Anne Els Van De Logt

assays, either chromogenic (bromocresol green (BCG), bromocresol purple (BCP)) or immunonephelometric. Differences between these assays have received limited attention thus far. In chapter 3 we evaluated bias and imprecision of BCG and BCP assays in comparison to the immunonephelometric assays using blood samples from patients with pMN and nephrotic syndrome. For BCG, mean bias was high (6.2 g/l, SD 2.4 g/l), for BCP mean bias was low (0.3 g/l, SD 1.5 g/l). Importantly, we illustrated that these differences are clinically relevant by evaluating the agreement of the decision toward the use of prophylactic anticoagulant therapy. Our study showed that nephrologists may reach different treatment decisions in up to 59% of patients. Unfortunately (and with current knowledge inexcusably) assay methods are often not reported in studies. Until standardization of the albumin assay is achieved, nephrologists/clinicians should realize that large discrepancies between assays exist and should be informed which assay is used in their hospital. Preferably, in patients with low albumin concentration the immunological method is the method of choice. When not available, the BCP method is more accurate compared to the BCG methods for the measurement of albumin at low concentration in patients with MN. In the new guideline it is recommended to use a 5g/l higher cut-off for serum albumin when using the BCG assay (3). We have extended the above studies in patients with chronic kidney disease (CKD). Preliminary analysis showed that the uncertainty of the target value when using the BCG assay was dependent on the assay platform/manufacturer with bias between BCG and Immunonephelometric assay ranging from 0 to 4 g/l. Most CKD patients had albumin values above 30 g/l in the BCP assay. In future research we will study the bias of different BCG platforms/manufacturers in nephrotic syndrome, and we expect that these differences will be even higher. In retrospect, the data obtained in chapter 3 were obtained with BCG assays using Beckton Dickinson. The guideline advise to subtract 5g/L when using BCG therefore is not correct and should be rephrased e.g. “when using BCG assay for serum albumin, cut-off values, used in definitions or treatment decisions, should be adapted taking into account the bias of the assay, which is dependent on the manufacturer, and may range from 0 to at least 7 g/L “.

Anti-PLA2R antibodies
We have studied the use of aPLA2Rab as a prognostic biomarker in pMN in chapter 4. Personalized treatment for pMN patients requires accurate prediction of the disease course at an early stage, before the start of immunosuppressive treatment. The current models (Toronto risk score, measurement of low-molecular-weight proteins) lack sufficient accuracy, with an AUC of 0.75-0.80. Our group previously showed that circulating aPLA2Rab correlated with clinical disease activity (8). However, the studied cohort was small and only univariate analysis was performed. Additional studies suggested the value of aPLA2Rab levels as prognostic marker, since low titers of aPLA2Rab were associated with a higher likelihood of remission (9-12). However, in these studies remission and renal progression were evaluated in mixed cohorts of treated and untreated patients. For the first time, we evaluated the prognostic value of aPLA2Rab titers at baseline in untreated patients. In our cohort of 156 patients, in univariate analysis, high aPLA2Rab levels were specific (>80 %) for predicting progression, with positive predictive values of 77 % for levels of 150-300 RU/ml and 87 % for antibody levels >300 RU/ml. However, we demonstrated that aPLA2Rab titer had no added value in a multivariate model over traditional biomarkers (serum creatinine, proteinuria) to predict progression in pMN. Although our data thus question the independent prognostic value of aPLA2Rab levels, measured once within few months after diagnosis, we suggest that future research should evaluate if repeated aPLA2Rab measurements during 6 months follow-up increase prognostic value.

There is some evidence from clinical trials that aPLA2Rab have predictive value and thus would allow to predict response to immunosuppressive therapy (10, 13). These two studies, both using rituximab therapy, analyzed aPLA2Rab levels at start of therapy and showed that non-response was particularly high in patients in the highest tertile of aPLA2Rab levels.

A Spanish group evaluated the predictive value of changes in the aPLA2Rab levels in patients with pMN treated with other immunosuppressive therapy (two different treatment regimens; CP and steroids according to Ponticelli protocol, and combination therapy of tacrolimus and rituximab). During treatment with both regimens the relative reduction of aPLA2Rab at resp. 3 and 6 months could be used to predict clinical remission after 6 and 12 months (e.g. respectively 44 and 48 % aPLA2Rab reduction at 3 months could predict remission at 6 months with an AUC of respectively 0.91 and 0.90)(14). However, no information is included whether changes in aPLA2Rab titer predict final outcome.

Toward individualized treatment of patients with membranous nephropathy

In chapter 5 of this thesis we discussed the different treatment options (up to 2016).