Table 2.

MicroRNAs in CSF of human patients with MS

Authors, country	Number of patients, gender, ages	Comparison	Changes in miRNAs in DR patients	Functional outcomes	Conclusion
Quintana et al., 2017; Spain	Discovery study: RRMS patients, 0 M/9 F, 4 LS_OCMB-/ 5 LS_OCMB+, median age 37 yrs. Validation study: 47 MS patients LS_OCMB-, 17 M/30 F, median age 32 yrs, presentation monofocal42/polyfocal2; 39 MS patients LS_OCMB+, 13 M/26 F, median age 34 yrs, presentation monofocal 35/polyfocal 4. All MS patients presented McDonald 2010 RRMS form and none was being treated with corticosteroids, immunosuppressants or immunomodulators. CSF was collected and centrifuged (400 × g, 15 minutes, 19oC) and then stored at –80oC until analysis. All MS patients were recruited at the beginning of the disease (most during the first year). Gadolinium-positive lesions were present in 57% of the patients.	Discovery study: 5 other neurological diseases (OND) controls, 0 M/5 F, median age 38 yrs. Validation study: 55 OND controls, 21 M/34 F, median age 36 yrs.	In the comparison of MS patients and OND controls, no differences were found for gender distribution or age in discovery cohort or in validation cohort. Similarly when separated into MS LS_OCMB-, MS LS_OCMB+, OND, no differences were found for gender distribution or age. Clinical and radiological data collected for MS patients from MRI performed within a mean of 3 months did not show any differences between LS_OCMB- and LS_OCMB+, except for onset symptoms where 50% of LS_OCMB+ patients presented medullar symptoms. In the discoverv study using RT-PCR, 62 miRNAs were detected in at least 75% of individuals. Of these miRNAs, 12 had significantly different expression for MS vs. controls: miR-203, miR-365, miR-21, miR-520c-3p, miR-191, miR-328, miR-30a-5p (upregulated in MS) and miR-140, miR-126, miR-199a-3p, miR-143, miR-19a (downregulated in MS). Greatest fold change values for upregulated miRNAs were for miR-30a-5p FC = 8.0 and miR-520c-3p FC = 6.4, while for downregulated miRNAs were for miR-140 FC = –3.8 and miR-19a FC = –3.4. In the validation study using RT-PCR, statistically significant differences in expression for MS vs. OND control were found for miR-328, miR-30a-5p, miR-150, miR-645 (upregulated in MS) and miR-365, miR-21, miR-191, miR-199a-3p, miR-106a, miR-146a (downregulated in MS). Four miRNAs showed significantly different expression in LS_OCMB+ vs. OND controls: miR-30a-5p, miR-150, miR-645 (upregulated in LS_OCMB+) and miR-191 (downregulated in LS_OCMB+). Two miRNAs had significantly different expression in LS_OCMB- vs. OND controls: miR-199a-3p and miR-106a (downregulated in LS_OCMB-). No significant differences were found for any miRNA between LS_OCMB+ and LS_OCMB-.	By ROC analysis, miR-150 had the greatest AUC value (0.684) for distinguishing MS patients from OND.	Upregulated expression of miR-150 was found for MS patients compared to OND controls and also for LS_OCMB+ compared to LS_OCMB-.
Bergman et al., 2016; Sweden	Validation cohort 1: 34 CIS patients, 10 M/24 F, median age 36 yrs, median disease duration from onset 1 yr, EDSS 2, percent sampled during relapse 20, percentage oligoclonal bands (OCB) positive 45, median CSF mononuclear cells 2.5; 43 MS patients, 17 M/26 F, median age 37 yrs, median disease duration from onset 7 yrs, EDSS 2.5, percent sampled during relapse 26, percentage OCB positive 88, median CSF mononuclear cells 6. Validation cohort 2: 96 CIS patients, 25 M/71 F, median age 32 yrs, median disease duration from onset 0 yr, EDSS 1, percent sampled during relapse 26, percentage OCB positive 71, median CSF mononuclear cells 4, percent converted CIS 63; 120 MS patients, 38 M/82 F, median age 37 yrs, median disease duration from onset 6 yrs, EDSS 2, percent sampled during relapse 31, percentage OCB positive 82, median CSF mononuclear cells 4. CSF samples were collected by lumbar puncture, centrifuged (440 × g, 10 minutes, room temperature) to separate cells and large particles from the CSF supernatant.	Validation cohort 1: 34 NINDC (noninflammatory neurologic disease controls) 10 M/24 F, median age 35 yrs, percentage OCB positive 0, median CSF mononuclear cells 2; 31 I inflammatory neurologic disease controls (INDC), 10 M/21 F, median age 45 yrs, percentage OCB positive 9, median CSF mononuclear cells 2. Validation cohort 2: 119 NINDC, 37M/82F, median age 38 yrs, median disease duration from onset 2 yrs, EDSS 1, percentage OCB positive 0, median CSF mononuclear cells 1; 95 INDC, 30 M/65 F, median age 40, median disease duration from onset 2 yrs, EDSS 1, percentage OCB positive 13, median CSF mononuclear cells 2.	By qRT-PCR, 15 miRNAs were selected from the analysis of pooled CIS, MS, NINDC and INDC cell-free CSF samples for examination in validation cohort 1. Of the tested miRNAs, only miR-145 and miR-150 were significantly different for MS compared to NINDC control. In the larger validation cohort 2, it was possible to replicate significantly higher levels of miR-150 in MS compared to both NINDC and INDC controls as well as being significantly higher between CIS and NINDC. Also a significantly higher level of miR-145 in MS compared to NINDC was found. Significantly higher miR-150 levels were found in CSF from patients with CIS who subsequently converted to MS compared to those who did not convert during follow-up (median period of 52 months). High levels of miR-150 correlated with higher CSF cell numbers and higher IgG index, indicating that miR-150 associated with active inflammation. In contrast, miR-150 levels did not correlate with the number of MRI T2 lesions and EDSS score, and there was only a tendency for higher miR-150 levels in relapse.	By ROC analysis, miR-150 level in CSF distinguished MS patients in relapse/remission from controls with AUC value 0.744 (sensitivity 0.890, specificity 0.500). In validation cohort 1, the ratio of miR-150 to miR-204 gave the largest difference between MS and NINDC. Also the ratio of miR-150 to miR-204 was significantly higher in MS compared to NINDC and INDC in validation cohort 2, as well as in patients with CIS who subsequently converted to MS compared to those who did not convert. By ROC analysis, the ratio of miR-150 to miR-204 had an AUC value of 0.811 (sensitivity 0.710, specificity 0.790) for differentiating MS from NINDC. In this respect, miR-150/miR-204 ratio performed as well as the best protein biomarker CXCL13. The miR-150/miR-204 ratio also differentiated patients with CIS who converted to MS compared to those who did not convert with an AUC value of 0.775.	CSF miR-150 is a biomarker for MS patients and for CIS patients who subsequently convert to MS.
Ahlbrecht et al., 2016; Germany	28 CIS-CIS patients, 7 M/21 F, 37.6 ± 11.5 yrs, EDSS 2.0 ± 1.1, onset of DMT after CIS diagnosis 35.7%, Qalb 5.2 ± 2.1, dysfunction of blood/CSF barrier 32.1%; 30 CIS-RRMS patients, 7 M/23 F, 31.1 ± 9.7 yrs, median time to RRMS 156 days, EDSS 2.4 ± 1.1, onset of DMT after CIS diagnosis 36.7%, Qalb 5.0 ± 1.7, dysfunction of blood/CSF barrier 23.3%. Inclusion criteria included CIS patients who underwent lumbar puncture and brain MRI scan at the time of CIS diagnosis whose residual CSF and serum samples were stored at –80oC and who had a documented CIS disease status 1 yr after CIS diagnosis or converted to RRMS within 1 yr. CSF and serum samples were collected simultaneously. CSF samples were centrifuged (170 × g, 10 minutes). Serum samples were centrifuged (1500 × g, 10 minutes). The cell-free supernatants of CSF and serum were stored frozen until the measurement of miRNAs. A dysfunction of the blood-CSF barrier was defined by CSF/serum albumin quotient (Qalb) higher than age-adjusted upper reference limit calculated as 4 + (age in years)/15 (Reiber, 1998).		By qRT-PCR, levels of miR-922 in CSF and serum were significantly higher in CIS-RRMS than CIS-CIS patients. While miR-181c levels in CSF were significantly higher in CIS-RRMS than CIS-CIS patients, there was no difference in serum levels of miR-181c between the two groups. No differences were found between CIS-CIS RRMS and CIS-CIS patients for miR-633 in both CSF and serum. Levels of the investigated miRNAs did not significantly correlate with age, EDSS, CSF cell count, total protein levels in CSF and Qalb. In univariate Cox regression analyses, lower levels of CSF miR-181c and serum miR-922 were significantly associated with a lower risk of conversion from CIS to RRMS. None of the other miRNAs were significantly associated with conversion to RRMS. Of the other baseline factors, younger age, and > 9 lesions on MRI, were significantly associated with conversion to RRMS. In the multivariate Cox regression analyses, CSF miR-181c, age, and MRI were significantly associated with conversion from CIS to RRMS, while serum miR-922 was not significantly associated with conversion to RRMS.	By ROC analysis, for CSF miR-181c the AUC value was 0.67. Combining MRI, age and CSF miR-181c resulted in an increased AUC value of 0.83 and improved values of specificity and positive predicted value for conversion to RRMS compared to CSF miR-181c alone or the combination of age and MRI (96% and 94%, respectively)	CSF miR-181c might serve as a biomarker for early conversion to RRMS.

AUC: Area under curve; CIS: clinically isolated syndrome; CSF: cerebrospinal fluid; F: females; FC: fold change; LS_OCMB: lipid-specific oligoclonal IgM bands; M: male(s); MS: Multiple sclerosis; ROC: receiver operating characteristics; RRMS: relapsing-remitting MS.