Toward a working definition of C3 glomerulopathy by immunofluorescence

Transcription

1 & 2013 International Society of Nephrology Toward a working definition of C3 glomerulopathy by immunofluorescence Jean Hou 1, Glen S. Markowitz 1, Andrew S. Bomback 2, Gerald B. Appel 2, Leal C. Herlitz 1, M. Barry Stokes 1 and Vivette D. D Agati 1 1 Department of Pathology, Columbia University, College of Physicians and Surgeons, New York, New York, USA and 2 Division of Nephrology, Department of Medicine, Columbia University, College of Physicians and Surgeons, New York, New York, USA Precise immunofluorescence criteria for C3 glomerulopathy remain to be defined. Here we tested hierarchical immunofluorescence criteria with varying stringency for C3 glomerulopathy in a cohort with dense deposit disease as the gold standard and then applied these criteria to analyze the incidence of C3 glomerulopathy in membranoproliferative glomerulonephritis (MPGN) types 1 and 3. Among 319 archived cases of primary MPGN types 1 3, immunofluorescence reports were retrospectively coded as glomerular deposits of the following: C3 only; C3 dominant with trace or 1 þ immunoglobulin (Ig)M only; and C3 dominant and at least two orders of intensity stronger than any combination of IgG, IgM, IgA, and C1q. The most restrictive criteria of C3 only captured only half of the cases with dense deposit disease (compared with 8% of type 1 and 10% of type 3). Adding the most liberal definition identified 88% of those with dense deposit disease (compared with 31% of type 1 and 39% of type 3). The unaccounted 12% had stronger intensity of Ig staining, but it never exceeded the intensity of C3. Among MPGN type 3, 90% of C3 glomerulopathy cases were the Strife and Anders variant. Repeat biopsies in C3 glomerulopathy revealed a change in immunofluorescence pattern in 10 of 23 biopsies. The prevalence of low serum C3 and/or low C4 did not significantly differ among the three immunofluorescence criteria. Thus, C3 only is an impractical definition of C3 glomerulopathy, and we propose a definition of C3 dominant and at least two orders of magnitude more intense than any other immune reactant, which requires validation by alternative pathway evaluation. These criteria provide a framework for identifying patients most likely to benefit from investigations of alternative complement pathway dysregulation. Kidney International (2014) 85, ; doi: /ki ; published online 25 September 2013 Correspondence: Vivette D. D Agati, Department of Pathology, Columbia University, 630 W. 168th Street Room VC14 224, New York, New York 10032, USA. vdd1@columbia.edu Received 10 May 2013; revised 27 June 2013; accepted 3 July 2013; published online 25 September 2013 The clinical and pathologic approach to membranoproliferative glomerulonephritis (MPGN) has evolved markedly in recent years. The term MPGN was originally introduced in the 1970s to describe a morphologic pattern of glomerular proliferation with mesangial interposition and duplication of glomerular basement membranes as defined by light and electron microscopy. 1 This proliferation occurs in response to deposition of immunoglobulins (Igs) and/or complement in the glomerular capillary walls. Historically, three major forms of MPGN have been recognized. In type 1, there is mesangial proliferation with regular circumferential mesangial interposition and double contouring of glomerular basement membranes in response to subendothelial electron-dense deposits. 2 Type 2, also known as dense deposit disease (DDD), is characterized by distinctive, highly electron-dense, ribbon-like intramembranous deposits that transform the lamina densa, often associated with ring-shaped mesangial deposits and variable hump-shaped subepithelial deposits. 3 In type 3 of Strife and Anders, there are more complex and illdefined intramembranous deposits associated with subendothelial and subepithelial deposits, causing disruption and fraying of the lamina densa. 4 By contrast, type 3 of Burkholder is characterized by more discrete subendothelial, intramembranous, and subepithelial deposits, often with overlapping features of MPGN type 1 and membranous glomerulopathy 4,5 (Figure 1). These categories were cumbersome to apply because of their complexity, the frequent occurrence of morphologic heterogeneity in a given case, and the lack of pathogenetic specificity. For example, many cases of DDD lack membranoproliferative features altogether, causing the term MPGN type 2 to be discarded. 6 With greater understanding of etiology over the past decade, there has been a major shift away from morphologic patterns to focus on composition of the deposits as defined by immunofluorescence (IF). Those cases with substantial deposits of Ig are considered immune complex mediated and should prompt an investigation of underlying autoimmune, infectious, or paraprotein-related disease. 7 The term C3 glomerulopathy (C3G), proposed to encompass those cases with deposition of C3 only, includes DDD and those forms of MPGN type 1 and type 3 with predominant deposits of C3. 8,9 A major breakthrough was the 450 Kidney International (2014) 85,

2 J Hou et al.: A working definition of C3G by immunofluorescence clinical investigation recognition that abnormal control of the alternative pathway (AP) of complement underlies C3G, irrespective of morphologic appearance. Potential abnormalities include genetic deficiencies in regulators of the alternative complement pathway (such as complement factor (CF)H, CFHR1 5, CFI, and CD46) 10,11 and autoantibodies to factor H, factor B, and a c e 50 µm b d f * 1 µm 10 µm 2 µm 1 µm 2 µm Figure 1 Light microscopy (LM) and electron microscopy (EM) findings in membranoproliferative glomerulonephritis (MPGN) types 1 3. (a) LM showing MPGN1 with lobular accentuation and membranoproliferative features characterized by glomerular basement membrane (GBM) duplication (arrow) and mesangial cell interposition (arrowhead; Jones methenamine silver, 400). (b) EM of MPGN1 showing electron-dense subendothelial deposits (asterisk) associated with GBM duplication (6000). (c) LM showing MPGN2/ DDD (dense deposit disease) with uniform thickening of the GBM (arrow; periodic acid Schiff, 600). (d) EM of MPGN2/DDD showing electron-dense deposits in the lamina densa causing alternating prominent thickening (asterisk) and thinning of the GBM (6000). (e, f) EMs of MPGN3: Burkholder (e) and Strife and Anders (f) subtype (6000). Note the discrete subepithelial and subendothelial deposits in Burkholder variant (e) as compared with the more ill-defined, complex intramembranous deposits in Strife and Anders variant (f). * the C3 convertase (known as C3 nephritic factors) Proteomic analysis by mass spectrometry performed on laser-captured glomeruli has identified complement debris (including C5, C6, C7, C8, and C9) in the glomerular deposits, consistent with activation of the AP in the fluid phase. 15 With the advent of specific clinical testing for abnormalities of the AP and specific therapies targeting the AP, 16 these distinctions are gaining greater clinical relevance. An impediment in the diagnosis of C3G is the lack of a working definition of IF criteria that has been developed and validated in a well-defined disease cohort. Although the term C3 only is often applied as a theoretical construct, it is not clear whether such a strict definition is practicable. For example, pathologists recognize that IgM is frequently trapped in areas of sclerosis and in thickened glomerular capillary walls in diverse glomerular diseases. 17 A major question is whether C3 dominance might occur on a background of less intense deposition or trapping of other immune reactants, including IgG, IgA, and C1q. We aimed to examine these questions by testing different set points for IF definition of C3G in a cohort of DDD as the gold standard. Once optimal criteria were found, we applied these criteria to a group of primary MPGN type 1 and type 3 to determine the incidence of C3G, define the relationship of C3G to the Strife and Anders versus Burkholder subgroups of primary MPGN type 3, and explore clinical and demographic correlates. RESULTS A total of 796 biopsies with diagnoses of MPGN 1 3 were identified from 1999 to After exclusion of 396 cases with clear underlying etiology (secondary MPGN due to hepatitis C infection, cryoglobulinemia, paraprotein deposition, and so on), 58 cases with inadequate tissue for IF or previous treatment with eculizumab, and 23 repeat biopsies, a total of 319 cases of primary MPGN were recruited into the study. As shown in Table 1, the largest number of primary cases was from MPGN1 (n ¼ 200), followed by MPGN3 (n ¼ 77) and MPGN2/DDD (n ¼ 42). Using MPGN2/DDD as the prototype for C3G, application of the strictest definition of C3-only staining (criterion 1) captured only 50% of the total cases. Allowing for low-level IgM deposition (criterion 2) increased the case recovery to 71.4%. Further broadening the criteria to permit low-level IgG, IgM, IgA, and/or C1q staining (criterion 3) identified 88.1% of total MPGN2/DDD cases. Broadening to criterion 4 added only Table 1 Primary membranoproliferative glomerulonephritis (MPGN) cases meeting criteria 1 3 Diagnosis Number of primary cases Criterion 1: C3 only (%) Criterion 2: C3 dominant and up to 1 þ IgM only (%) Criterion 3: C3 dominant and X2 orders of intensity greater than any combination of IgG, IgM, IgA, and C1q (%) Criteria 1, 2, or 3 (%) MPGN (8%) 13 (6.5%) 32 (16%) 61 (30.5%) MPGN2/DDD (50%) 9 (21.4%) 7 (16.7%) a 37 (88.1%) MPGN (10.4%) 11 (14.3%) 11(14.3%) 30 (39%) Total (14.1%) 33 (10.3%) 50 (15.7%) 128 (40.1%) Abbreviations: DDD, dense deposit disease; Ig, immunoglobulin. a Seven cases that fulfilled criterion 3 displayed C3-dominant staining with the following immunofluorescence patterns: two cases with IgG only; one case with IgM only; two cases with C1 only; and two cases with IgG, IgM, IgA, and C1. Kidney International (2014) 85,

3 J Hou et al.: A working definition of C3G by immunofluorescence 4.8% of DDD cases, while this change added 26.5% of MPGN1 cases and 13% of MPGN3 cases, suggesting loss of specificity. Therefore, the first three criteria were chosen as cutoffs for the gold-standard group of DDD and used as optimal defining criteria for C3G. We then applied these criteria to cases of MPGN1 and MPGN3 to determine the incidence of C3G in these groups. Sixty-one cases (30.5%) of MPGN1 and 30 cases (39%) of MPGN3 successfully met the defining criteria for C3G (Table 1). Of the 30 C3G cases recovered from MPGN3, 90% were Strife and Anders variant and 10% were Burkholder variant. Cases of MPGN 1 3 that did not meet the third cutoff for C3G are outlined in Table 2. Five MPGN2/DDD cases failed to meet the defining C3G IF criteria: two cases with C3 dominance and 41 þ IgG or IgM, two cases with C3 codominance with IgG or IgM, and one case (an outside consult) reported as negative for C3 and other immune reactants. Of note, no DDD case exhibited stronger staining intensity for Ig than C3. A significant subset of MPGN1 (26.5%) and MPGN3 (13%) cases also exhibited C3- dominant staining that failed to meet the C3G criteria. Codominant staining of C3 with IgG or IgM was observed more frequently in primary MPGN1 and MPGN3 than in MPGN2/ DDD. The incidence of IgG-dominant staining, which was never observed in MPGN2/DDD, was higher in cases of MPGN3 than in those of MPGN1. Analysis of repeat biopsies Twenty-three patients underwent repeat biopsies, including 12 with MPGN1, 8 with DDD, and 3 with MPGN3. Subsequent biopsies for each patient were then separately categorized and summarized. Overall, 13 of the 23 patients maintained the same IF staining patterns on initial and follow-up biopsies (Table 3). On repeat biopsy, 2 of the 23 patients met a more stringent criterion for C3G, and 4 of the 23 met a more liberal criterion for C3G. Only two C3G patients had follow-up biopsies not meeting the C3G criteria, although in both cases C3 staining was stronger than IgG. Two patients showed the opposite trend, with initial biopsies that showed C3-dominant staining not meeting the criteria but with repeat biopsy meeting the C3G criteria. Of the three patients initially diagnosed with acute post-infectious glomerulonephritis who had repeat biopsies diagnosed as C3G, two had MPGN1 and one had MPGN3 on repeat biopsy. Clinical features Of the 128 cases that met classification criteria as C3G, 72 (56%) were male with a mean age of 34.4 years (Table 4). In this cohort, 65 of the 98 patients (66%) with reported data on race/ethnicity were white. On average, these patients presented with renal insufficiency (mean serum creatinine 2.6; median 1.6 mg/dl and mean estimated glomerular filtration rate 59.3; median 48.8 ml/min per 1.73 m 2 ) and nephrotic-range proteinuria (mean 24-hour proteinuria 4.5; median 3.6 g/day). Serum complement levels were low in B75% of patients, primarily in the pattern of low C3 with normal-range C4 levels. These presenting historical and clinical data were examined according to original histopathologic diagnoses MPGN1, MPGN2, or MPGN3 as well as by our proposed IF criteria schemes for diagnosing C3G. Age and sex distribution were relatively similar among the C3G patients originally classified as MPGN1, MPGN2/DDD, and MPGN3. In contrast, Hispanic patients were more likely to demonstrate the MPGN2 and MPGN3 pattern than MPGN1. Patients with MPGN3 were also marked by significantly lower presenting serum creatinine (and correspondingly higher estimated glomerular filtration rate) than patients with MPGN1 or MPGN2. Although the presence of low C3 levels overall did not differ between the MPGN Table 3 Primary membranoproliferative glomerulonephritis (MPGN) cases with serial biopsies IF criterion of initial biopsy IF criterion of repeat biopsy a 5b a a 4 1 b 5a 1 b 5b Abbreviation: IF, immunofluorescence. Thirteen of 23 patients (57%) had no change in immunofluorescence criteria (bolded). a Two patients (8.7%) had an initial biopsy that failed to meet criteria 1 3 and a subsequent biopsy that did meet criteria. b Two patients (8.7%) had an initial biopsy that did meet criteria 1 3 and a subsequent biopsy that did not. Table 2 Primary membranoproliferative glomerulonephritis (MPGN) cases not meeting criteria 1 3 Diagnosis Number of cases NOT meeting criteria 1 3 (%) Criterion 4: C3 dominant and 41 þ any Ig/C1q (%) Criterion 5: C3 co-dominant 5a: C3/IgM co-dominant (%) 5b: C3/IgG co-dominant (%) Criterion 6: IgM dominant IgM4C3 (%) Criterion 7: IgG dominant IgG4C3 (%) Criterion 8: C3 negative (%) MPGN1 139 (69.5%) 53 (26.5%) 18 (9%) 25 (18%) 14 (7%) 16 (8%) 13 (6.5%) MPGN2/DDD 5 (11.9%) 2 (4.8%) 1 (2.4%) 1 (2.4%) (2.4%) MPGN3 47 (61%) 10 (13%) 4 (5.2%) 15 (19.5%) 1 (1.3%) 16 (20.8%) 1 (1.3%) Total 191 (59.9%) 65 (20.4%) 23 (7.2%) 41 (12.9%) 15 (4.7%) 32 (10%) 15 (4.7%) Abbreviations: DDD, dense deposit disease; Ig, immunoglobulin. 452 Kidney International (2014) 85,

4 J Hou et al.: A working definition of C3G by immunofluorescence clinical investigation Table 4 Clinical and biological features of C3G patients All C3G (n ¼ 128) C3G in MPGN1 (n ¼ 61) C3G in MPGN2 (n ¼ 37) C3G in MPGN3 IF criterion 1 (n ¼ 30) a P-value (n ¼ 45) IF criterion 2 (n ¼ 33) IF criterion 3 (n ¼ 50) P-value a Mean age (years) Female (%) Race/ethnicity (%) White Hispanic Black Asian Unknown Mean disease duration (months) b Median serum creatinine (mg/dl) b Median egfr (ml/min per 1.73 m 2 ) b Median 24-h proteinuria (g/day) b Median serum albumin (g/dl) b Serum complements (%) b Normal Low C3 and normal C Low C3 and low C Low C3 overall Hematuria (%) b None Microscopic Gross Hypertension (%) Abbreviations: C3G, C3 glomerulopathy; egfr, estimated glomerular filtration rate; IF, immunofluorescence; MPGN, membranoproliferative glomerulonephritis. a P-value from analysis of variance, testing null hypothesis that mean is equal among all three categories. b Excludes missing data for subjects on disease duration (n ¼ 51), serum creatinine (n ¼ 11), proteinuria (n ¼ 49), serum albumin (n ¼ 59), serum complements (n ¼ 26), and hematuria (n ¼ 36). subtypes, interestingly, no patients with MPGN2 demonstrated low C3 and low C4 in contrast to a small proportion of those patients with MPGN1 and MPGN3 with this complement profile. Women were overrepresented in the subgroup of C3G patients who were included based on IF criterion 3. This subgroup also included the only Asian patients in the cohort. Patients who met the strictest IF criterion 1 for diagnosis of C3G presented with the highest incidence of gross hematuria and the most severe renal dysfunction, with mean serum creatinine 3.5; median 1.9 mg/dl and mean estimated glomerular filtration rate 50.2; median 39.0 ml/min per 1.73 m 2 (which, in turn, likely influenced their significantly lower level of proteinuria than IF criteria 2 and 3 subgroups). Notably, the prevalence of low C3 and/or low C4 serum levels did not significantly differ among these IF criteria subgroups. Analysis of subset of secondary MPGN We retrospectively reviewed a random subset of 125 biopsies of secondary MPGN, including 45 hepatitis C associated MPGN (of which 18 also had cryoglobulinemia), 30 cryoglobulinemic glomerulonephritis, 30 lupus nephritis with membranoproliferative pattern, 10 hepatitis B associated MPGN, 5 combined hepatitis B and HIV associated MPGN, and 5 combined hepatitis B and C associated MPGN. Only one case of hepatitis C associated MPGN (without cryoglobulinemia) met the IF criterion 2; one case of hepatitis B associated MPGN met IF criterion 3; and no case met IF criterion 1. These data indicate that our proposed IF criteria for C3G are rarely found in biopsies with secondary MPGN. DISCUSSION C3G is a histopathologic diagnosis that aims to be both descriptive, by relaying the findings of IF microscopy, and physiologic, by implicating a complement-mediated (rather than immune complex mediated) glomerulonephritis. Although C3G is defined by IF findings in the proper clinical and morphologic context, no prior studies have investigated what the precise IF criteria for diagnosis should be in pragmatic terms. We applied a hierarchical set of IF criteria to define the optimal cutoff for the diagnosis of C3G using MPGN2/DDD as a gold standard. A definition of C3 only identified only half of the DDD patients, whereas a definition of C3 dominance of at least two orders of magnitude stronger than any other immune reactant captured 88.1% of DDD patients. Further relaxing the defining criteria captured a few additional DDD patients but many more MPGN1 and MPGN3 patients, suggesting reduced specificity. Our results indicate that the theoretical definition of C3G as C3 only staining is too stringent if the goal of diagnosing C3G is to identify suitable candidates for AP evaluation. Allowing for IgM staining increased the retrieval to 71.4% of MPGN2/DDD. The presence of IgM staining in other glomerular diseases has been attributed to nonspecific trapping in areas of sclerosis or glomerular capillary wall thickening, without necessarily contributing to disease Kidney International (2014) 85,

5 J Hou et al.: A working definition of C3G by immunofluorescence pathogenesis. 17 The potential for low-level IgG staining also may exist in C3G. Serologic evidence of AP dysregulation has been recently identified in more than half of a single cohort of patients with MPGN1, 18 with C3NeF, an acquired autoantibody of C3 convertase, identified in MPGN1 as frequently as it was in C3G. In that study, IF revealed concomitant C3 and IgG deposits in a subset of patients who had clear serologic evidence of AP abnormalities. Our findings agree that the presence of IgG does not exclude C3G. Our proposed working definition of C3G as C3 dominance of at least two orders of magnitude more intense than any other immune reactant, beyond identifying 88.1% of MPGN2/DDD cases, qualified 30.5% of cases of MPGN1 and 39% of cases of MPGN3 as C3G. Among cases of MPGN3, the vast majority was Strife and Anders (90%) variant, which is consistent with the previously reported finding of C3-dominant deposits in Strife and Anders variant in a smaller cohort. 19 Although our biopsy database archived cases according to historical MPGN categories, it should be emphasized that C3G is not restricted to a membranoproliferative pattern and may exhibit other histological phenotypes. A diagnosis of C3G should prompt an investigation of the alternative complement pathway, specifically testing for mutations in regulators (e.g. factor B) and inhibitors (e.g. factor H) of the AP, autoantibodies directed at such regulators and inhibitors, and a functional assessment of AP activity. These investigations are noninvasive and, in most cases, can influence therapeutic decisions, allowing for the opportunity to tailor therapies to detected defects in the AP. 20 Given the importance of such testing, any diagnostic criteria for C3G should veer toward being more inclusive rather than exclusive. In the largest study of 56 cases of C3G to date, for example, acquired and genetic complement abnormalities were found in patients with phenotypes of MPGN1 and MPGN2/DDD. 18 Clinical features of C3G have also been reported in a pediatric cohort 21 and familial C3G. 22,23 Analysis of a subset of patients with diverse causes of secondary MPGN found only 2 of 125 (1.6%) cases meeting these proposed IF criteria for C3G. Our data now provide a practicable, nonconstrictive, pathologic definition of C3G that identifies those patients most likely to benefit from genetic and functional studies of AP dysregulation. Analysis of repeat biopsies in 23 patients provides further support for our definition of C3G. After initial biopsy diagnosis of C3G by meeting IF criteria 1 3 in 23 patients, repeat biopsies in the majority (13/23) met the identical IF criterion as the first biopsy. On the other hand, 2/23 C3G patients met a more stringent criterion and 4/23 met a more liberal criterion, while still fulfilling the definition for C3G on repeat biopsy. Further, two patients meeting criterion 4 on initial biopsy met C3G criteria 2 and 3, respectively, on repeat biopsy. Conversely, only two patients fulfilling C3G criterion 1 on initial biopsy met criteria 4 and 5a, respectively, on repeat biopsy. These data support fluidity between these subgroups and the ability to transform from one criterion class to another over time. These data also support the importance of a less constrictive definition that allows for such fluidity. It is not clear as to why some patients had a repeat biopsy that transformed to a more liberal criterion beyond criterion 3. This finding is consistent with the small percentage of MPGN2/DDD patients who manifested variable intensities of Ig staining on initial biopsy while never exceeding the staining intensity for C3. Conceivably, glomeruli may be subjected to other forms of immunologic injury, and harboring an abnormality in AP regulation does not exclude the possibility of other immune assaults that could promote IgG or other Ig deposition, including classical pathway activation. More studies are needed to address the potential for multiple immunologic hits in this population. Whether glomerular staining for IgG represents inconsequential passive trapping of a circulating plasma protein or superimposed active immune-mediated injury capable of promoting glomerular leukocyte infiltration and proliferation requires further study. Clinical analysis of this C3G cohort revealed that patients with MPGN2/DDD were more likely to have isolated reduction in serum C3 (82.6%) compared with those with MPGN1 and MPGN3 (54.6% and 54.2%, respectively). No patient with MPGN2/DDD had low serum C3 and C4, as compared with a small percentage (23.6% and 8.3%, respectively) of MPGN1 and MPGN3 with this complement profile. These findings support the concept that classical complement activation may contribute to glomerular injury in some C3G patients with MPGN1 and MPGN3 phenotypes. 24 In other respects, those patients meeting defining criteria of C3G within the three major pathologic subcategories of MPGN had similar demographic and clinical presenting features, supporting the thesis that the IF pattern is a more important diagnostic criterion than the light and electron microscopic patterns. Interestingly, C3G patients who met the strictest IF criterion 1 for diagnosis of C3G presented with the highest incidence of gross hematuria, more severe renal dysfunction, and lower-level proteinuria than those meeting criteria 2 and 3. This tendency to an acute nephritic (as opposed to nephrotic) presentation resembles the typical clinical presentation of postinfectious glomerulonephritis. Further study is needed to determine differences in the incidence of preceding infections among these IF criteria subgroups. Notably, the prevalence of low C3 and/or low C4 levels did not significantly differ among these IF criteria subgroups. The most liberal criterion (criterion 3) for diagnosing C3G demonstrated a trend toward greater prevalence of low C4 than stricter criteria (1 and 2); however, that again raises the issue of whether a defect in AP may heighten susceptibility to or amplify injury from immune complex deposition in some C3G cases. The finding of several cases initially diagnosed as postinfectious glomerulonephritis but fulfilling criteria for C3G on subsequent biopsy underscores the potential difficulty in diagnosing C3G in such patients without clinical follow-up and repeat biopsy. 25, Kidney International (2014) 85,

6 J Hou et al.: A working definition of C3G by immunofluorescence clinical investigation Our study has a number of limitations. First, as a retrospective study, it is based on IF recordings in the biopsy report rather than reevaluation of the original IF glass slides. Thus, the precise character and distribution of the staining for each immune reactant and in each glomerulus cannot be evaluated. Because the IF findings are reported as glomerular intensity of staining, it is not clear as to how the distribution of the staining for C3 and the staining for other immune reactants compare. In particular, whether there is complete or partial colocalization in individual glomeruli cannot be assessed. Second, the relationship of IgM staining with the presence of sclerosing features also cannot be defined without examining the same glomerulus by light microscopy and IF using serial sections. Finally, although we provide presenting clinical data on our cohort of C3G, our study does not include workup of the AP system, treatment, or follow-up information, which will be the subject of future investigations. Thus, we cannot speak to the true sensitivity or specificity of the proposed criteria. Nevertheless, using the cases of DDD (diagnosed by EM) as a gold standard or true positive, we have clearly shown that a C3 only criterion has far too low sensitivity (50%) and that a more inclusive criterion, allowing for some low-level Ig staining, increases this sensitivity. In conclusion, we performed the first study to assess IF criteria for C3G in a large, well-defined biopsy cohort. By using unbiased hierarchical definitions to find an optimal working definition of C3G for practicing pathologists, we provide a framework for triage of patients who are most likely to benefit from molecular and functional assays of AP dysregulation. An optimal definition of C3G, one that is both descriptive and physiologic, should cast a relatively wide net to maximize opportunities to identify AP abnormalities. METHODS The Columbia Renal Pathology Laboratory database was queried for cases diagnosed as MPGN1, 2/DDD, or 3 from the years 1999 to Detailed pathology reports were retrospectively reviewed. We excluded the following cases: (1) all cases with a clear etiology (such as hepatitis C infection, cryoglobulinemia, and dysproteinemia); (2) cases where complete IF findings were not detailed; (3) cases inadequate for IF on frozen tissue; and (4) cases treated with eculizumab, because of potential confounding IF features. 27 A total of 319 biopsies of primary MPGN types 1 3 were identified, including 23 cases with multiple biopsies. For patients with repeat biopsies, only the first biopsy diagnostic of MPGN was analyzed in the initial criteria study. Three additional patients with initial biopsy diagnosis of postinfectious glomerulonephritis and subsequent biopsy diagnosis of MPGN were included among the repeat biopsy analysis. All biopsies were stained at the time of initial biopsy diagnosis according to standard techniques applied to frozen sections using fluorescein isothiocyanate conjugated antisera to IgG, IgM, IgA, C3c, C1q, fibrin, albumin, kappa, and lambda light chains (Dako, Carpinteria, CA). IF findings were graded on a scale of 0 to 3 by one of four renal pathologists as follows: 0; trace; 1 þ ;2þ; and 3 þ.for this study, the glomerular IF findings recorded in the biopsy reports were retrospectively reviewed. Using DDD as the validation group, we proposed hierarchical IF criteria with decreasing stringency for diagnosis of C3G: criterion 1, glomerular deposits of C3 only; criterion 2, C3-dominant deposits and p1 þ IgM only; and criterion 3, C3-dominant deposits and X2 orders of intensity stronger than any combination of IgG, IgM, IgA, and C1q. Illustrative examples are shown in Figure 2. These criteria were then applied to cases of primary MPGN1 and MPGN3 to determine the incidence of C3G so defined and the relationship to the Strife and Anders versus Burkholder variants of MPGN3. Cases not meeting the third cutoff were further classified as fulfilling criterion 4: C3-dominant deposits and o2 orders of intensity stronger than any combination of IgG, IgM, IgA, and C1q; criterion 5, C3 Criterion 1 Criterion 2 Criterion 3 C3 C3 C3 IgG IgM IgG Figure 2 Illustrative examples of immunofluorescence (IF) staining meeting criteria 1 3. IF microscopy findings (400). (a, b) Criterion 1: C3G case with features of membranoproliferative glomerulonephritis (MPGN)1 showing 2 3 þ mesangial and capillary wall staining for C3 (a), with negative immunoglobulin (Ig)G (b). (c, d) Criterion 2: dense deposit disease (DDD) case showing 2 þ glomerular capillary wall and mesangial staining for C3 with mesangial ring forms (c), with trace (±) IgM (d). (e, f) Criterion 3: DDD case showing 3 þ glomerular capillary wall and mesangial staining for C3 (e) with 1 þ IgG (f). Bar ¼ 100 mm. Kidney International (2014) 85,

7 J Hou et al.: A working definition of C3G by immunofluorescence co-dominant with IgM (5a) or with IgG (5b); criterion 6, IgM4C3; criterion 7, IgG4C3; and criterion 8, no C3 staining. For those cases defined by IF criteria as C3G, the medical record was reviewed for presenting demographic and clinical features including age, gender, race/ethnicity, duration of proteinuria or renal dysfunction, significant comorbidities (e.g. diabetes and hypertension), and laboratory and serologic findings (including serum creatinine, serum albumin, 24-hour urine protein excretion, urinalysis, and serum C3 and C4 levels). Statistical analysis For demographic, clinical, and laboratory findings, continuous variables were expressed as mean values and categorical variables were expressed as prevalence rates. Comparisons were made using analysis of variance (STATA version 11.0, StataCorp, College Station, TX) between three subgroups: MPGN1 versus MPGN2 versus MPGN3 by original diagnostic criteria, and IF criterion 1 versus IF criterion 2 versus IF criterion 3 by the proposed inclusion criteria for C3G detailed above. DISCLOSURE All the authors declared no competing interests. REFERENCES 1. Mandalenakis N, Mendoza N, Pirani CL et al. Lobular glomerulonephritis and membranoproliferative glomerulonephritis: a clinical and pathologic study based on renal biopsies. Medicine 1971; 50: Levy M, Gubler MC, Sich M et al. Immunopathology of membranoroliferative glomerulonephritis with subendothelial deposits (Type I MPGN). Clin Immunol Immunopathol 1978; 10: Habib R, Gubler MC, Loirat C et al. Dense deposit disease: a variant of membranoproliferative glomerulonephritis. Kidney Int 1975; 7: Strife CF, McEnery PT, McAdams AJ et al. Membranoproliferative glomerulonephritis with disruption of the glomerular basement membrane. Clin Nephrol 1977; 7: Burkholder PM, Hyman LR, Krueger RP. Characterization of mixed membranous and proliferative glomerulonephritis: recognition of three varieties. Perspect Nephrol Hypertens 1973; 1(Pt 1): Walker PD, Ferrario F, Joh K et al. Dense deposit disease is not a membranoproliferative glomerulonephritis. Mod Pathol 2007; 20: Sethi S, Fervenza FC, Zhang Y et al. C3 glomerulonephritis: clinicopathological findings, complement abnormalities, glomerular proteomic profile, treatment, and follow-up. Kidney Int 2012; 82: Sethi S, Fervenza FC. Membranoproliferative glomerulonephritis: pathogenetic heterogeneity and proposal for a new classification. Semin Nephrol 2011; 31: Fakhouri F, Fremeaux-Bacchi V, Noel LH et al. C3 glomerulopathy: a new classification. Nat Rev Nephrol 2010; 6: Habbig S, Mihatsch MJ, Heinen S et al. C3 deposition glomerulopathy due to a functional factor H defect. Kidney Int 2009; 75: Abrera-Abeleda MA, Nishimura C, Frees K et al. Allelic variants of complement genes associated with dense deposit disease. JAmSoc Nephrol 2011; 22: Strobel S, Zimmering M, Papp K et al. Anti-factor B autoantibody in dense deposit disease. Mol Immunol 2010; 47: Goodship TH, Pappworth IY, Toth T et al. Factor H autoantibodies in membranoproliferative glomerulonephritis. Mol Immunol 2012; 52: Ohi H, Watanabe S, Fujita T et al. Significance of C3 nephritic factor (C3NeF) in non-hypocomplementaemic serum with membranoproliferative glomerulonephritis (MPGN). Clin Exp Immunol 1992; 89: Sethi S, Gamez JD, Vrana JA et al. Glomeruli of dense deposit disease contain components of the alternative and terminal complement pathway. Kidney Int 2009; 75: Zuber J, Fakhouri F, Roumenina LT et al. Use of eculizumab for atypical haemolytic uraemic syndrome and C3 glomerulopathies. Nat Rev Nephrol 2012; 8: Bolton WK, Benton FR, Maclay JG et al. Spontaneous glomerular sclerosis in aging Sprague-Dawley rats. I. Lesions associated with mesangial IgM deposits. Am J Pathol 1976; 85: Servais A, Noel LH, Roumenina LT et al. Acquired and genetic complement abnormalities play a critical role in dense deposit disease and other C3 glomerulopathies. Kidney Int 2012; 82: Bomback AS, Appel GB. Pathogenesis of the C3 glomerulopathies and reclassification of MPGN. Nat Rev Nephrol 2012; 8: Nester CM, Smith RJ. Treatment options for C3 glomerulopathy. Curr Opin Nephrol Hypertens 2013; 22: He RJ, Xiao HJ, Wang SX et al. [Characteristics of pediatric C3 glomerulopathy with decreased factor H in 3 cases]. Zhonghua Er Ke Za Zhi 2012; 50: Athanasiou Y, Voskarides K, Gale DP et al. Familial C3 glomerulopathy associated with CFHR5 mutations: clinical characteristics of 91 patients in 16 pedigrees. Clin J Am Soc Nephrol 2011; 6: Gale DP, de Jorge EG, Cook HT et al. Identification of a mutation in complement factor H-related protein 5 in patients of Cypriot origin with glomerulonephritis. Lancet 2010; 376: D Agati VD, Bomback AS. C3 glomerulopathy: what s in a name? Kidney Int 2012; 82: Sandhu G, Bansal A, Ranade A et al. C3 glomerulopathy masquerading as acute postinfectious glomerulonephritis. Am J Kidney Dis 2012; 60: Zand L, Kattah A, Fervenza FC et al. C3 glomerulonephritis associated with monoclonal gammopathy: a case series. Am J Kidney Dis 2013; 62: Herlitz LC, Bomback AS, Markowitz GS et al. Pathology after eculizumab in dense deposit disease and C3 GN. J Am Soc Nephrol 2012; 23: Kidney International (2014) 85,