Clinical utility of the Lumipulse™ immunoassay for plasma neurofilament light chain in multiple sclerosis

Objective : Blood neurofilament light chain (NfL) is robustly associated with disease worsening in multiple sclerosis (MS), though potentially affected by concomitant factors also determining neuro-axonal loss. We investigated the association between plasma NfL (pNfL) measured with Lumipulse™ immunoassay and demographic and clinical variables in MS Methods: This cross-sectional study included 685 people with MS (


Background
Neurofilaments are neuron-specific cytoskeletal proteins that are released after neuroaxonal damage.The availability of newer immunoassays has allowed the measurement of neurofilament light chain (NfL) in the blood, which, in turn, has been associated with clinical features of different neurological conditions [1].However, NfL levels are heavily influenced by age and comorbidities, including kidney function, cardiovascular risk factors, and body mass index (BMI).Also, several reference ranges for blood NfL levels have been proposed, but it is unclear how they compare between different analytic platforms [2,3].
In multiple sclerosis (MS), NfL levels increase with relapses and new MRI lesions, and decrease following effective treatment [4][5][6][7].In studies with a majority of patients with relapsing MS, higher NfL levels were also associated with disability accrual and brain tissue loss over time (i.e., brain atrophy on MRI) [8,9].Most studies in MS used the Single Molecular Array (Simoa™, Quanterix) to quantify NfL in the blood [10,11].Lumipulse™ is an automated system based on a two-step sandwich chemiluminescent enzyme immunoassay (CLEIA), that has been recently implemented to analyze plasma NfL (pNfL), but its clinical utility in MS remains unexplored.Hereby, we aim to evaluate demographic and clinical correlates of pNfL levels measured with Lumipulse™ in MS.

Study design and population
This is a cross-section study conducted at the MS Clinical Unit, of the Federico II University Hospital, Naples, Italy.The study was approved by the Federico II Ethics Committee (332/21).All patients signed informed consent authorizing the use of anonymized data in line with data protection regulation (GDPR EU2016/679).The present study was performed in accordance with good clinical practice and Declaration of Helsinki.
We included consecutive people with a diagnosis of MS [12], from Sep to Nov 2023, regardless of age, expanded disability status scale (EDSS), or treatment status.Patients were asked to participate to the study at their scheduled neurological consultation and blood drawn.

Demographic and clinical variables
Demographic and clinical variables were age, sex, height and weight (from which we calculated the body mass index (BMI)), smoking (ever or never smoker), cardiovascular comorbidities (high blood pressure, high cholesterol, diabetes, atrial fibrillation, stroke, coronary disease and/or related medications).We further classified patients into normal weight (BMI < 25), overweight (BMI 25-30) and obese (BMI > 30).
Relapses and MRI activity (i.e.new/enlarging T2 lesions, T1 gadolinium-enhancing lesions) were recorded on the occasion of clinical consultations.Relapses were defined as new, worsening, or recurrent neurological symptoms that lasted for at least 24 h in the absence of infection, fever, or adverse reaction to a prescribed medication [12].The presence of T1 gadolinium-enhancing or new/enlarging T2 lesions was documented by review of the official MRI report, as determined by a neuroradiologist.We identified patients with active MS based on clinical relapse in the previous year and/or new MRI lesions and/or newly diagnosed MS patients [13].
We also identified patients with EDSS progression by comparing EDSS at the time of the study with EDSS in the previous year.EDSS progression was defined as increase in EDSS by 1.5 points if baseline EDSS was 0, increase in EDSS by 1 point if baseline EDSS was between 1 and 5.5, or increase in EDSS by 0.5 points if baseline EDSS was above 5.5 [14].

Neurofilament light chain measurements
Fasting blood samples were obtained on the same day of the clinical assessments.Plasma and serum samples were taken at the same time in BD Vacutainer™ anti-coagulated ethylene-diamine-tetra-acetic acid (EDTA) and BD SST II Advance Vacutainer™ Serum Analysis tubes, respectively.
Blood samples were centrifuged within 3 h after draw at 1100 rpm × 10 min, aliquoted into polypropylene tubes, and stored at − 80 • C. pNfL and serum NfL (sNfL) levels were evaluated using fully automated chemiluminescent enzyme immunoassay (Lumipulse™ Fujirebio, Tokyo, Japan), consisting of two steps immunoassay method on the Lumipulse G system, analytical information about the assay is available on the manufacturer's website [15].Results were analysed in singlicate, and expressed in picograms per milliliter (pg/mL).
Age-and BMI-adjusted z scores and percentiles were generated from the pNfL measurement using previously validated algorithm for sNfL using Simoa™ (Quanterix) [11].NfL z scores are a measure of deviation from values observed in healthy controls (i.e., NfL z-score of 1 indicates that the NfL concentration deviates of 1 standard deviation from values in the reference database after adjusting for age and BMI).Percentiles express the percentage of the general population expected to have a pNfL value (adjusted for age and BMI) lower than a given value (i.e., NfL 99 percentile indicates that the NfL concentration is higher than 99% of the reference database after adjusting for age and BMI).
To preliminary explore association between pNfL and sNfL, we used a univariable linear regression model including pNfl as dependent variable and sNfL as independent variable.We used a univariable linear regression model to evaluate associations between raw pNfL measurements and derived metrics (pNfL z-score and pNfL percentiles); then, the same model was run including age and BMI as covariates (which correspond to the variables used for computing pNfL z-score and pNfL percentiles).We used a paired t-test to investigate if there were significant differences between pNfL and sNfL levels.
To evaluate clinical utility of pNfL, we used different univariable linear regression models including each demographic and clinical variable, in turn, as independent variable (age, sex, BMI, smoking, presence of cardiovascular comorbidities, MS disease duration, descriptor of disease progression, EDSS, EDSS progression, disease activity, SDMT, SDMT impairment, current DMT group, DMT duration, number of previous DMTs), and pNfL (raw) values, as dependent variable.Then, the same models were run including the full set of covariates (age, sex, disease duration, EDSS, presence of cardiovascular comorbidities, smoking, and BMI).
Results were reported as coefficients (Coeff), 95% confidence interval (95%CI), and p-values, as appropriate.Distribution of variables and residuals was checked using both graphical and statistical methods.Statistical analyses were performed using Stata 15.0 (StataCorp, College Station, TX, USA).Results were considered statistically significant if p < 0.05.

Results
We included 685 MS patients (age 49.7 ± 12.4 years; 65.55% females), with 17.15 ± 10.47 years of disease duration and median EDSS of 3.0 (ranging from 1.0 to 8.0).The mean pNfL value was 14.48 pg/ml (ranging from 2.00 to 247.21 pg/ml).Out of 685 patients, BMI was available for 454, and SDMT for 393.Ten patients had EDSS progression; 17 had recent disease activity (13 were newly diagnosed, 4 showed new radiological and/or clinical activity).Demographic, clinical and treatment variables are reported in Table 1.
When comparing pNfL levels and derived metrics (percentiles [11]) with previously established cutoffs [16], we found that 437 patients (55.91%) had pNfL levels within normal limits and below the 90th percentile, while 150 (21.90%) had pNfL levels above both normal limits and the 95th percentile.Correspondence between different classifications for blood NfL and respective of pNfL measured using Lumi-pulse™ immunoassay are reported in Table 2.

Discussion
Our study confirmed that NfL in the peripheral blood is a biomarker of MS-related clinical features, including clinical/MRI activity, disability and treatment, also when using the Lumipulse™ immunoassay.We have also provided some additional hints on the use of NfL, including factors affecting NfL disease-specificity in MS and comparability between different classification systems.
Our study confirmed clinical utility of pNfL in MS [4,5], also when using Lumipulse™ assay.In our study, we found significant associations between pNfL and disability status (EDSS) and clinical/MRI activity in the past year (that would be within the predicted half-life of blood NfL) [17].On the contrary, we failed to find higher pNfL in the small subset of our population (1.49%) with EDSS progression in the previous year.However, based on the cross-sectional design of the study, we were unable to confirm disability progression on follow-up, which is in turn associated with baseline NfL by previous studies [4,5].Interestingly, we found a gradient of NfL across DMTs, with lower values in people treated with more effective DMTs, thus suggesting more complete disease control.Previous studies have already shown that higher levels of NfL are found in MS patients with clinical activity, radiological activity and disease progression, and have measured decrease in NfL following DMT initiation [5,6].Taken together, blood NfL reflects the occurrence of neuro-axonal loss within different clinical expressions (inflammatory relapses or progressive disability), and related treatment response.
Looking at factors independent from MS that can affect NfL levels, we found a direct association between pNfL levels and age, as in some previous studies [3,9].However, in our population, this association was lost when covariates are added to the regression model, suggesting that age acts as a proxy for other factors not always being accounted for.Considering the raising prevalence of comorbidities with age [18], higher pNfL levels might be driven by comorbidities.In keep with this, we found significant association between NfL and cardiovascular comorbidities.This result provides biomarker-based evidence for wellknown associations between cardiovascular comorbidities and MS outcomes [19,20,21].Besides, these results should encourage caution when interpreting cut-off values/meaningful changes of NfL in the presence of cardiovascular comorbidities [22].
Some previous studies showed lower blood NfL levels in the presence of higher BMI (and therefore larger distribution volume), as well as higher levels in people with cardiovascular risk factors [23][24][25][26].Intriguingly, we observed a significant association with cardiovascular comorbidities, but not with BMI.This could be explained by the fact that NfL levels are inversely related to BMI, whereas they are directly related to cardiovascular comorbidities.The two variables would therefore interact against each other, with our results indicating greater association of pNfL with cardiovascular comorbidities, which would in turn associate with higher BMI, as already suggested by Beydoun et al. in people without neurological diseases [27].Notably, we described for the first time this interaction in people with MS.
Vecchio and colleagues found strong correlation between Simoa™ and Lumipulse™ blood assays for NfL in MS [28], with the caveat of some minor difference whose relevance will need to be further explored.Simrèn and colleagues previously derived specific cutoffs of plasma NfL in a cohort of healthy participants aged between 5 and 90 years using Simoa™ NfL measurements (but did not include BMI) [16].In keep with this, we tried to apply percentiles, z-score and cutoffs developed for Simoa™ NfL measurements to our population, by using the same set of adjusting factors 11,16. .We found a clear correspondence between pNfL levels above the normality cutoff and the 95th percentile (27.40 pg/mL), and between pNfL levels below the normality cutoff and the 90th percentile (9.24 pg/mL), thus suggesting good correspondence between different classification methods.Notably, about two-thirds of our MS population presented with pNfL levels below both the cutoff and the 95th percentile, possibly reflecting an effective disease control even at a subclinical level.However, Simoa™ derived cut-points for sNfL and pNfL cannot be directly applied to Lumipulse™ pNfL, and we preferred raw NfL measurements, pending proper validation using both assays on a population of controls and MS patients.Besides, some previous studies have already preferred the use of raw values [29].
In a random subset of our population, we measured NfL in both serum and plasma.Overall, measurements showed high correlation, but  pNfL had numerically higher values than sNfL, thus suggesting plasma could be preferred to increase statistical power.Very few studies have compared serum and plasma measurements of NfL and have reported conflicting results [30][31][32].For instance, O'Connel and colleagues found lower levels of tau and amyloid in serum than plasma in 8 healthy donors, but failed to find any differences for NfL, thus suggesting low risk of clotting in serum preparations [30].Rübsamen and colleagues have derived a formula for conversion between plasma and serum NfL, with serum levels being numerically higher [31].In our study, regression coefficient is close to 1 (0.95), thus suggesting there is no need for conversion.In the future, comparisons between CSF, serum and plasma are granted, including greater number of samples from MS, other neurological diseases and controls, also to achieve standardization of NfL measurements and utilization in clinical practice [31].
Limitations of our study include the cross-sectional design, with longitudinal assessments being needed to evaluate predictive values of NfL on disease outcomes.Also, we did not have a design suitable for computing cut-off values, which will warrant further investigation.We did not include NfL measurements other than Lumipulse™, because aiming to provide clinical utility rather than comparison (that has recently been performed [28]).Finally, future studies might consider additional outcomes, including more complete cognitive evaluation, patient-reported outcome measures, and imaging [33].
In conclusion, this is the first study showing clinical utility of NfL measured using Lumipulse™ immunoassay in MS.We confirmed that higher values of blood NfL are associated with inflammatory activity and disability, as from neuro-axonal loss.Lumipulse™ is a novel methodology to analyze blood NfL levels, whose utilization in clinical practice will require integration with individual characteristics, including age and presence of cardiovascular comorbidities.

Fig. 2 .
Fig. 2. Clinical utility of plasma neurofilament light chain.Scatter plots (grey shades represent confidence intervals) show the associations between plasma neurofilament light chain (pNfL) levels and age (A), disease duration (B), and EDSS (D).Box plots show the associations between pNfL levels and descriptor of disease progression (C), presence of cardiovascular comorbidity (E), use of DMT (F), DMT group (G), recent activity(H), and EDSS progression (I).Coefficients (Coeff), 95% confidence intervals (95% CI), and p values are presented for significant associations.Ref -Reference; nsnon significant.

Table 1
Demographic, clinical and treatment features.

Table 2
Correspondence between different blood NfL classifications.

Table 3
pNfL and demographic, clinical and treatment correlates.
Table shows coefficients (Coeff), 95% confidence intervals (95% CI), and p values from linear regression models including, in turn, demographic, clinical and treatment features.as independent variable, and pNfL levels, as dependent variable; in the adjusted models covariates were age, sex, disease duration, EDSS, presence of cardiovascular comorbidities, smoking, and BMI.Significant results (p < 0.05) are reported in bold.