Circulating microRNAs as biomarkers for rituximab therapy, in neuromyelitis optica (NMO)

Our results demonstrate for the first time that a set of miRNAs with elevated expression in the blood of NMO patients revert to the levels of matched healthy controls following effective therapy with rituximab. The majority of these “normalized” miRNAs are known as brain-specific/enriched miRNAs. Following further validation, the results of this study might lead to the establishment of a diagnostic test to monitor treatment response in NMO.

The six brain-enriched miRNAs, miR-135a, miR-135b, miR-125b, miR-134, miR-138, and miR-760 [24‐27], that revert to healthy control levels after treatment are expressed abundantly or predominantly in the brain. The higher quantity of these miRNAs in the circulation of patients with NMO might imply that they were shed to the circulation by degenerative brain cells, with a corresponding decrease in the amount of these miRNAs in the circulation observed following treatment, when the destructive inflammatory process diminishes. Therefore, monitoring the levels of brain-specific/enriched miRNAs in the blood might reflect the degree of disease activity in the CNS of demyelinating disorders.

miRNAs are highly stable elements that can be detected in body fluids as well as in different blood cells [28]. Their stability, along with the recent development of sensitive detection and quantification methods, has positioned miRNAs as great candidate biomarkers for diagnosis of specific diseases. Moreover, miRNAs might play a major role in initiating and/or maintaining autoimmunity processes [29, 30].

For many years, NMO was considered as a severe variant of MS [1]. Both diseases have autoimmune inflammatory lesions in the CNS. In MS, there is an abundance of data on miRNA expression in whole blood, peripheral blood mononuclear cells (PBMCs), plasma, cerebrospinal fluid (CSF), and peripheral blood T cells [31‐34]. It is widely accepted that miRNAs are dysregulated in MS, but their use as biomarkers is still evolving [31]. In a recently published review on the available data regarding miRNAs in MS, Xinting et al. summarized that miR-15a, miR-19a, miR-22, miR-210, and miR-223 were upregulated in T-reg cells, plasma, blood cells, PBMCs, and brain white matter tissues from MS patients; miR-21, miR-142-3p, miR-146a, miR-146b, miR-155, and miR-326 were upregulated and miR-181c and miR-328 were downregulated in PBMCs and brain lesions; and miR-15a and miR-15b were downregulated in blood, peripheral T cells and B cells, or plasma samples from MS patients. None of the upregulated miRNAs were increased in our study; the distinct miRNA expression profiles provide further support for the concept that NMO has a distinct pathogenesis from MS.

In MS, there have been limited attempts to study the use of miRNAs in treated versus untreated patients, evaluating commonly used therapies. Waschbisch et al. studied five miRNAs (miR-20b, miR-142-3p, miR-146a, miR-155, and miR-326) by qPCR and found that there was significantly lower expression of miR-142-3p and miR-146a in glatiramer acetate (GA)-treated patients (no differences in expression were detected between untreated and IFN-β-treated patients) [35]. In an investigation of miRNA and mRNA expression in PBMCs of MS patients before and after IFN-β therapy, Hecker et al. found that IFN-β-responsive genes were upregulated in parallel to downregulation of miRNAs. Among the miRNAs, they identified altered expression among members of the miR-29 family [36]. Given the low clinical response rate of MS patients to IFN-β and GA therapy, the implications of these studies are unclear. For the more effective MS therapy, natalizumab, Sievers et al. found that 10 miRNAs, out of 1059 tested, were differentially expressed in B cells of natalizumab-treated patients versus untreated patients and Ingwersen et al. found that natalizumab therapy restored aberrant blood miRNA expression profiles in MS patients [37, 38]. The 10 most strongly upregulated miRNAs from MS lesions were expressed in astrocytes. This is noteworthy since in NMO the autoantibodies that are directed against AQP4 are expressed by astrocytes. This supports the rationale that miRNAs may contribute to pathogenesis of diseases in which astrocytes play a major role, like NMO.

In contrast to MS, in NMO, there is only one recent publication by Keller et al., which studied miRNAs in the blood of patients [9]. Their study included 11 patients with NMO, 60 patients with MS, and 43 HCs. They identified 141 differentially expressed miRNAs in NMO patients compared with HCs and 115 miRNAs in NMO versus MS patients. These results did not concur with the results from our study; not all the miRNAs that were found by Keller et al. were detected in our study and vice versa. The difference between the results of these studies may be partly attributed to the differences in patient cohorts. Our study included only untreated patients while most of the patients in the Keller et al. study were treated with immunosuppressive medications. Given the known influence of the immune system on miRNAs [39], it is conceivable that miRNA expression profiles would differ in immunosuppressant-treated and untreated patients. The use of different biostatistical method analyses and different RNAseq library preparation methods might also contribute to the different results [40]. Interestingly, one of the two RT-PCR-validated miRNAs identified by Keller et al. [9] was also highly significant in our cohort—miR-1180 (p = 0.005). Although our study demonstrated differentially expressed miRNAs in NMO patients and HCs, the main focus of our study was on differential miRNA expression after treatment. Using this analysis, we found a clear signature of miRNA expression that not only differed between NMO patients and HCs but also reverted to normal in response to rituximab therapy. Our study has its limitations and currently we cannot ascertain whether some of the above described changes in the miRNA levels following rituximab therapy are secondary to B cell depletion. It is noteworthy that the blood samples taken at the 6-month time-point were taken following analysis of positive CD19 cell detection in the blood, indicating that patients were not depleted of B cells. This supports the idea that the miRNA signature could represent a distinct phenomenon, rather than simply being a reflection of decreased B cell counts.

Rituximab is probably the most effective and commonly used therapy in NMO. Although most patients with NMO have detectable anti-AQP4 autoantibodies in the serum, the mechanism by which B cell-depleting activity mediates the beneficial effect is not clearly understood. Previous studies testing rituximab in NMO found a high response rate and most of the patients remained relapse free or experienced decreases in the annualized relapse rate (ARR) (60–100 % of disease-free patients in cohorts with 5–30 NMO patients) [22, 23].

Using 18 samples obtained from the nine responding patients in our study (two from each participant), we identified 46 miRNAs that significantly changed following rituximab therapy (p < 0.05) (Table 1). One of the significantly changed miRNA is miR-125. This miRNA has previously been described as a biomarker of rituximab therapy in B cell lymphoma and rheumatoid arthritis (RA) [41]. These findings give further validation to the authenticity of our analysis and suggest that the effect of rituximab on miR-125 levels may not be disease-specific.

Several attempts have been made to develop biomarkers in order to individualize rituximab therapy in NMO patients. CD19+ and CD27 + CD19 + B cell counts were studied by Kim et al. in NMO patients treated with rituximab [22, 42]. CD19 is expressed on B cells and CD27 on memory B cells. They reported that monitoring CD19- and CD27-positive B cell counts in the blood could help dosing the next rituximab administration. However, this assay of CD27-positive B cell suppression by rituximab is not in widespread clinical use. Recently, the same group found that, like in patients with non-Hodgkin lymphoma and RA [43], NMO patients carrying the V158F allele of the fragment c gamma receptor 3A (FCGR3A-F) have increased likelihood of incomplete B cell depletion and fivefold increase chance of relapse following rituximab treatment. This effect was more significant in patients that were not treated previously with methotrexate [43].