Refining Our Understanding of the PLA2R-Antibody Response in Primary Membranous Nephropathy: Looking Forward, Looking Back

Since the M-type phospholipase A2 receptor (PLA2R) was described as the major autoantigen in membranous nephropathy (MN) a decade ago, many studies have advanced the clinical utility of circulating autoantibodies to PLA2R (PLA2R-ab) in primary MN and started to disclose its genetic and pathophysiologic underpinnings. With each step forward, new questions arise that open new avenues for research into this fascinating autoimmune disease.

Two such articles are published in this month’s issue of JASN. The first, by Reinhard et al.,¹ describes a novel fourth epitope-containing region within PLA2R and challenges the notion that epitope spreading independently associates with worsened clinical prognosis. The second, by Burbelo et al.,² provides a glimpse of the natural history of PLA2R-ab before clinical diagnosis of MN by analyzing Department of Defense Serum Repository (DoDSR) samples from military recruits.

Let us first revisit the discovery of PLA2R epitopes. PLA2R has ten extracellular domains: an N-terminal cysteine-rich domain (CysR, also called a Ricin B domain), a fibronectin type 2 domain, and eight C-type lectin-like domains (CTLD1–8). These domains fold into a compact structure that is modulated by pH and likely regulates ligand binding and release.³ Fresquet et al. identified an immunodominant, 31-amino acid conformational epitope in CysR⁴; antibodies to this region (or at least to the larger CysR domain) are found in all patients with circulating PLA2R-ab and have a very high affinity for their target. Others had found that the immunodominant epitope is within a PLA2R fragment that includes CysR through CTLD1.⁵ Seitz-Polski et al.⁶ subsequently demonstrated that CTLD1 contains an epitope distinct from that in CysR, and reported a third epitope within CTLD7. They attributed the variable PLA2R-ab specificities to spreading beyond the immunodominant CysR epitope, as demonstrated in the Heymann nephritis model in rats immunized with an N-terminal epitope of megalin.⁷ They also found that the severity of MN was worse in patients in whom antibody specificity extended beyond CysR (so-called “spreaders”).⁶ Whereas patients with only anti-CysR (one-third of the cohort) were more likely to remit spontaneously, those with spreading to CTLD1 and/or CTLD7 had more proteinuria at last follow-up and were more apt to progress to ESKD. Notably, spreaders also tended to have higher overall PLA2R-ab titers. Multivariate analysis (within a small additional cohort of patients who were PLA2R-ab positive drawn from the GEMRITUX study, and adjusting for age, sex, treatment group, and PLA2R-ab) showed that epitope spreading correlated with fewer clinical remissions at 6 months and last follow-up independently of overall PLA2R-ab titer.⁸ In a more recent study of cohorts combined from GEMRITUX and NICE, epitope spreading at diagnosis was again found to be an independent risk factor for lack of clinical remission, despite a very strong correlation between spreading and overall titer.⁹

Enter the current article by Reinhard and colleagues¹ to the fray. These authors identified 150 patients seropositive for PLA2R-ab whose initial samples were obtained within 6 months of diagnostic kidney biopsy and followed them for clinical remission with a median follow-up of 54 months. To identify the epitope-containing domains, they used the same PLA2R deletion-construct strategy as did Seitz-Polski et al.,⁶ using a Western blot assay with a standardized 1:100 serum dilution and a new ELISA for the epitope-containing domains. The authors make several new assertions and challenge the independent role of epitope spreading as a prognostic indicator. First, they identify a proportion of subjects (16% by Western blot and 10% by ELISA) whose sera appear to recognize an epitope within CTLD8, thereby defining a fourth epitope within PLA2R. In this cohort, 67% of subjects exhibited a Western-blot pattern suggestive of reactivity to CTLD7, bringing the total of those with recognition of either of these C-terminal domains to 83% of the entire cohort. When the remaining patients were reanalyzed at a serum titer of 1:25, they too exhibited apparently specific reactivity to these C-terminal epitopes, because control sera did not yield a signal at this higher concentration. Because all 150 patients additionally had evidence of antibodies to the N-terminal (CysR or CTLD1) epitopes, these authors argue that all patients already had broad reactivity to PLA2R by the time of clinical diagnosis.

Moreover, those subjects who exclusively exhibited an N-terminal pattern at the standard Western blot dilution of 1:100 (17%) also had lower median total PLA2R-ab titers (19 RU/ml) compared with those who exhibited C-terminal patterns (median titer of 175 RU/ml). This is not inconsistent with the previous work from the French studies. However, in this analysis using this larger prospective cohort followed for a median of 54 months, total PLA2R-ab level was predictive of clinical remission rather than the presence or titer of the additional epitope specificities.

Although it is conceptually simple to define epitope spreading, measuring discrete specificities in a clinically meaningful and standardized manner is technically challenging due to the variable sensitivities of the detection assays. This may account for the apparent discrepant conclusions of the Seitz-Polski et al. and Reinhard et al. studies. The conformation or accessibility of each potential epitope may be altered by differences in construct design, expression system, protein stability, denaturing conditions (e.g., Western blot), and manner in which the protein is bound to substrate. Increasing the sensitivity of the assay may also increase the risk of detecting irrelevant or crossreactive epitopes. Although there are some data that the CTLD specificities are discrete,⁶ crossreactivity of a single anti-CTLD–directed autoantibody with epitopes in other CTLDs might further confound assessment of the humoral repertoire. It should be noted that CTLD7, recently resolved at the crystal structure level,¹⁰ is most similar by sequence to CTLD8; Reinhard and colleagues did not attempt to determine through competition assays whether the anti-CTLD7 and anti-CTLD8 specificities were discrete.

The technical possibility of crossreactivity notwithstanding, it is likely that there are indeed multiple distinct PLA2R-ab specificities, implying the existence of distinct pools of B cells, which likely evolved over time with maturation of the humoral immune response. Longer durations of disease would allow for epitope spreading, oligoclonal expansion, and affinity maturation of these separate pools of autoantibody-producing B cells, all of which should translate into higher overall PLA2R-ab titers. High titers in turn are known from several studies to be associated with less responsiveness to treatment and worse clinical outcome.

Natural history studies of MN, now enriched with data from studies that incorporate circulating PLA2R-ab and tissue staining for PLA2R, suggest that disease typically plays out over years and that kidney biopsy is performed relatively late in the process. To date, PLA2R-ab studies (as discussed above) have assessed baseline levels near the time of biopsy and looked for correlations in future clinical outcomes. What if researchers were to look for autoantibodies in the other direction, before the biopsy diagnosis of MN?

Such a study was done by Burbelo et al.,² using banked DoDSR samples from military personnel. The investigators retrospectively identified 134 cases with at least one serum sample available before biopsy-confirmed primary MN and asked whether pre-existing PLA2R-ab could be detected at any point before clinical diagnosis. They used a luciferase immunoprecipitation system comprising the N-terminal half of PLA2R (including CysR and CTLD1), which they had previously shown to be sensitive and specific for PLA2R-ab,¹¹ and detected PLA2R-ab in at least one sample in 44% of the MN cases (and in none of the longitudinal samples from the 134 matched controls). This overall prevalence is lower than the typical 70%–80% prevalence rate for PLA2R-ab in primary MN; however, sera were generally not available at the time of kidney biopsy. In fact, 30% of cases did not have an available sample within 2 years of the biopsy; when these cases were excluded, the prevalence was 63%.

The authors make a heroic attempt to assemble the piecemeal clinical data points available to them, limited both by the sporadic nature of the samples and strict DoDSR privacy considerations, into a rational model of disease activity before clinical diagnosis. The four-sample-per-subject maximum that was allowed by DoDSR resulted in data points often years apart and/or quite distant from the time of diagnosis. Moreover, proteinuria data were limited and thus serum-albumin measurements of the available samples served as surrogates for the nephrotic state. Despite these hurdles, the authors show evidence that PLA2R-ab often first appears within the year before diagnosis; the median time from first positive sample to biopsy was 274 days. In one sense, this result is not surprising. We know that in recurrent MN it may take as long as a year for clinical disease to manifest when an allograft is transplanted into a recipient with circulating PLA2R-ab.

The finding that PLA2R-ab was occasionally found decades before initial biopsy diagnosis was unexpected. Correlation with other data points such as hypoalbuminemia and episodes of subnephrotic proteinuria suggest that these remote periods of seropositivity might have represented subclinical flares of disease activity that resolved spontaneously before a full clinical presentation years later. Other possibilities, such as the presence of nonpathogenic antibodies or smoldering disease activity can also be imagined, but in the absence of more granular data—especially relating to proteinuria—we can only speculate.

What is the overall significance of these findings? Given the rarity of MN and its often-slow progression once detected, it is clearly impractical to screen all adults for PLA2R-ab. And, if one were to find it in a patient who was asymptomatic, other than following the patient for the development of proteinuria, one would not “treat the test” anyway. A suggestion is made by the authors that patients with otherwise unexplained hypoalbuminemia or subnephrotic proteinuria should be investigated for PLA2R-ab. This is echoed by retrospective evidence by Hoxha et al.¹² that such patients (with non-nephrotic proteinuria) who are seropositive often progress to full nephrotic syndrome and require treatment, whereas those that are negative have spontaneous remission of proteinuria.

The takeaway messages from this pair of studies are twofold. From a practical and clinical standpoint, because there are no readily available epitope-specific immunoassays, and antibody titer can serve as a surrogate for epitope diversity, clinicians can be guided by antibody titer in their decision making. At the same time, research should continue to trace the antibody repertoire, including epitope specificity, back to its origins. One can envision using the DoDSR or similar biorepository with more granular samples and data points to investigate the evolution of the humoral response and autoantibody repertoire to the point of its earliest detection, much as deeper space probes and knowledge of the expanding universe are bringing us closer to the “big bang.”

All in all, these two studies are welcome additions to the literature and will surely not be the last in our ever-evolving understanding of the humoral response in primary MN. Our slow-but-steady accumulation of knowledge about pathogenesis, prognosis, and treatment in MN has always involved a Janus-like approach of simultaneously looking forward in new directions and also backward to pioneering studies in the field. Now, our study of the evolution and characteristics of PLA2R-ab can be described as doing the same, looking forward and looking back.

Disclosures

Dr. Beck Jr. reports grants from Sanofi Genzyme and Pfizer CTI; personal fees for consulting/advisory board from Achillion Pharmaceuticals, Genentech and Visterra. In addition, Dr. Beck Jr. has a patent Diagnostics for membranous nephropathy with royalties income through Boston University and royalty payments from UpToDate, Inc. relating to three topic cards on the subject of membranous nephropathy. Dr. Salant reports grants from Sanofi Genzyme and Pfizer CTI; consulting and advisory board fees from Advance Medical, Chugai, Pfizer and Visterra. In addition, Dr. Salant has a patent Diagnostics for membranous nephropathy with royalties income through Boston University and royalty payments from UpToDate, Inc. relating to two topic cards on the subject of membranous nephropathy.

Funding

This work was supported by discretionary research funds from the Department of Medicine at Boston Medical Center, Boston, MA, and a research training grant from the National Institutes of Health (T32 DK007053).