Comparison of SimoaTM and EllaTM to assess serum neurofilament‐light chain in multiple sclerosis

Abstract We compared SimoaTM and EllaTM immunoassays to assess serum neurofilament‐light chain levels in 203 multiple sclerosis patients from the OFSEP HD study. There was a strong correlation (ρ = 0.86, p < 0.0001) between both platforms. The EllaTM instrument overestimated values by 17%, but as the data were linear (p = 0.57), it was possible to apply a correction factor to EllaTM results. As for SimoaTM, serum neurofilament‐light chain levels measured by EllaTM were correlated with age and EDSS and were significantly higher in active multiple sclerosis, suggesting that these assays are equivalent and can be used in routine clinical practice.


Introduction
Neurofilaments (Nf) are major components of the neuronal cytoskeleton, consisting predominantly of three subunits: Nf-light (NfL), Nf-medium and Nf-heavy chains. 1 Upon neuro-axonal damage of the central nervous system (CNS), NfL is released into the extracellular space and is detectable in the cerebrospinal fluid and blood. 2 Thus, NfL levels are increased proportionally to the degree of damage, 2 making serum NfL levels a useful biomarker for diagnosing and predicting disease progression of a variety of CNS disorders, including multiple sclerosis (MS). 3 In MS, serum NfL is correlated with several factors including age, Expanded Disability Status Scale (EDSS), disease activity and disease-modifying treatments. 4 Several ultrasensitive immunoassay technologies are available for quantification of serum NfL. The current reference method is the Single Molecular Array (Simoa TM , Quanterix) 5 using an antibody developed by Uman Diagnostics. Recently, several companies have acquired this antibody, allowing NfL quantification using the Simple Plex TM Ella (Ella TM ) microfluidic platform (ProteinSimple). The Ella TM instrument allows rapid and ultra-sensitive measurement of biomarkers. 6 This platform allows quantitation of an analyte from 72 samples in a single disposable microfluidic cartridge, within 90 minutes (ProteinSimple, 2020). However, the comparability of the two technologies in measuring serum NfL levels in patients with MS remains to be determined.
The objective of this study was to compare the NfL values obtained using the Simoa TM platform with Ella TM instrument in MS patients and healthy controls (HCs). Correlations of the serum NfL measures were performed to evaluate whether Ella TM had good clinical performance in reflecting age, EDSS and disease activity, and could be routinely used to monitor MS patients in clinical practice.

Serum samples
Anonymized serum samples were taken from 203 of the 1800 anticipated patients ≥15 years old with MS according to the revised McDonald diagnosis criteria included in the OFSEP "High Definition" cohort (NCT03981003), and from 30 HCs. Ethics approvals were obtained, and all patients and controls participated voluntarily in the study and provided written informed consent (Details in Supplementary materials and methods).

Simoa TM and Ella TM NfL assay
Serum NfL concentrations were prospectively determined in parallel with the Simoa TM Human Neurology 4-Plex "A" kit (Quanterix Corp, Boston, MA) on Simoa TM HD-1 analyzer and Simple Plex TM NfL Assay (ProteinSimple, CA, USA) on Ella TM instrument, according to the manufacturers' instructions. Ella TM was calibrated using the incartridge factory standard curve and Simoa TM using the provided standards. All samples were measured in simplicate, on the same day, after a single thaw, with a 1:2 dilution for Ella TM and 1:4 for Simoa TM . In each run, the HC, one control patient with active relapsing remitting MS (RRMS), and one high and one low concentration control sample provided with the kits were assayed. The lower limit of quantification is 0.241 pg/ml for Simoa TM and 2.70 pg/ml for Ella TM .

Statistical analysis
The intra-assay coefficients of variation (CV) of manufacturer-provided controls were automatically calculated in duplicate (Simoa TM ) or internal triplicate (Ella TM ). Repeatability tests were performed with samples at high (RRMS patient) and low (HC) concentrations by repeated measures for Simoa TM (30 times each) and for Ella TM (28 times and 25 times, respectively). Intra-assay CV was calculated from the standard deviation of the average concentrations divided by the overall mean of the average concentrations.
Median NfL values obtained by each platform were compared using the Wilcoxon-Mann-Whitney test. Spearman correlation coefficients were calculated to assess the association between concentrations obtained by each platform, presented with 95% confidence interval (95% CI). The Bland-Altman method 7 was used to measure mean difference and 95% limit of agreement between logtransformed concentrations obtained by each platform. The regression relationship between the two platforms was evaluated using Passing-Bablok. 8 Finally, correlations of serum NfL levels with clinical parameters were analyzed using linear regression (age, EDSS) or Wilcoxon-Mann-Whitney (e.g. RRMS vs. progressive MS).

Data availability statement
Anonymized data will be shared by request from any qualified investigator.

Results
Repeatability tests were performed by measuring 25-30 times one sample at low concentration (HC) and one sample at high concentration (RRMS patient) and showed similar CVs with both platforms ( Supplementary Figure A). The mean [min-max] intra-assay CVs on Ella TM technology was 2.12% [1.53-2.70] vs 3.78% [2.93-4.63] on Simoa TM platform. The mean [min-max] inter-assay CV of the three runs was 12.93% [7.59-18.27] on Ella TM and 5.54% [5.08-6.00] on Simoa TM . In MS patients, median serum NfL levels [interquartile range] measured by Ella TM were higher than by Simoa TM (13.90 pg/ml [10.73-18.48] for Ella TM vs. 9.46 pg/ml [6.94-12.9] for Simoa TM , p < 0.001) ( Figure 1A). Serum NfL levels were strongly correlated between the two technologies in MS patients (  The Bland-Altman method depicted a mean bias of 17.6% for the NfL concentrations between the assays performed with the two technologies. Thus, Ella TM showed a 17.6% "overestimation" compared with Simoa TM . Overall, 95% of observations were within the limit of agreement ( Figure 1C). The slope of the Passing-Bablok regression line was 1.161 (95% CI [1.091-1.240], p < 0.0001) and the intercept was 2.917 pg/ml (95% CI [2.132-3.676], p < 0.0001). The 95% CI of intercept and slope values differ from zero and one, respectively, indicating a method agreement and allowing application of a correction coefficient. 9 Moreover, the linearity test demonstrated no significant deviation from linearity between the two datasets (p = 0.57), suitable for concluding on method agreement ( Figure 1D).
Both platforms exhibited significant correlations of serum NfL with age, EDSS and disease form ( Figure 2). Especially, serum NfL levels were higher in RRMS patients than in age-matched HCs, higher in active MS than in inactive MS, higher during relapses than in patients with a stable disease and higher in PMS than in RRMS patients with both platforms ( Figure 2B). The last comparison was no longer significant in a multivariate model including age. Figure 2. Comparison of serum NfL values measured by the Simoa TM and Ella TM platforms. A, Association of age with NfL concentration (pg/ml, shown in logarithmic scale) in serum determined by Ella TM (light gray) and Simoa TM (dark gray) platforms were estimated using the linear regression with 203 samples (b = 0.18, p = 0.002, r 2 = 0.045 in Simoa TM and b = 0.21, p < 001, r 2 = 0.057 in Ella TM ). B: Comparison of NfL levels (pg/ml, shown in logarithmic scale) in serum for HCs and MS patients, obtained by the Simoa TM (dark gray, left) and the Ella TM (light gray, right) instruments. Serum NfL levels were higher in RRMS patients than in HCs (p = 0.021 and p < 0001, respectively), higher in active MS than in inactive MS (p = 0.0080 and p = 0.0356, respectively), higher during relapses than in patients with a stable disease (p = 0.0153 and p = 0.0373, respectively), and lower in RRMS than in PMS patients (p = 0.0007 and p = 0.0021, respectively) (*p < 05, **p < 01, ***p < 001, ****p < 0001). C: Association of EDSS with NfL concentration (pg/ml, shown in logarithmic scale) in serum determined by Simoa TM (left, dark gray boxplots) and Ella TM (right, light gray boxplots) platforms were estimated using linear regression with 203 samples (b = 0.83, p = 0.026, r 2 = 0.026 in Simoa TM and b = 0.96, p = 0.015, r 2 = 0.031 in Ella TM ).

Discussion
Blood NfL is a biomarker associated with several clinical parameters in MS. 10 We showed that both Ella TM and Simoa TM platforms offer excellent sensitivity, detecting serum NfL concentrations in the picogram range, Simoa TM platform offering the lowest inter-assay imprecision at low analyte levels. A limitation of our study was restricting the analysis to three runs, making the inter-assay CV harder to accurately define. The two systems use different methods to determine intra-assay CV, using technical duplicate or triplicate readings, preventing direct comparison. However, Simple Plex TM runs the samples in parallel at the same time, assuring the exact same conditions for replicate analysis, an advantage over the Simoa TM platform that processes serial measures. Moreover, calibrators are directly integrated in the Simple Plex TM cartridges, providing best calibration for each run.
The main finding of this study is the demonstration of a concordance between NfL levels measured using both platforms, even at low levels in the HC group. This is potentially the result of using the same anti-NfL antibody and of heterophilic blockers limiting potential cross-reaction between anti-NfL antibody and antibodies in the serum for both platforms. However, we observed significant differences in absolute biomarker concentrations between these two instruments. Using different calibrators (naturally derived bovine NfL for Ella TM and a recombinant human NfL for Simoa TM ) has been associated with differences in NfL measure and could explain the differences in absolute values obtained by both assays. 2 The NfL raw concentrations measured by Simoa TM were globally lower vs Ella TM , as confirmed by the Bland-Altman plot. The "spike recovery" reported in the data sheet of the two assays is 68% for Simoa TM NfL kit and 108% for Simple Plex TM NfL, suggesting that Simoa TM could underestimate the values of NfL by 17% compared to Ella TM due to a greater effect of the serum matrix than in the Simple Plex TM method. Passing-Bablok allowed the bias to be evaluated over the entire measurement range and the linear test shows that the data are linear (p = 0.57). Thus, it is possible to apply a correction factor 2.917. Therefore, Ella TM technology, with the advantage of small footprint and a robust and cheaper platform, represents a reliable substitute for Simoa TM to measure serum NfL.
Moreover, we demonstrate that serum NfL levels determined by Ella TM show the same properties, concerning correlation of serum NfL with age, EDSS and disease activity. This is crucial, since future studies with Ella TM can directly resume previous results already published using Simoa TM . However, NfL cannot be used in combination with other brain biomarkers that remain unavailable on this platform, such as glial fibrillary acidic protein, available on the Simoa TM platform which currently has a larger range of biomarkers.
Although the Ella TM platform showed a greater interassay variation compared to Simoa TM , it seems an attractive choice for routine quantification of serum NfL considering the reduced cost, high performance and small footprint while maintaining a high concordance with Simoa TM . Serum NfL biomarker can be quantified using automated Ella TM instrument to reliably and rapidly monitor disease activity and treatment in MS as well as in many other CNS pathological conditions, thus optimizing quality of care.

Acknowledgments
This work was conducted using data from the Observatoire Franc ßais de la Scl erose en Plaques (OFSEP) which is supported by a grant provided by the French State and handled by the "Agence Nationale de la Recherche," within the framework of the "Investments for the Future" program, under the reference ANR-10-COHO-002, by the Eug ene Devic EDMUS Foundation against multiple sclerosis and by the ARSEP Foundation." The authors thank Sarah Kabani (BESPIM, CHU de Nîmes) for substantive editing of the manuscript.

Conflict of interest
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. Audrey Gauthier: nothing to disclose; S ebastien Viel: nothing to disclose; Magali Perret: nothing to disclose; Sabine Laurent-Chabalier: nothing to disclose; Marc Debouverie: nothing to disclose; Gilles Edan: consultancy and lecturing fees from Bayer-Schering, Biogen, LFB, Merck, Novartis, Roche, Sanofi; research grants from Bayer, Biogen, Genzyme, Mercks, Novartis, Roche, Teva, and the ARSEP foundation. He has been principal investigator in phase 2 and 3 clinical studies conducted by Bayer, Biogen, Merck, Novartis, Sanofi-Aventis Teva, and 4 academic programs (programmes hospitaliers de recherche clinique, PHRC) on MS sponsored by Rennes University Hospital; Sandra Vukusic: grants, personal fees and non-financial support from Biogen, grants and personal fees from Geneuro, grants, personal fees and non-financial support from Genzyme, grants and personal fees from Medday, grants, personal fees and non-financial support from Merck-Serono, grants, personal fees and non-financial support from Novartis, grants, personal fees and non-financial support from Roche, grants, personal fees and nonfinancial support from Sanofi, personal fees from Teva; Christine  Thouvenot : consulting and lecturing fees, travel grants or unconditional research support from the following pharmaceutical companies: Actelion, Biogen, Celgene, Genzyme, Merck Serono, Novartis, Roche, Teva pharma.

Funding information
The study was funded by CHU de Nimes and has also been supported by a grant provided by the French State and handled by the "Agence Nationale de la Recherche," within the framework of the "Investments for the Future" programme, under the reference ANR-10-COHO-002 Observatoire Franc ßais de la Scl erose en plaques (OFSEP).

Supporting Information
Additional supporting information may be found online in the Supporting Information section at the end of the article.
Supplementary Figure S1. Comparison of Simoa TM and Ella TM platforms at low serum NfL levels. A: Repeatability tests of both platforms using samples from one HC and from one RRMS patient tested 30 times. For Simoa TM , average NfL concentrations were 6.55 pg/ml and 14.22 pg/ml and CVs were 11.3% and 8.1%, respectively. For Ella TM , average serum NfL concentrations were 8.60 pg/ml and 38.38 pg/ml and CVs were 12.8% and 8.9%, respectively, as indicated on the graph. B: Spearman correlation (r) between NfL concentration values obtained by the Ella TM compared to the Simoa TM instruments in a cohort of 29 HCs (r = 0.76, p < 0.0001).