CSF Oligoclonal Bands in MS and Clinically Isolated Syndrome
CSF Oligoclonal Bands in MS and Clinically Isolated Syndrome
Following the initial screening, 350 unique papers were identified. The abstracts and full text of these papers were then hand searched for papers meeting the inclusion criteria. Seventy-one articles were selected for inclusion in the final analysis (see online supplementary appendix 1). The reasons for rejecting papers at this stage were varied, but most commonly included papers that selected small numbers of patients for methodological studies (n=58), papers that selected OCB-positive or OCB-negative patients only (n=15), review papers (n=18) and papers examining IgM OCB only (n=12). The selection process is summarised in figure 1.
(Enlarge Image)
Figure 1.
Study selection.
Forty-eight studies were used to assess the prevalence of OCB in MS and CIS, with 32 giving information on OCB prevalence in MS and 21 in CIS (some papers covered both MS and CIS). 36 papers were used to calculate the relationship between OCB and clinical outcomes and 12 papers gave qualitative information regarding the relationship between clinical outcomes and OCB status (see online supplementary appendix 1). Of the 36 papers used to calculate outcomes, 18 used IEF with immunofixation, in 12 the technique was not specified and the remaining 6 used a specified technique other than IEF with immunofixation, most commonly electrophoresis with silver staining. Ten studies used in the clinical outcomes data analysis studied outcomes in MS, and of the remaining 26 examining CIS, 9 selected patients with ON. Forty-four papers were used in the latitudinal analysis, with some papers providing information on both MS and CIS. 28 papers were used to determine the relationship between latitude and OCB status in MS, and 19 in CIS.
There was OCB data meeting the inclusion criteria in a total of 12 253 MS patients, of whom 10 751 were OCB positive and 1577 OCB negative; overall, 87.7% patients with MS were OCB positive. When the three Asian studies were excluded, 10 719 of 12 171 (88.1%) MS patients were found to be OCB positive. When all studies were included, regardless of the population and the technique used to detect OCB, 16 678 of 19 773 (84.3%) MS patients were OCB positive. A conservative analysis, where only those papers using IEF with immunofixation were included, and all papers possibly using duplicate cases (ie, those originating from the same centre) and the Asian studies were excluded, showed 5495 of 6118 (89.8%) patients with MS were OCB positive. There was a significant difference in the OCB positivity rate between the 'all studies' and the conservative analysis (p<0.0001; χ with Yates' correction).
There was OCB data meeting the inclusion criteria in a total of 2685 patients with CIS, of whom 1841 were OCB positive and 844 OCB negative; overall, 68.6% patients with CIS were OCB positive. There were no studies examining OCBs in CIS in Asian patients. When all studies were included regardless of techniques, 3580 of 5154 (69.5%) patients were OCB positive. A conservative analysis showed 1489 of 2205 (67.5%) CIS patients to be OCB positive. There was no significant difference in the OCB positivity rate between the 'all studies' and the conservative analysis.
Ten studies gave data about clinical outcomes in patients with MS. Of these, four used IEF with immunofixation. In all of the studies using IEF with immunofixation, Expanded Disability Status Scale (EDSS)-related outcome measures were used to define clinical outcomes; one used EDSS of 4 at 10 years disease duration, two used EDSS 6 during follow-up and one used an increase of ≥1 EDSS point in 5 years. When the results were combined, 667 of 1764 (37.8%) OCB-positive patients reached the disability outcome measures specified in the study compared with 42 of 154 (27.2%) OCB-negative patients (p<0.0001, Fisher's exact test). When the meta-analysis was performed, this gave an OR of reaching the disability outcome of 1.96 (95% CI 1.31 to 2.94; p=0.001) with no between-study heterogeneity (I=0%; X=2.95, df=3, p=0.40) (figure 2). There was no significant publication bias (Egger p value=0.12). A subgroup analysis of the two studies using EDSS 6 as an endpoint gave an OR of reaching EDSS 6 of 2.03 (95% CI 1.24 to 3.33; p=0.005) with no heterogeneity (I=0%; X=0.87, df=1, p=0.35).
(Enlarge Image)
Figure 2.
Relationship between oligoclonal band status and clinical outcomes in multiple sclerosis.
When the six studies using other techniques for measuring OCBs were included (see online supplementary appendix 1), the range of outcome measures used increased. EDSS outcome was used by a number of studies, including EDSS 6, 7.5 or 8 between 5 and 10 years disease duration. One study used worsening of EDSS by 1 point over 2 years, and another study used 'poor recovery from relapses'. Seven hundred and seventy of 2202 (35.0%) OCB-positive patients reached the disability outcome measure associated with the study compared with 66 of 333 (19.8%) OCB-negative patients (p<0.0001, Fisher's exact test). Inclusion of the studies using alternative techniques and EDSS-defined outcome measurements gave an OR of meeting the study endpoint of 1.65 (95% CI 1.27 to 2.13; p=0.0002) with moderate heterogeneity (I=48%; X=0.22.97, df=13, p=0.03) (data not shown). This was not significantly different from when only IEF with immunofixation was used.
Thirteen studies gave narrative results without absolute numbers. One study found a significantly lower relapse rate in OCB-negative patients (relapse rate 1.45±0.69 in OCB positive and 0.58±0.64 in OCB negative, p=0.001). None of the other studies demonstrated any relationship between the presence of OCBs and the disability outcomes collected, including relapse rate, EDSS and MS severity score.
Fourteen studies examined the relationship between OCB detected by IEF with immunofixation and outcomes in CIS. Two of these studies specified ON, and one study specified a brainstem syndrome as the CIS. Twelve studies used conversion to clinically definitive multiple sclerosis (CDMS) as the outcome, one used radiological conversion to MS and one used the number of patients reaching EDSS 6 at 5 years. The study using EDSS 6 as the outcome measures was excluded given the very different outcome measure, leaving 13 studies in the analysis (see online supplementary appendix). Seven hundred and thirty-three of 1143 (64.1%) OCB-positive patients converted to MS compared with 139 of 616 (22.6%) OCB-negative patients (p<0.0001, Fisher's exact test). This gave a sensitivity of 0.84 and a specificity of 0.54 when using OCB to predict conversion to CDMS. The positive predictive value was 0.64 and the NPV was 0.77.
When the meta-analysis was performed, there was an OR of conversion to MS of 9.88 (95% CI 5.44 to 17.94; p<0.00001) in the OCB-positive patients (figure 3). However, there was significant between-study heterogeneity (I=71%; X=40.79, df=12, p<0.0001). There was no evidence of publication bias (Egger p value=0.20). Excluding the study using radiological conversion did not significantly alter this result. Other attempts to explore the underlying causes of the heterogeneity observed were similarly unsuccessful.
(Enlarge Image)
Figure 3.
Relationship between oligoclonal band status and conversion to multiple sclerosis in clinically isolated syndrome.
When all of the studies examining the relationship between OCB and conversion to MS (regardless of the technique used to detect OCB) were considered (an additional 12 studies; see online supplementary appendix 1), 973 of 1584 (61.4%) OCB-positive CIS patients converted to MS compared with 173 of 927 (18.7%) OCB-negative CIS patients (p<0.0001, Fisher's exact test). This gave an OR of conversion to MS of 9.99 (95% CI 6.54 to 15.27; p<0.00001) in the OCB-positive patients (data not shown). There was significant between-study heterogeneity (I=57%; X=56.27, df=24, p=0.0002), which proved impossible to eliminate. This result was not significantly different from that when only those studies using IEF with immunofixation were used.
When all studies were included, there appeared to be a relationship between the proportion of OCB-positive patients converting to CDMS and the duration of follow-up (using linear regression, p=0.042, R=0.1833) (figure 4). However, when only those studies using IEF with immunofixation were included, this relationship was no longer significant. There was no relationship between the proportion of OCB-negative patients converting to CDMS and the duration of follow-up. Given the low conversion rate in the OCB-negative group together with the lack of any relationship between conversion rate and duration of follow-up in the OCB-negative group, it was not possible to determine whether conversion occurs sooner in those who are OCB positive.
(Enlarge Image)
Figure 4.
Graph demonstrating the relationship between duration of follow-up and conversion rates in those who are oligoclonal band positive (data from all studies).
Given the large number of studies examining outcomes in ON, these were studied separately. The majority (7 of 9) did not specify that IEF with immunofixation was used. The results were similar to those obtained with all CIS, with 474 of 743 (63.8%) OCB-positive patients developing MS compared with 98 of 429 (22.8%) OCB-negative patients (p<0.0001, Fisher's exact test). The meta-analysis gave an OR of conversion to MS of 10.13 (95% CI 7.11 to 14.44; p<0.00001) in the OCB-positive patients with no heterogeneity (I=0%; X=6.96, df=8, p=0.54).
Twenty-eight studies giving data on OCB in MS were used to determine the effect of latitude on the proportion of MS samples positive for OCBs. Only studies using IEF with immunofixation were included in this section of the analysis. Linear regression revealed a significant relationship between OCB positivity and latitude (p=0.002, figure 5) with a correlation coefficient (R) of 0.31. This relationship was maintained when an additional variable for sample size was included in the model (for effect of latitude, p=0.009; for effect of sample size, p=0.833). When the Asian studies were excluded, a significant relationship remained (p=0.005 in the linear regression model; R=0.169).
(Enlarge Image)
Figure 5.
Graph demonstrating the relationship between latitude and proportion of multiple sclerosis patients who are oligoclonal band positive.
Nineteen studies were included in the latitudinal regression model for CIS. There was no significant relationship between the proportion of OCB-positive samples and the latitude (p=0.099; data not shown); this was not altered by the inclusion of sample size in the model (for effect of latitude, p=0.119; for effect of sample size, p=0.856).
Results
Included Papers
Following the initial screening, 350 unique papers were identified. The abstracts and full text of these papers were then hand searched for papers meeting the inclusion criteria. Seventy-one articles were selected for inclusion in the final analysis (see online supplementary appendix 1). The reasons for rejecting papers at this stage were varied, but most commonly included papers that selected small numbers of patients for methodological studies (n=58), papers that selected OCB-positive or OCB-negative patients only (n=15), review papers (n=18) and papers examining IgM OCB only (n=12). The selection process is summarised in figure 1.
(Enlarge Image)
Figure 1.
Study selection.
Forty-eight studies were used to assess the prevalence of OCB in MS and CIS, with 32 giving information on OCB prevalence in MS and 21 in CIS (some papers covered both MS and CIS). 36 papers were used to calculate the relationship between OCB and clinical outcomes and 12 papers gave qualitative information regarding the relationship between clinical outcomes and OCB status (see online supplementary appendix 1). Of the 36 papers used to calculate outcomes, 18 used IEF with immunofixation, in 12 the technique was not specified and the remaining 6 used a specified technique other than IEF with immunofixation, most commonly electrophoresis with silver staining. Ten studies used in the clinical outcomes data analysis studied outcomes in MS, and of the remaining 26 examining CIS, 9 selected patients with ON. Forty-four papers were used in the latitudinal analysis, with some papers providing information on both MS and CIS. 28 papers were used to determine the relationship between latitude and OCB status in MS, and 19 in CIS.
OCB Prevalence in MS and CIS
There was OCB data meeting the inclusion criteria in a total of 12 253 MS patients, of whom 10 751 were OCB positive and 1577 OCB negative; overall, 87.7% patients with MS were OCB positive. When the three Asian studies were excluded, 10 719 of 12 171 (88.1%) MS patients were found to be OCB positive. When all studies were included, regardless of the population and the technique used to detect OCB, 16 678 of 19 773 (84.3%) MS patients were OCB positive. A conservative analysis, where only those papers using IEF with immunofixation were included, and all papers possibly using duplicate cases (ie, those originating from the same centre) and the Asian studies were excluded, showed 5495 of 6118 (89.8%) patients with MS were OCB positive. There was a significant difference in the OCB positivity rate between the 'all studies' and the conservative analysis (p<0.0001; χ with Yates' correction).
There was OCB data meeting the inclusion criteria in a total of 2685 patients with CIS, of whom 1841 were OCB positive and 844 OCB negative; overall, 68.6% patients with CIS were OCB positive. There were no studies examining OCBs in CIS in Asian patients. When all studies were included regardless of techniques, 3580 of 5154 (69.5%) patients were OCB positive. A conservative analysis showed 1489 of 2205 (67.5%) CIS patients to be OCB positive. There was no significant difference in the OCB positivity rate between the 'all studies' and the conservative analysis.
Relationship Between OCB Status and Clinical Outcomes in MS
Ten studies gave data about clinical outcomes in patients with MS. Of these, four used IEF with immunofixation. In all of the studies using IEF with immunofixation, Expanded Disability Status Scale (EDSS)-related outcome measures were used to define clinical outcomes; one used EDSS of 4 at 10 years disease duration, two used EDSS 6 during follow-up and one used an increase of ≥1 EDSS point in 5 years. When the results were combined, 667 of 1764 (37.8%) OCB-positive patients reached the disability outcome measures specified in the study compared with 42 of 154 (27.2%) OCB-negative patients (p<0.0001, Fisher's exact test). When the meta-analysis was performed, this gave an OR of reaching the disability outcome of 1.96 (95% CI 1.31 to 2.94; p=0.001) with no between-study heterogeneity (I=0%; X=2.95, df=3, p=0.40) (figure 2). There was no significant publication bias (Egger p value=0.12). A subgroup analysis of the two studies using EDSS 6 as an endpoint gave an OR of reaching EDSS 6 of 2.03 (95% CI 1.24 to 3.33; p=0.005) with no heterogeneity (I=0%; X=0.87, df=1, p=0.35).
(Enlarge Image)
Figure 2.
Relationship between oligoclonal band status and clinical outcomes in multiple sclerosis.
When the six studies using other techniques for measuring OCBs were included (see online supplementary appendix 1), the range of outcome measures used increased. EDSS outcome was used by a number of studies, including EDSS 6, 7.5 or 8 between 5 and 10 years disease duration. One study used worsening of EDSS by 1 point over 2 years, and another study used 'poor recovery from relapses'. Seven hundred and seventy of 2202 (35.0%) OCB-positive patients reached the disability outcome measure associated with the study compared with 66 of 333 (19.8%) OCB-negative patients (p<0.0001, Fisher's exact test). Inclusion of the studies using alternative techniques and EDSS-defined outcome measurements gave an OR of meeting the study endpoint of 1.65 (95% CI 1.27 to 2.13; p=0.0002) with moderate heterogeneity (I=48%; X=0.22.97, df=13, p=0.03) (data not shown). This was not significantly different from when only IEF with immunofixation was used.
Thirteen studies gave narrative results without absolute numbers. One study found a significantly lower relapse rate in OCB-negative patients (relapse rate 1.45±0.69 in OCB positive and 0.58±0.64 in OCB negative, p=0.001). None of the other studies demonstrated any relationship between the presence of OCBs and the disability outcomes collected, including relapse rate, EDSS and MS severity score.
Relationship Between OCB Status and Outcomes in CIS
Fourteen studies examined the relationship between OCB detected by IEF with immunofixation and outcomes in CIS. Two of these studies specified ON, and one study specified a brainstem syndrome as the CIS. Twelve studies used conversion to clinically definitive multiple sclerosis (CDMS) as the outcome, one used radiological conversion to MS and one used the number of patients reaching EDSS 6 at 5 years. The study using EDSS 6 as the outcome measures was excluded given the very different outcome measure, leaving 13 studies in the analysis (see online supplementary appendix). Seven hundred and thirty-three of 1143 (64.1%) OCB-positive patients converted to MS compared with 139 of 616 (22.6%) OCB-negative patients (p<0.0001, Fisher's exact test). This gave a sensitivity of 0.84 and a specificity of 0.54 when using OCB to predict conversion to CDMS. The positive predictive value was 0.64 and the NPV was 0.77.
When the meta-analysis was performed, there was an OR of conversion to MS of 9.88 (95% CI 5.44 to 17.94; p<0.00001) in the OCB-positive patients (figure 3). However, there was significant between-study heterogeneity (I=71%; X=40.79, df=12, p<0.0001). There was no evidence of publication bias (Egger p value=0.20). Excluding the study using radiological conversion did not significantly alter this result. Other attempts to explore the underlying causes of the heterogeneity observed were similarly unsuccessful.
(Enlarge Image)
Figure 3.
Relationship between oligoclonal band status and conversion to multiple sclerosis in clinically isolated syndrome.
When all of the studies examining the relationship between OCB and conversion to MS (regardless of the technique used to detect OCB) were considered (an additional 12 studies; see online supplementary appendix 1), 973 of 1584 (61.4%) OCB-positive CIS patients converted to MS compared with 173 of 927 (18.7%) OCB-negative CIS patients (p<0.0001, Fisher's exact test). This gave an OR of conversion to MS of 9.99 (95% CI 6.54 to 15.27; p<0.00001) in the OCB-positive patients (data not shown). There was significant between-study heterogeneity (I=57%; X=56.27, df=24, p=0.0002), which proved impossible to eliminate. This result was not significantly different from that when only those studies using IEF with immunofixation were used.
When all studies were included, there appeared to be a relationship between the proportion of OCB-positive patients converting to CDMS and the duration of follow-up (using linear regression, p=0.042, R=0.1833) (figure 4). However, when only those studies using IEF with immunofixation were included, this relationship was no longer significant. There was no relationship between the proportion of OCB-negative patients converting to CDMS and the duration of follow-up. Given the low conversion rate in the OCB-negative group together with the lack of any relationship between conversion rate and duration of follow-up in the OCB-negative group, it was not possible to determine whether conversion occurs sooner in those who are OCB positive.
(Enlarge Image)
Figure 4.
Graph demonstrating the relationship between duration of follow-up and conversion rates in those who are oligoclonal band positive (data from all studies).
Given the large number of studies examining outcomes in ON, these were studied separately. The majority (7 of 9) did not specify that IEF with immunofixation was used. The results were similar to those obtained with all CIS, with 474 of 743 (63.8%) OCB-positive patients developing MS compared with 98 of 429 (22.8%) OCB-negative patients (p<0.0001, Fisher's exact test). The meta-analysis gave an OR of conversion to MS of 10.13 (95% CI 7.11 to 14.44; p<0.00001) in the OCB-positive patients with no heterogeneity (I=0%; X=6.96, df=8, p=0.54).
Relationship Between OCB and Latitude
Twenty-eight studies giving data on OCB in MS were used to determine the effect of latitude on the proportion of MS samples positive for OCBs. Only studies using IEF with immunofixation were included in this section of the analysis. Linear regression revealed a significant relationship between OCB positivity and latitude (p=0.002, figure 5) with a correlation coefficient (R) of 0.31. This relationship was maintained when an additional variable for sample size was included in the model (for effect of latitude, p=0.009; for effect of sample size, p=0.833). When the Asian studies were excluded, a significant relationship remained (p=0.005 in the linear regression model; R=0.169).
(Enlarge Image)
Figure 5.
Graph demonstrating the relationship between latitude and proportion of multiple sclerosis patients who are oligoclonal band positive.
Nineteen studies were included in the latitudinal regression model for CIS. There was no significant relationship between the proportion of OCB-positive samples and the latitude (p=0.099; data not shown); this was not altered by the inclusion of sample size in the model (for effect of latitude, p=0.119; for effect of sample size, p=0.856).