IgG subclasses in cryoglobulins: link to composition and clinical manifestations

Background Cryoglobulins (CG) are immunoglobulins which precipitate at low temperature. The analysis of IgG subclass composition of CG is poorly reported. The aim of this study was to determine the subclasses of IgG-containing type I and mixed type II and III CG in relation to clinical manifestations. Methods Out of a previous series of 1675 patients, inclusion criteria were a cryoprecipitate > 1 mL and a total IgG > 300 mg/L. IgG subclasses were quantified by immunoturbidimetry, rheumatoid factor (RF), and C4 by immunonephelometry. Clinical parameters were collected from hospital charts. Results CG samples from 86 patients were included, 10 type I CG and 76 mixed CG. Type I CG subclasses were IgG1 (6/10) and IgG2/IgG3 (4/10), never IgG4. IgG subclass in type II vs. III CG were 73.3 ± 15.2% vs. 52.5 ± 20.7% for IgG1 (p < 0.0001), 15.4 ± 8.2% vs. 25.9 ± 14% for IgG2 (p < 0.0001), 8.4 ± 12.4 vs. 21.2 ± 14% for IgG3 (p < 0.0001), and 3 ± 5.2% vs. 0.5 ± 1.2 for IgG4 (p < 0.0001). In mixed CG, the higher proportion of IgG4 was associated with RF positive CG (p = 0.01) and low C4 (p = 0.01). In type I CG, IgG1 were associated with severe vasculitis manifestations, IgG2/IgG3 with cutaneous or renal manifestations. In mixed CG, IgG2 was the only subclass associated with CG manifestations, with a higher concentration in asymptomatic (162.6 ± 29.5 mg/L) vs. symptomatic patients with cutaneous (103 ± 17.8 mg/L, p = 0.04) and neurological (108 ± 24 mg/L, p = 0.04) manifestations. Conclusion In type I IgG CG, IgG1 was the main CG subclass associated with CG vasculitis. In mixed CG, low IgG2 concentration was linked to CG cutaneous and neurological manifestations.


Introduction
Cryoglobulins (CG) are immunoglobulins (Ig) which precipitate at low temperature. They are classified as type I CG, composed of a monoclonal Ig (IgM or IgG), type II CG composed of a monoclonal and polyclonal Ig, and type III CG composed of polyclonal Ig. Type II and type III CG are called mixed CG, with monoclonal or polyclonal IgM with rheumatoid factor (RF) activity (IgM anti-IgG) [1].
IgG is the most important Ig class of the humoral immune response, with 4 subclasses with structural differences of the heavy chain, with distinct physicochemical, biological, and immunological properties. The IgG effector functions, such as immune complex formation, complement binding, and activation, depend on the Fc fragment [2]. Larger immune complexes are formed with IgG1 and IgG3, which efficiently bind C1q and other complement fractions to activate the classical pathway. Due to its spatial structure, CH2 glycosylation, and difficult accessibility of CH domains, IgG4 cannot form immune complexes and activate complement by the classical pathway [3]. Immune complex formation and complement activation are involved in CG pathogenesis [4]. Different IgG subclasses are involved in the pathogenesis of immune complexes, in particular in autoimmune renal diseases [5,6]. This topic has been rarely reported in the literature for cryoglobulinemia.
The purpose of the present study was to focus on the IgG subclasses and their associated clinical manifestations, in type I and mixed CG from a previous series of 1675 characterised CG [22].

Population and data collection
The present retrospective study was conducted at the University Hospitals of Lyon, France. From 1675 patients with positive detection of CG tested in a previous study [22], this study included samples for whom a cryoprecipitate was kept in the cryoglobulin bank. Inclusion criteria for subclass identification were a cryoprecipitate volume > 1 mL and a total IgG concentration > 300 mg/L. Cryoprecipitate volume needed to determine subclasses was at least 300 μL. These inclusion criteria related to volume and IgG concentration in the cryoprecipitate may have induced some patient selection bias, in particular for the relationships between subclasses and clinical signs.
The protocol was approved by the Ethics Committee of the Hospitals of Lyon for the protection of the people participating in biomedical research under the number AC-2016-2729.

CG characterisation
All CG were detected and analysed as previously described [22]. Total IgG concentration in the cryoprecipitate was measured by immunonephelometry (BN ProSpec®, Siemens Healthcare, St Denis, France), and the four IgG subclasses were assessed by immunoturbidimetry (Optilite®, The Binding Site, St Egrève, France). The sensitivity limit inherent to this immunoturbidimetric method (150 mg/L for IgG1, 20 mg/L for IgG2, 10 mg/L for IgG3, 5 mg/L for IgG4) prevented assessment of interpretable subclass repartition when the total IgG concentration in the cryoprecipitate is less than 300 mg/L. The proportion of IgG subclass was calculated by dividing the concentration of each IgG subclass by the concentration of total IgG in each sample. Mean proportions of IgG subclasses in normal adult serum are IgG1 60%, IgG2 32%, IgG3 4%, and IgG4 4% [2].
RF activity of CG (IgM anti-IgG) was measured in the cryoprecipitate by immunonephelometry (normal < 10 IU/ mL, BN ProSpec®, Siemens). Complement fraction C4 concentration was measured with BN ProSpec® (Siemens) in serums collected simultaneously with the CG testing sample. C4 normal range in the laboratory is 0.14-0.32 g/L.

Statistical analysis
For continuous variables, the results were expressed as mean ± SEM or median (range) when the distribution was not normal. Distribution normality was tested using the D'Agostino Pearson omnibus normality test. The Kruskal-Wallis test was used for the variance analysis and the Mann-Whitney test for the comparison of quantitative variables. The Student t test and Wilcoxon test were used for the comparison of quantitative parameters. Chisquared test and Fisher's exact test were used to analyse qualitative differences. P < 0.05 was considered statistically significant. Calculations were performed with GraphPad Prism version 5.01 software (La Jolla, CA, USA).

Population and IgG subclass in the cryoprecipitates
From the 1675 patients, 86 were included in this study based on inclusion criteria (cryoprecipitate > 1 mL and cryoprecipitate IgG concentration > 300 mg/L). There was no significant difference for demographical data of this cohort compared to the initial larger cohort [22]. Sex ratio in the larger cohort was 1.55 (1018F/657M) compared to 1.39 in the subclass series (50F/36M, p = 0.57). The mean age in the larger cohort was 54.0 ± 17.4 years compared to 58.7 ± 15.5 years in the subclass series (p = 0.25). All type I IgG CG were in the two cohorts. For the mixed CG, 40.8% were secondary to infectious diseases (mostly HCV) in this cohort vs. 49.5% in the larger cohort (p = 0.18); 28.9% were secondary to auto-immune diseases in this cohort vs. 25.4% in the larger cohort (p = 0.49).
Among the 86 patients included, there were 10 patients with type I IgG CG (6 IgG kappa and 4 IgG lambda), and 50 patients with type II CG (including 41 RF-positive CG) and 26 patients with type III CG (including 8 RF-positive CG) (supplementary Fig. S1). Out of these 86 patients, 100 samples were analysed (12 samples from 10 patients with type I CG, 62 samples from 50 patients with type II CG, and 26 samples from 26 patients with type III CG), because some patients had several CG detections and the cryoprecipitates have been banked. These replicates were used to test the assay reproducibility and stability of CG composition. The comparison of paired values did not show any difference for IgG1 (p = 0.7), IgG2 (p = 0.7), IgG3 (p = 0.2), and IgG4 (p = 0.6).
Among the 50 patients with type II CG, 46 CG were monoclonal IgM (IgM kappa and IgM lambda) associated with polyclonal IgG, and 4 CG were monoclonal IgG (3 IgG kappa and 1 IgG lambda) associated with polyclonal Ig (IgG, IgA, or IgM). All 26 type III CG samples were polyclonal IgG by definition.
The proportion of CG types in the four subclasses showed that IgG1 were found in all types of CG (p = 0.25), IgG2 were more frequent in type III CG vs. type II and type I CG (p < 0.001), IgG3 were more frequent in type I CG vs. type III and type II CG (p < 0.001), and IgG4 were more frequent in type II CG vs. type III CG and never in type I CG (p < 0.0001).

IgG subclasses in type I cryoglobulins
Out of the 10 patients with type I IgG CG, IgG subclasses were 5 IgG1 (2 IgG1 kappa and 3 IgG1 lambda), 2 IgG2 (1 IgG2 lambda and 1 IgG2 kappa for 84% associated for 16% with an IgG3), and 2 IgG3 (1 IgG3 kappa and 1 IgG3 lambda for 82.2% associated for 17.8% with an IgG2). One patient had a biclonal type I CG, with the same proportion of IgG1 and IgG3. No type I CG contained IgG4 (Table 1).

IgG subclass and RF activity
For the mixed CG, there were 26 RF-negative CG (9 type II, 17 type III) and 50 RF-positive CG (41 type II, 9 type III). IgG subclass proportions in RF-negative CG compared to RF-positive CG were for IgG1 57. 6 Fig. S3).

IgG subclasses and clinical manifestations of type I CG
Patients with type I IgG CG were all symptomatic. Among the 6 patients with IgG1, 4 presented cutaneous signs associated with neurological (3/4), articular (1/4), and/or renal (3/4) signs, and 2 had no cutaneous signs but neurological, renal, and/or articular manifestations. The 4 patients with IgG2 and/or IgG3 had cutaneous manifestations only (3/4), and one with renal manifestation (proteinuria) ( Table 1).

Discussion
Type I IgG CG were mostly IgG1, in relation with cutaneous, renal, and neurologic manifestations. Some IgG2 and IgG3 type I CG were observed, associated with cutaneous or renal manifestations. In mixed CG, IgG1 were more frequent in type II CG, associated with RF-positive CG, and IgG2 and IgG3 in type III CG. A higher proportion of IgG4 was associated with RF-positive CG and a low level of C4. IgG2 concentration was lower in the mixed CG of symptomatic patients. For type I IgG CG, few studies have reported on the determination of IgG subclasses and their relation to the associated clinical manifestations [7][8][9][10]13]. In this study, IgG1 was the most common cryoprecipitating type I IgG and no monoclonal IgG4 was found, as reported previously [7,10]. The distribution of monoclonal IgG subclasses was different between cryoprecipitating IgG and normal IgG, with more IgG2 and IgG3 precipitating in the cold and rarely, if any, IgG4 [23,24]. Monoclonal IgG1 and IgG3 with their high capacity to aggregate are good candidates for cryoprecipitation. In contrast, cryoprecipitation of IgG4 seems more difficult due to its peculiar conformation and lower capacity to self-aggregate and to form large size complexes [2,3,25].
An association of two subclasses in type I CG was reported in this series: monoclonal IgG1 and IgG3 in the same proportion for one patient, and IgG2/IgG3 association for two patients. The association of more than one IgG subclass within type I CG and normal monoclonal IgG could result from one plasma cell clone producing several gamma heavy chains associated with the same light chain [11] or, that more than one pathologic plasma cell clones were involved in monoclonal gammopathies [24]. Another hypothesis for the association with monoclonal IgG3 was the RF activity of IgG3 recognising IgG Fc (IgG3 RF anti-IgG) leading to the formation of larger size complexes when temperature decreases [12,20].
Regarding the link to clinical manifestations in this study of type I CG, IgG1 was associated with multiple vasculitis manifestations (cutaneous, neurological, rheumatological, and/or renal), and IgG2 and IgG3 were associated with only one type of manifestation (cutaneous or renal). Only few clinical cases were previously reported concerning patients with type I IgG CG [8,9,13], not allowing a conclusion on a relationship between IgG subclass and clinical manifestations. Cutaneous manifestations were the main clinical signs reported in this series; nephrotoxicity was rarely observed, only in patients with IgG1 as reported in another case [13]. In the autoimmune mouse MRL model and in humans, IgG3 CG was described as nephrotoxic [9, 14-17, 26, 27]. Mouse IgG1 was reported as protecting against renal disease in a mouse model of cryoglobulinemia, unlike IgG3 [28]. Mouse IgG3 can selfassociate to form large immune complexes which could precipitate in glomeruli capillaries [28]. These mouse models cannot be transferred to humans, because mouse IgG1 resembles human IgG4, not involved in type I CG pathophysiology. The involvement of one specific IgG subclass in renal manifestations in humans is not clear because renal manifestations appear several years after CG diagnosis [29]. This delay could explain differences in the reported results and justify a long follow-up study of patients.
Regarding mixed CG, IgG1 was the main subclass especially in type II CG, IgG2, and IgG3 were overrepresented in type III CG, and IgG4 was less represented. In studies reporting on mixed CG, all IgG subclasses were found with similar distribution as reported here, especially in RF-positive CG [16,27,28].
The significant association of IgG2 and IgG3 in mixed CG observed in this study may be linked to RF activity of IgG3 (IgG3 anti-IgG2), responsible for immune complex formation at low temperature [12,14,20,26]. In this series, IgG3 were found in higher proportion and concentration in RF-negative mixed CG. This higher representation of IgG3 was not in relation with their association with RF IgM as previously reported [18,30,31], but could be due to their greater ability to form spontaneous aggregates and to cryoprecipitate.
There was a trend for a higher concentration of total IgG in mixed CG in asymptomatic patients, compared to symptomatic ones that could be explained by an earlier diagnosis of CG in symptomatic patients. However, in patients with cutaneous and neurological manifestations, IgG2 was found less represented in mixed CG than in asymptomatic patients. Regarding IgG3, no direct relationship to a clinical manifestation was found herein. However, in the literature in the mouse MRL model [14,15,17] and humans [9,16,17], IgG3 has been linked with renal and cutaneous manifestations of CG. In another model of human mixed IgM-IgG CG transfer in Balbc mice, the crucial role of in situ deposits of CG complexes has been linked to the development of CG glomerulonephritis, independently of RF activity or IgG subclass composition [21]. Clinical manifestation of human mixed CG did not clearly appear to be related to their IgG subclass composition.
Surprisingly, in mixed CG, higher proportion and concentration of IgG4 were associated with RF-positive CG and low C4 concentration, in favour of complement system activation. IgG4 subclass cannot activate complement classical pathway, unlike IgG1, because mutations in the γ4 CH2 domain inhibit the binding of C1q [2]. In this study, IgG4 concentration was significantly associated with that of IgG1 in CG immune complexes, which favour complement activation.