Introduction
Children represent a group of patients with low levels of sensitisation against human leukocyte antigens (HLA) as they have generally not been previously exposed to multiple blood products, pregnancies or previous transplants [
1]. However, they do have a more naive immune compartment and are prone to developing infections which carry a small risk of cross-reaction with the allograft through heterologous immunity [
2]. Children face lifelong immunosuppression and potential multiple re-transplants. Therefore, finding the correct balance between the suppression of alloimmunity and the side effects of immunosuppression is even more important [
3].
We previously published the largest cohort of paediatric renal transplant recipients screened prospectively for de novo donor-specific HLA antibodies (DSA) [
4]. DSA-positive patients were found to have a faster decline in allograft function and more features of antibody-mediated rejection (AMR) on biopsies done ‘for-cause’. Also, the level of allograft dysfunction correlated with rising mean fluorescence intensity (MFI) levels for Class II DSA. In the study reported here, we investigated further the capability of DSA to activate the complement cascade through in vitro assays detecting complement binding at the levels of C1q (first subcomponent of the C1 complex of the classical pathway of complement activation) and C3d (subcomponent of complement component 3) [
5,
6]. We hypothesised that patients with complement-activating DSA would have poorer renal allograft outcomes.
Discussion
We investigated the utility of complement binding assays to further stratify DSA+ patients at risk of worse renal allograft outcomes. In vitro, a larger proportion of DSA fixed C1q compared to C3d (49 vs. 35%). Complement positivity correlated poorly with IgG MFI, and a clear threshold could not be defined. C1q+ DSA were associated with a higher risk of tubulitis, but the long-term renal allograft function of these patients was not significantly different to those with C1q− DSA. C3d+ DSA were associated with more C4d staining on ‘for-cause’ biopsies and significantly worse renal allograft function.
We hypothesised that patients producing complement-fixing DSA would have a worse allograft outcome because of the implication that in vivo, binding to the allograft endothelium would enable recruitment of additional immune activation pathways through activation of the complement cascade. The ability of antibody to activate the complement cascade depends on the IgG subtype and corresponding Fc portion [
10], in addition to the target antigen density and proximity of multiple target HLA molecules to allow cross-linking and subsequent binding of C1q [
10]. This in turn triggers the classical complement pathway through a series of regulated intermediate steps, leading to the formation of C4bC2b, which catalyses the breakdown of C3 into its active components [
11]. C3 is the converging point of three complement pathways, ultimately leading to the generation of C5b-9, the membrane attack complex (MAC) and cell lysis. The breakdown products of C4b and C3 include C4d and C3d, respectively, which covalently bind to cell membranes [
11]. Therefore, in vitro tests have been designed to probe the complement binding capabilities of DSA at the check-points C1q, C4d and C3d [
12]. Fixing of C1q is a pre-requisite for initiation of the complement cascade. C3d itself is dissociated from IgG, i.e. C3d is not bound to antibody but rather is attached to cell membranes or complement receptors. Interestingly, we showed that 41% of antibodies positive for C1q binding do not concomitantly fix C3d, despite agreement between the pan-IgG kits. Whether this is a specific difference in functional characteristics of the antibody detected, a reflection of antibody titre or a difference inherent in the assays needs addressing in the future.
This study focused on DSA+ patients initially identified using the pan-IgG LABScreen SAB. We therefore cannot rule out patients who might have tested DSA IgG+ using the alternative Immucor SAB assay. However, our laboratory experience and a review of the literature suggests agreement rates of 90% [
13]. In our subset of C1q−/C3d+ patients, there was only 50% agreement between kits, i.e. additional IgG+ patients were identified on the Immucor SAB. However, long-term renal allograft function was not worse. Therefore, the discrepancy in results could be due to kit-related factors, such as the presence of denatured HLA antigen or the detection of potentially non-significant antibody [
14]. In addition, complement-binding DSA tended to be detected later post-transplant, in agreement with previous studies showing that patients with early onset DSA and AMR had better responses to treatment and favourable clinical outcomes [
15‐
17].
This study is limited by its retrospective nature and by the relatively small number of patients, which precluded further multi-variable survival analysis. In addition, rejection and reduced allograft function was more prevalent among the group of patients for whom sera were unavailable for testing, whom we hypothesised would have complement-binding DSA. With these caveats, our data do suggest an association between C1q+ DSA and increased episodes of acute tubulitis. This was also reported by Lefaucheur et al. using a non-supervised principal component analysis examining IgG subclasses and C1q [
18]. C4d staining in vivo correlated with C3d testing in vitro which is downstream of C4 in the complement cascade. In vivo C3d staining has rarely been reported in the literature. One histological study of biopsies during acute allograft rejection showed C3d peritubular capillary staining in 30% of samples and increased renal allograft loss in the C3d+ patients [
19]. MAC deposition has been shown in patients with HLA incompatible transplants, but its role in non-sensitised de novo AMR has not been studied [
20]. MAC deposition results in cell lysis, although sublytic concentrations also can induce pro-inflammatory changes in glomerular cells and endothelial mesenchymal transition of tubular cells [
21,
22].
The results of our study are in concordance with those of a recently published study in paediatric renal transplant recipients that also concomitantly assessed C1q and C3d binding [
23]. Both studies showed a lower proportion of C3d binding and better prognostic predictability with the C3d assay. Comoli et al. also showed that patients could progress from C1q−/C3d− to C1q+/C3d+ and from C1q+/C3d− to C1q+/C3d+. Some patients were intermittently positive for complement binding associated with a low MFI of <2000 [
23]. The current studies highlight the complexities of assessing DSA. MFI has often been used in studies as a quantitative measure of DSA although there is significant inter-assay variability and the assay is not licenced clinically as a quantitative measure [
13]. It is also subject to the prozone effect which can give an artificially low MFI [
24]. This can be overcome by dilutional titering. although this adds additional time and cost factors. In addition, studies have shown an association between C1q binding, IgG subtypes and IgG MFI, thus limiting the extra information obtainable in performing all three assays [
25]. In our study, sera were obtained prospectively as per guidelines regardless of renal allograft function; as compared to studies which were done at the time of graft dysfunction and biopsy ‘for cause’ [
6,
26]. In addition, our patients had low pre-transplant HLA antibody sensitisation rates which are not comparable to those of adult studies which include highly-sensitised and HLA-incompatible transplants [
4,
5,
17]. We showed that the C3d assay potentially further stratified patients at the highest risk of renal allograft failure. This is independent of IgG MFI as the correlation between C3d and IgG MFI was poor (adjusted
R
2 0.11), with a significant overlap of C3d+ and C3d– patients in the moderate MFI range of between 1000 and 8000. Nonetheless, the results would be strengthened by being validated in a prospective study.
In conclusion, our study adds to the evidence of the potential importance of determining complement binding capabilities when testing for de novo DSA. Of the DSA we detected 49% could bind C1q, and its presence was associated with an increased proportion of ‘for-cause’ biopsies showing acute tubulitis, but not with worse long-term outcome. 35% of the DSA fixed C3d, and these DSA were associated with an increased proportion of ‘for-cause’ biopsies demonstrating positive C4d histological staining and significantly worse long-term renal allograft dysfunction. With the increasing financial pressures on healthcare provision, along with the significant costs of performing these tests, we believe these data may aid the decision-making behind the choice of tests used for post-transplant DSA monitoring.
Compliance with ethical standards