MSPrecise: A molecular diagnostic test for multiple sclerosis using next generation sequencing

Abstract

Background: We have previously demonstrated that cerebrospinal fluid-derived B cells from early relapsing-remitting multiple sclerosis (RRMS) patients that express a VH4 gene accumulate specific replacement mutations. These mutations can be quantified as a score that identifies such patients as having or likely to convert to RRMS. Furthermore, we showed that next generation sequencing is an efficient method for obtaining the sequencing information required by this mutation scoring tool, originally developed using the less clinically viable single-cell Sanger sequencing.

Objective: To determine the accuracy of MSPrecise, the diagnostic test that identifies the presence of the RRMS-enriched mutation pattern from patient cerebrospinal fluid B cells.

Methods: Cerebrospinal fluid cell pellets were obtained from RRMS and other neurological disease (OND) patient cohorts. VH4 gene segments were amplified, sequenced by next generation sequencing and analyzed for mutation score.

Results: The diagnostic test showed a sensitivity of 75% on the RRMS cohort and a specificity of 88% on the OND cohort. The accuracy of the test in identifying RRMS patients or patients that will develop RRMS is 84%.

Conclusion: MSPrecise exhibits good performance in identifying patients with RRMS irrespective of time with RRMS.

Keywords: B cell; Biomarker; Genetics; High-throughput nucleotide sequencing; Multiple sclerosis; Neurological disease.

Figure 1. VH4 and JH gene distributions of CSF B cells from RRMS patients are more divergent from healthy control naïve peripheral B cell repertoires than those from OND patients

VH4 (a) and JH (b) gene calls were obtained by IgBlast alignment (see methods). Total unique sequences used in cohort databases are indicated inside the pie charts. Chi-squared analysis values between cohort gene distributions are shown above the bars. Gene frequencies are shown in the table. Abbreviations: RRMS, relapsing-remitting MS; OND, other neurological disorder; HCN, healthy control naïve peripheral B cells. HCN samples are all replicates from a single patient.

Figure 2. Mutation characteristics of VH4 sequences in RRMS and OND patients

(a) Mutation frequency (MF) analysis was done by nucleotide; boxes indicate total unique sequences in each cohort and sample numbers are marked under cohort names. (b) Replacement mutation frequency (RMF) analysis was done by codon. RRMS sequence data includes 119,483 total point mutations and 62,749 total replacement mutations (RM); OND sequence data includes 74,769 total point mutations and 39,324 total replacement mutations (RM); RRMS sequence data includes 51,238 total point mutations and 17,375 total replacement mutations (RM). MF and RMF were calculated by sample and bar graphs show median (indicated on the bar graphs) and interquartile range (statistical significance of the difference between RRMS and OND was tested by Mann Whitney test). MF, RMF and R:S ratios for CDR and FR regions were calculated independently by region for each sample and are shown as cohort medians. HCN samples are all replicates from a single patient.

Figure 3. MSPrecise scores in RRMS and OND patients

Each data point represents a single sample sequence pool (median and interquartile range are marked on the figure). The dashed line represents the MSPrecise cut-off point of 6.8 above which patients are expected to have or convert to relapsing-remitting MS (RRMS). The dotted line delineate an indeterminate range (-1) below the 6.8 cut-off where the results of the MSPrecise score test are less clear cut. Samples are grouped by most current diagnosis as RRMS, other neurological diseases (OND), and healthy control naïve (HCN). Only samples that pass our filtering criteria are displayed with their calculated MSPrecise scores. Statistical significance of the difference between cohorts was calculated by Mann Whitney test. HCN samples are all replicates from a single patient.

Figure 4. Low diversity correlates with high MSPrecise score in the RRMS cohort but not in the OND cohort

Each data point represents a single sample sequence pool from (a) the RRMS cohort or (b) the OND cohort. The diversity index was calculated as described in the methods section and high values indicate a more even distribution across the VH4 genes. Pearson's correlation coefficient (R) indicates the linear correlation between MSPrecise and the diversity index, and the two-tailed p-value of the correlation is also indicated. The dashed line represents the MSPrecise cut-off point of 6.8 above which patients are expected to have or convert to relapsing-remitting MS (RRMS). The dotted lines delineate an indeterminate range (-1) below the 6.8 cut-off where the results of the MSPrecise score test are less clear cut. (c) Distribution of the diversity index is shown here with the median marked on the graph. HCN samples are all replicates from a single patient. Statistical significance of the difference between cohorts was tested by Mann Whitney test.