Oligoclonal IgG and free light chains – comparison between agarose and polyacrylamide isoelectric focusing

Summary

Aim:

To compare agarose and polyacrylamide isoelectric focusing for oligoclonal imunoglobulin G (o-IgG) and oligoclonal free light chains (o-fLC) detection.

Patients and methods:

Oligoclonal bands were detected in 106 (o-IgG) and 48 (o-fLC), resp. consecutive paired cerebrospinal fluid and serum samples. Kappa statistics was used for method comparison and inter-observer agreement.

Results:

When results were expressed as negative or positive, only three samples (2.8%) for o-IgG, three (6.2%) for o-fLC kappa and one (2.1%) for o-fLC lambda were evaluated differently. Maximum difference between methods was three bands in these discrepant cases. Inter-observer agreement for o-fLC was very good (κ = 0.906–1.000).

Conclusions:

Although polyacrylamide gel might be advantageous due to smaller pore size and hence better resolution, results for both o-IgG and o-fLC were very similar to agarose isoelectric focusing in our series. Both methods performed equally well and discrepances were restricted to borderline positive cases only.

Key words:

cerebrospinal fluid – free light chains – immunoglobulin G – isoelectric focusing – agarose – polyacrylamide

The authors declare they have no potential conflicts of interest concerning drugs, products, or services used in the study.

The Editorial Board declares that the manu­script met the ICMJE “uniform requirements” for biomedical papers.

---

寡克隆IgG和自由光链。琼脂糖和聚丙烯酰胺等电聚焦的比较

目的:

比较琼脂糖和聚丙烯酰胺等电聚焦检测寡克隆免疫球蛋白G (o-IgG)和寡克隆自由轻链(o-fLC)的效果。

患者和方法:

在106例(o-IgG)和48例(o-fLC)脑脊液和血清样品中检测到寡克隆条带。采用Kappa统计量进行方法比较和观察者之间的一致性。

结果:

当结果表达为阴性或阳性时，只有3个样本(2.8%)的o-IgG, 3个(6.2%)的o-fLC kappa和1个(2.1%)的o-fLC lambda得到不同的评价。在这些差异的情况下，方法之间的最大差异是三个波段。Inter-observer协议o-fLC非常好(κ= 0.906 - -1.000)。

结论:

虽然聚丙烯酰胺凝胶由于孔径较小，因此分辨率较高，但在我们的系列中，o-IgG和o-fLC的结果与琼脂糖等电聚焦非常相似。两种方法均表现良好，差异仅局限于阳性的边缘病例。

关键词:

无脑脊液轻链-免疫球蛋白G -等电聚焦-琼脂糖-聚丙烯酰胺

 

Introduction

Oligoclonal IgG (o-IgG) is a principal test used to demonstrate intrathecal antibody synthesis in chronic inflammatory CNS diseases, MS in particular [1]. Although replacing the o-IgG test by a more convenient free kappa light chains (fKLC) quantitation [2,3] has been suggested, o-IgG gained on formal importance for MS diagnosis in the recent revision of McDonald diagnostic criteria [4], therefore it is likely to be performed whenever CSF analysis is required for the support of MS diagnosis.

Besides fKLC quantitation, cerebrospinal fluid (CSF)-restricted fKLC oligoclonality showed to be a sensitive marker of intrathecal inflammation, perhaps even slightly more sensitive than o-IgG [5–7]. However, no between-method comparison is available for this test that is rarely performed due to its labouriousness, especially compared to new automated quantitative methods. Free lambda light chains (fLLC) intrathecal synthesis also occurs in MS and although being less frequent than that of IgG or fKLC [5,7,8], recent studies have shown that simultaneous analysis of fKLC and fLLC may be of prognostic relevance [9–11].

Oligoclonal free light chain (o-fLC) tests gained popularity among the clinicians sending samples to our laboratory, that did not diminish after the introduction of quantitative fLC tests. After the discontinuation of the Multiphor II apparatus production (GE Healthcare Life Sciences, Buckinghamshire, United Kingdom), we looked for an adequate substitute and chose the EDC Flatbed Professional (Electrophoresis Development and Consulting, Tübingen, Germany). Then we decided to assess the performance of the manufacturer´s polyacrylamide gels (PAG) for o-IgG and o-fLC separation. Finally, we aimed at evaluating inter-observer agreement on o-fLC results.

Materials and methods

Clinical specimens

Consecutive paired CSF and serum samples were examined for o-IgG (N = 106) and o-fLC (N = 48) according to a clinician´s request. No clinical data were available for patients followed in other hospitals whose samples had been sent for routine CSF analyses.

Patients followed in our hospital with available clinical data (N = 42) were diagnosed MS (N = 12, six of them after a clinically isolated syndrome according to the latest revision of McDonald diagnostic criteria), clinically isolated syndrome (CIS) (N = 3, not fulfilling MS diagnostic criteria), CNS inflammatory diseases (one case each of herpetic encephalitis, varicella zoster meningoencephalitis, and neuromyelitis optica), peripheral nervous system inflammatory diseases (one case each of acute inflammatory demyelinating polyneuropathy, multifocal motor neuropathy, and paraproteinaemic neuropathy), non-inflammatory neurological diseases (dementia, N = 4; vertigo, N = 2; migraine, N = 2; cryptogenic polyneuropathy, N = 2; and one case each of psychosis, stroke, Parkinson disease, spinal canal stenosis, cervical myelopathy, steroid myopathy) or as symptomatic controls (N = 5).

Oligoclonal IgG and fLC analyses

Agarose isoelectric focusing (IEF) o-IgG detection with immunofixation (IF) was performed with commercially available Hydragel 9 CSF Isofocusing Kit (Sebia, Evry Cedex, France) using Hydrasys apparatus according to the manufacturer´s instructions.

Polyacrylamide IEF was performed on commercially available gels pH 6-11 40S (EDC, Tübingen, Germany) using EDC Flatbed Professional apparatus under the conditions recommended by the manufacturer (Step 1: 500 V, 25 mA, 10 W, 20 min; Step 2: 1350 V, 25 mA, 22 W, 90 min; Step 3: 1750 V, 20 mA, 26 W, 20 min). CSF and serum samples were diluted to 10 mg/L IgG with 0.1% NaCl, and 10 µL were applied into slots in the gel. Sensitive immunodetection based on the principle of a single alkaline-phosphatase labelled anti-IgG antibody method described by Sádaba et al. [12] was performed. Briefly, after capillary blotting for 50 min, the membrane was blocked in 3% bovine serum albumin (Serva Electrophoresis, Heidelberg, Germany) for 45 min and incubated with an alkaline-phosphatase labelled goat anti-human IgG Fc antibody (Bio-Rad, Prague, Czech Republic) for 75 min. Colour reaction was developed using BCIP/NBT substrate (Vector Laboratories, Burlingame, USA).

Agarose IEF for the demonstration of o-fLC was performed using Multiphor II apparatus as described previously [7,8]. PAG fLC IEF was performed on gels pH 3-10 on the Flatbed Professional device under the conditions recommended by the manufacturer (Step 1: 500 V, 12 mA, 10 W, 30 min; Step 2: 1700 V, 12 mA, 18 W, 90 min; Step 3: 2000 V, 8 mA, 20 W, 30 min). For both agarose and PAG fLC IEF, serum samples were diluted 1/80 with 0.7% NaCl and 7 μl of paired samples were applied using the application mask. Subsequent blotting and immunodetection procedure were identical as those for agarose gels.

Oligoclonal IgG evaluation was performed by one of us, whereas inter-observer agreement between two of the authors was studied for o-fLC. IgG bands without serum counterparts or clearly stronger in the CSF than in serum were considered intrathecally synthesized (‘CSF-restricted’); in the latter case, their faint serum counterparts were disregarded for the five-pattern consensus classification. On the contrary, fLC bands were considered intrathecally synthesized only if no serum counterpart was visible because the fLC amounts applied in paired CSF and serum samples were not equalized. The gels and membranes were evaluated in a blinded fashion. Numbers and percentages of discrepantly classified cases were evaluated; chi-squared test and kappa statistics were used for between-method and inter-observer agreement. Spearman´s rank correlation, weighted κ and Wilcoxon test were used to compare the number of CSF-restricted bands. Statistical analyses were performed using the MedCalc Statistical Software version 18.5 (MedCalc Software bvba, Ostend, Belgium).

Albumin, IgG and free light chain quantitation

Cerebrospinal fluid and serum albumin and IgG were measured on BN ProSpec nephelometer (Siemens Healthcare, Prague, Czech Republic). fLC were quantitated by turbidimetry on SPA_PLUS analyser using kits certified for CSF and serum measurements (LK016.L.S and LK018.L.S) (The Binding Site, Birmingham, United Kingdom).

Ethical approval

Informed consent was obtained from all patients followed in our hospital that were included in the study. The study has been approved by the local Ethics Committee as a part of the project ‘CSF biomarkers of multiple sclerosis’ (Ref. No. 400/2017).

Results

Oligoclonal IgG

106 consecutive samples were compared. The results are presented in Tab. 1 and a representative example is provided in Fig. 1.

In total, 3 samples (2.8 %) were classified differently. Numbers of CSF-restricted IgG bands in these cases are presented in Tab. 2.

1 CSF-restricted IgG band was found in 7/106 (6.6%) samples on Sebia IEF/IF. On PAG IEF/IB, no band was seen in 4 of these cases, 1 band in 2, and 4 bands in 1 case. An example of such a borderline result is provided in Fig. 2.

On PAG IEF/IB, 1 CSF-restricted IgG band was found in 10 cases (9.4%). On Sebia IEF/IF, no band was seen in 6 of these cases, 1 band in 2, and 2 and 3 bands in one case each.

A more elaborate o-IgG pattern classification containing five types according to two consensus statements [13,14] was compared as well. An agreement was reached on the o-IgG pattern in 75 samples (70.8 %). κ statistics was 0.503, i.e. only moderate agreement. It can be seen that pattern 1 using one of the methods was classified as pattern 4 using another method in 19 cases, and analogous exchange of patterns 2 and 3 contributed to another 8 differently classified patterns. Detailed results are presented in Tab. 3.

The number of CSF-restricted IgG bands was compared by Spearman´s rank correlation coefficient (ρ 0.861; 95% CI 0.802–0.903, P < 0.0001) and weighted κ statistics (0.795; 95% CI 0.7432–0.8473) indicating good between-method agreement. No systematic difference was revealed by the Wilcoxon test in the number of CSF-restricted IgG bands between the two methods (16 positive and 17 negative differences, P = 0.8442).

Oligoclonal fLC

Oligoclonal-fLC were detected in 48 samples. Results are presented in Tab. 4 and 5 and a representative example is provided in Fig. 3. Both between-methods and inter-observer agreement were very good. Inter-observer κ values were 0.9064 and 0.9091 for agarose and PAG o-fKLC and 0.9286 and 1.000 for agarose and PAG o-fLLC.

A different classification was observed in 3 samples (6.2%) for o-fKLC and 1 (2.1%; observer A) or 2 (4.2%) samples (observer B) for fLLC. The number of CSF-restricted fKLC bands in discrepant cases is presented in Tab. 2. For fLLC, the only discrepant case had no fLLC bands in agarose but 2 CSF-restricted bands on PAG.

Only one CSF-restricted fLC band was noted in one to four samples (2.1–8.3%) depending on the observer and separation method used. The number of CSF-restricted fLC bands was compared by Spearman´s rank correlation coefficient (ρ 0.847–0.978; P < 0.0001 for all comparisons) and weighted κ (0.804–0.912) indicating very good agreement. Notably we observed a higher number of CSF-restricted fKLC bands using PAG compared to agarose (Wilcoxon test, P = 0.0063 and P = 0.0023 for observers A and B, respectively), and a tendency for a higher number of CSF-restricted fLLC bands (Wilcoxon test significant only for observer B, P = 0.0137).

Comparison with quantitative measures of intrathecal IgG and fLC synthesis

Comparing quantitative IgG measurements and calculations with agarose IEF/IF, there were no false positives using the formulas of Reiber [15], Auer et al. [16] or Öhman et al. [17], while there was one false positive result for IgG index (using a conventional criterion > 0.7). However, the sensitivity of all the calculations was low (43.5% for IgG index and the formulas of Reiber and Auer et al., and 52.2% for the formula of Öhman et al.).

All 10 samples with intrathecal IgG synthesis according to the formulas of Reiber and Auer et al. displayed CSF-restricted oligoclonal IgG bands using both methods, while one sample with intrathecal IgG synthesis, according to the formula of Öhman, was positive for o-IgG by agarose IEF/IF but negative by PAG IEF/IB. Notably, the optimum cut-off value for IgG index obtained from the analysis of ROC curve (not shown) was > 0.589 (sensitivity: 69.6% and 72.7%; specificity: 96.4% and 96.4% compared to agarose IEF/IF and PAG IEF/IB, resp.), while the conventional criterion > 0.7 resulted in sensitivity of 43.5% and 45.5% and specificity of 98.8% and 98.8% compared to agarose IEF/IF and PAG IEF/IB, resp..

Although fLC quantitation was requested in 10 samples only, we consider the results to be interesting enough to warrant their presentation (Tab. 6). Using the cut-offs for the presence of intrathecal fLC synthesis previously determined in our laboratory (CSF fKLC 0.54 mg/L; CSF fLLC 0.30 mg/L; fKLC index 6.07; fLLC index 6.27), there was 100% agreement between quantitative and qualitative values for fKLC, 90% for CSF fLLC and 80% for fLLC index. One patient with normal CSF fLLC yet increased fLLC index had no CSF-restricted fLLC bands, while one patient with normal CSF fLLC and fLLC index had 2 CSF-restricted fLLC bands in PAG and 1 (observer B) or 2 (observer A) faint bands in agarose. Notably, while 100% agreement was obtained for fKLC using both non-linear formulas for intrathecal synthesis [18,19], the recently proposed cut-off of 4.2 for fLLC index or the formula for intrathecal fLLC synthesis [19] resulted in 3 and 2 presumably false positives, resp.

Correlation with clinical data

Although this study has not been aimed at correlations of o-IgG and o-fLC findings with clinical data, we analysed the results of patients with available clinical diagnoses (N = 42 for o-IgG; N = 22 for o-fLC) as an additional test for the plausibility of our laboratory findings. All 12 patients diagnosed with MS were positive for o-IgG on agarose IEF/IF; one of them was negative on PAG IEF/IB (1 CSF-restricted IgG band only). This patient displayed 3 CSF-restricted o-IgG bands on agarose IEF/IF but was negative for o-fKLC and o-fLLC. All 3 patients diagnosed as CIS were positive for o-IgG using both methods. 27 non-MS patients were uniformly negative for o-IgG. Out of 6 patients with MS evaluated for o-fLC, o-fKLC were positive in 5 and negative in 1 MS patient. 3 of these patients were also positive for o-fLLC. Both CIS patients evaluated for o-fLC were positive for o-fKLC as well as o-fLLC (in one case, o-fLLC were visible only on PAG; otherwise, there was complete agreement between methods and observers). Out of 14 non-MS patients, 3–5 were positive for o-fKLC (maximum of 5 bands on PAG and 7 bands on agarose in a patient with multifocal motor neuropathy; one and two patients were discrepantly classified by the two observers using PAG and agarose IEF/AIB, resp.) but only one patient (with a diagnosis of cryptogenic polyneuropathy) was positive for o-fLLC. Interestingly, using a higher cut-off of ≥ 6 CSF-restricted o-fKLC bands, as proposed in an earlier study [7], would only result in 1 non-MS patient, but at the same time only 3 MS patients classified as o-fKLC positive for agarose IEF/AIB. By contrast, the same cut-off value applied to PAG IEF/AIB would preserve the o-fKLC positivity in all 5 MS patients, whereas leaving all 14 non-MS patients o-fKLC negative. There was 100% inter-observer agreement on these cases.

Discussion     

Comparison of various o-IgG separation and specific detection methods in contemporary use has been reported in several studies [12,20–23], but the use of agarose versus polyacrylamide IEF was not systematically studied in this context. For CSF o-fLC analysis, we were not able to find a single comparative study.

We aimed to compare two commercially available methods for o-IgG and separation of o-fLC in agarose vs. PAG, followed by identical detection procedure established for o-fLC [7,8]. The strength of our study lies in the recruitment of consecutive samples sent for o-IgG and o-fLC analysis, preventing any possible bias, and potentially providing some estimate of an expected proportion of positive samples. We admit that the absence of clinical information for more than 50% of the studied patients might be considered a weakness of our study. Nevertheless, the correlations between various tests aimed at detection of intrathecal immunoglobulin synthesis were very good, and results for a subgroup of patients with known diagnoses were consistent with previous reports. Hence, we consider our results to be reliable. However, clinical-laboratory correlations observed in our study should be interpreted with great caution since the o-IgG and o-fLC results were available to clinicians and could thus influence their diagnostic decisions. Although it should be kept in mind that intrathecal immunoglobulin synthesis is not disease-specific and may occur in many inflammatory CNS diseases, it appears that the o-IgG test is not only highly sensitive, but also quite specific for MS in an unselected series of patients indicated for o-IgG testing.

A much larger sample size would be required to reveal subtle but significant differences. Using κ statistics, however, it can be concluded that there was very good agreement (κ > 0.80) in all comparisons except for o-IgG pattern classification. Moreover, the differences concerning positivity/negativity of the o-IgG test as well as o-fLC tests were observed in borderline cases only.

The agreement on o-IgG patterns according to international consensus classification was less pronounced, in agreement with earlier studies [20,22,23]. The significance of this more elaborate classification is controversially discussed [1,24,25]. We speculate that only moderate agreement on these five patterns (opposed to very good agreement on a simple negative/positive classification) might be the main reason why this classification is of rather limited clinical utility.

Borderline results are important since when reported, they may contribute to uncertainty and resulting stress of a patient. Single CSF-restricted band may represent a ‘tip of the iceberg’ of the monoclonal as well as oligoclonal response [25]. In routine practice, the presence of only 1 CSF-restricted band is considered either negative or borderline. For o-IgG, we have noted such a result in 7 and 9% of samples, an incidence higher than reported in previous studies [26–29]. We classify such a result as negative; however, unlike others [1] but in accordance with the results of a recent multicentre study [29], we do report such a finding. If MS is suspected in these cases, close follow-up is recommended since it has been shown that a certain proportion of these patients develop oligoclonal profile over time [26,27,29]. At least 2 bands are required for the statement of o-IgG positivity and similar cut-off is used for o-fLC. In several reports, even higher cut-offs of three [30,31] or four [3,7,28,32] CSF-restricted IgG bands or six [7] fKLC bands have been proposed. Further studies might warrant comparing contemporary o-IgG and o-fLC separation/detection methods, perhaps including comparison with quantitative fLC results and focusing on ‘critical samples’ with one or a few CSF-restricted bands.

It has been proven once again that calculation methods are less sensitive for intrathecal IgG synthesis detection. No definite conclusions can be made for quantitative analysis of fLC in this study due to a low number of analysed samples. However, our results suggest apossibility of both false negative and false positive results of the quantitative fLC test. fLC quantitation using different reagents/instruments might not provide comparable values to allow the universal use of cut-off values for intrathecal fLC synthesis calculations. Discrepant results between quantitative and qualitative fKLC tests were also reported in a recent study [33]. In general, o-fLC tests, although previously shown to be slightly more sensitive for the demonstration of intrathecal inflammatory response than o-IgG, might be less specific for MS than o-IgG [5-8]. We therefore agree with Bayart et al. [33] that fLC tests might be used as a complement to rather than a substitute for o-IgG analysis that is still listed as a key element of CSF analysis in published local recommendations for CSF analysis [34,35].

Polyacrylamide IEF might be particularly advantageous for smaller fLC molecules because of a smaller pore size. Indeed, we observed a higher number of CSF-restricted fLC bands using PAG compared to agarose. In addition, we felt that fLC bands were more even and the overall o-fLC pattern was sharper in PAG compared to agarose where distorted bands were common. We admit, however, that these differences might have been influenced not only by the separation medium, but also by different programmes used for the IEF run. In any case, these minor differences seem not to be critical for the final qualitative evaluation of the o-fLC results.

In conclusion, the agreement between agarose and polyacrylamide IEF for o-IgG and o-fLC detection was very good and we consider both methods to be suitable for routine practice. Inter-observer agreement on o-fLC results was as good as demonstrated earlier for o-IgG.

Accepted for review: 4. 10. 2018

Accepted for print: 10. 12. 2018

Acknowledgements

We wish to thank laboratory technicians L. Fürstová, I. Faruzelová, R. Malečková, R. Výtisková, B. Strakošová and O. Michnová for their skillful technical assistance; to Prof. D. Stejskal, Assoc. Prof. K. Šafarčík and Dr. Z. Švagera (Institute of Laboratory Diagnostics, University Hospital Ostrava), Dr. J. Jurčíková and Dr. K. Vítková (Dept. for Science and Research, University Hospital Ostrava) for their continuous support; and to M. Hradecká for language corrections.

Supporting agencies

The study was supported in part by an Institutional grant From Ministry of Health, Czech Republic (07/RVO-FNOs/2017).

Ing. MUDr. David Zeman, Ph.D.

Ústav laboratorní diagnostiky FN Ostrava

17. Listopadu 1790

708 52 Ostrava

e-mail: david.zeman@fno.cz