most associated with proteinuria and studied their prognostic value.
Results
Glomerular determinants of initial proteinuria
Two-hundred and sixty five patients were included in the Oxford cohort. Using the entire Oxford cohort and the initial scoring of lesions as continuous variables, the glomerular lesions independently associated with initial proteinuria were mesangial hypercellularity (ß=0.267 p<0.001), endocapillary hypercellularity (ß=0.201 p=0.001) and segmental glomerulosclerosis or adhesion (ß=0.138 p=0.03). Crescents were not predictive of the initial proteinuria independently of these lesions.
Substudy of lesions within segmental sclerosis
We focused on segmental glomerulosclerosis and determined which lesions were most associated with initial proteinuria. The clinical and demographic features are summarised in table 1.
Of the 137 biopsies with S1 with slides available for review, 38% showed podocyte hypertrophy, 10% hyalinosis, 9% resorption droplets within podocytes, 7% tip lesions, 3% perihilar location, 2% endocapillary foam cells in sclerosis, there were no collapsing lesions.
Reproducibility of lesions within segmental sclerosis
The various features used to subclassify FSGS lesions were scored independently by two pathologists in 121 biopsies available for second review. Reproducibility was excellent for perihilar sclerosis (k=0.86), good for tip lesions (k=0.65) and hyalinosis (0.60), moderate for podocyte hypertrophy (0.52) and poor for podocyte resorption droplets and endocapillary foam cells.
Correlation of segmental sclerosing lesions with clinical features at time of biopsy
The presence of tip lesions, podocyte hypertrophy and adhesions correlated with greater proteinuria at the time of biopsy (table 2) while segmental sclerosis without adhesions, hyalinosis and resorption droplets within podocytes did not. These 3 lesions remained associated with proteinuria independently of one another (data not shown).
We then assessed whether a dose dependent relationship existed. For podocyte hypertrophy, (median of 9.1% of total glomeruli when present), the initial proteinuria was 1.45 g/day (1.10-2.80) in S1 without podocyte hypertrophy, and 2.06 (1.26-3.29) and 3.02 (1.50-5.12) for S1 with less or more than the median of glomeruli with podocyte hypertrophy (p=0.008, trend test). The same could not be tested for tip lesions, found in a single glomerulus in all biopsies except for one, where 2 glomeruli with tip lesions existed. For adhesions (median 7.5% of total glomeruli when present), the initial proteinuria was 1.25 g/day (1.00-1.70) in S1 without adhesion and 1.95 (1.21-3.63) and 2.06 (1.20-3.67) for S1 with less or more than the median percent of glomeruli with adhesions (p=0.03, trend test).
There was no correlation between any of the above features and initial GFR or initial MAP. Patients with perihilar lesions and endocapillary foam cells were too few to test associations with clinical features.
Individual role of segmental sclerosing lesions on the initial proteinuria
To illustrate the influence of lesions within S1 on proteinuria, we classified patients into mutually exclusive groups defined starting with the most influential lesion by univariate analysis: Tip lesion, podocyte hypertrophy (without tip lesion), adhesion (without tip lesion or podocyte hypertrophy) and, finally, segmental glomerulosclerosis, hyalinosis or resorption droplets, which were not associated with proteinuria. We added the group with S0 for comparison. A higher level of initial proteinuria existed when tip lesions and podocyte hypertrophy without tip lesions were present (figure 1). Using multivariate analysis, tip lesions, podocyte hypertrophy and adhesions were associated with the initial proteinuria independently of other lesions within S1 (data not shown).
Predictive value of lesions within segmental sclerosis
We first assessed the predictive value of lesions within segmental sclerosis in patients without immunosuppression to assess their “natural” predictive value. We combined podocyte hypertrophy and tip lesion together since they were distinctly associated with greater proteinuria at the time of biopsy compared to segmental sclerosis lesions and adhesions (figure 1). Individuals with podocyte hypertrophy or tip lesions experienced a greater follow-up proteinuria, faster rate of renal function decline (table 3) and worse survival from a combined event (p=0.001, figure 2) despite receiving no less renin-angiotensin system blockade (50% with S0, 76% with S1 and podocyte hypertrophy or tip lesions, and 80% with S1 without these features). These findings remained independent of the follow-up proteinuria (data not shown).
Associations between lesions within FSGS and the response of immunosuppressive therapy
We tested the associations between immunosuppressive therapy and outcome in individuals with different FSGS features (table 4 and figure 3). Groups who received such treatments tended to be younger with greater proteinuria at the time of biopsy, and biopsies showing endocapillary hypercellularity and less tubular atrophy/interstitial fibrosis. Individuals with podocyte hypertrophy or tip lesion who received immunosuppression had better outcomes compared to those without treatment. Both groups had a similar use of renin angiotensin system blockade.
Discussion
We verified the association between segmental sclerosis and initial proteinuria independently of other glomerular lesions and identified tip lesions and podocyte hypertrophy as its strongest determinants. These features also correlated with a worst outcome in those who did not receive immunosuppressive treatment compared to S1 patients without these features and S0 patients. Finally, in patients with tip lesions or podocyte hypertrophy, those who received immunosuppression experienced a greater reduction in proteinuria and better survival compared to those who were not treated.
There are at least three mechanisms implicated in the development of segmental sclerotic lesions. They can form as a result of glomerular inflammation or necrosis as seen in inflammatory nephritides which subsequently heals by sclerosis. These lesions typically appear as bland hypocellular scars with broad-based adhesions to Bowman’s capsule. Segmental sclerosis is associated with adaptive haemodynamic changes and hyperfiltration changes in patients with a reduced number of nephrons (secondary FSGS). The characteristic appearance of such lesions is sclerosis and hyalinosis adjacent to the vascular pole of the glomerulus with glomerulomegaly of non-sclerosed glomeruli. Alternatively, segmental sclerosis can form as a result of podocyte injury (“podocytopathies”), as has been shown using animal models[8], [9]. In these studies, loss of podocytes with denudation of the glomerular basement membrane is followed by an adhesion to Bowman’s capsule and subsequently the formation of sclerotic lesions. Nephrotic FSGS that develops in the absence of glomerular immune deposits may result from viral or drug-induced glomerular injury or arise as a result of genetic abnormalities, principally involving proteins of the slit diaphragm[10]. Circulating “permeability factors” have been implicated in primary FSGS[11]. Morphological features of FSGS secondary to podocytopathies include glomerular tip lesions, podocyte hypertrophy, endocapillary foam cells and collapsing lesions[12]. We use the term podocyte hypertrophy by convention to describe the cells overlying segmental sclerosing lesions, reflecting their visceral epithelial location, although it is now recognised that these typically show a parietal epithelial phenotype. We describe these lesions in the context of IgAN as “podocytopathic” features. Of these lesions, podocyte hypertrophy was most frequently seen, followed by tip lesions. There were no collapsing lesions and only two with endocapillary foam cells. These lesions are associated with severe proteinuria but the relationship is not necessarily causal. Tip lesions are a frequent finding in patients with the nephrotic syndrome associated with many different glomerular diseases and might be a manifestation, rather than a cause, of the severe glomerular protein loss.
Segmental glomerulosclerosis is seen in various glomerular diseases associated with immunoglobulin deposits, including lupus nephritis and IgAN, and the glomerular morphology may closely resemble primary FSGS; no morphological type of segmental lesion is specific for a single disease process. FSGS-like lesions constitute one of the classes (Class II) in the Haas classification of IgAN[13]. In the Oxford Classification study it was demonstrated that the presence of segmental sclerosis (S) was an independent prognostic marker5, in addition to mesangial hypercellularity (M), endocapillary hypercellularity (E) and tubular atrophy/interstitial fibrosis (T). Segmental sclerosis associated with immune complex-mediated glomerulonephritis may develop as a result of healed necrotising or proliferative lesions. However, Hill and co-workers4 [14] have provided evidence that a significant proportion of FSGS seen in association with IgAN represents podocyte injury similar to that seen in primary nephrotic FSGS. They noted that adhesions without inflammation in the glomerular tuft occurred most commonly in primary FSGS (69%), followed by IgAN (41%), and were least common in lupus nephritis (8%), thus showing that, as in primary FSGS, adhesions in IgAN may result from podocyte injury and not merely secondary to inflammation. In the Oxford Classification cohort, adhesions without sclerosis in the underlying tuft were seen in 7% of biopsies. Hill et al16 studied the immunohistochemical features of 15 cases of IgAN with FSGS, demonstrating focal loss of the podocyte markers glomerular epithelial protein (GLEPP-1) and nephrin, particularly at sites of capsular adhesions. They demonstrated proliferation of cells that are positive for parietal epithelial cell antigens, paired box antigen 2 (PAX2) and cytokeratin, at sites of adhesions and overlying sclerosed segments, again pointing to a podocytopathy similar to primary FSGS. These pieces of evidence provide further support to earlier studies demonstrating that IgA stimulates mesangial cell proliferation with release of mediators that cause podocyte injury[15], [16] and also the fact that excretion of podocytes in the urine reflects disease progression[17].
In the Oxford classification study, segmental sclerosis and adhesions were merged as a single entity for the purpose of analysis in order to improve interobserver agreement. Segmental sclerosis involving at least one glomerulus was seen in 76% biopsies from the 265 patients3, but was not further subclassified. A schema for the subclassification of primary FSGS, the Columbia classification12, is widely used and has been demonstrated to be of prognostic value[18], [19]. Patients with tip variant FSGS have the highest frequency of remission following steroid therapy and a better long term outcome, whereas patients with collapsing variant are most likely to progress to end stage chronic kidney disease. In the Oxford Classification cohort, tip lesions and podocyte hypertrophy were associated with higher levels of proteinuria at the time of diagnosis, supporting the view that these lesions reflect podocyte injury in IgAN. The Columbia classification has not been validated in IgAN and to simply apply this schema would require potentially incorrect assumptions to be made, and risk losing valuable information from the histological analysis. We therefore chose to score the individual histological features seen within segmental sclerosing lesions, including those which form the basis of the Columbia schema.