Clinical Features and Biomarkers to Differentiate Primary and Amyotrophic Lateral Sclerosis in Patients With an Upper Motor Neuron Syndrome

Abstract

Objectives

Differentiation between primary (PLS) and amyotrophic lateral sclerosis (ALS) entails relevant consequences for prognosis and management but is mostly unreliable at early stages. The objectives of the study are (1) to determine the features at onset that could help to differentiate between PLS and ALS, (2) to evaluate the diagnostic performance of an integrated serum biomarker panel, and (3) to identify the prognostic factors for patients presenting with upper motor neuron (UMN) syndrome.

Methods

We selected and retrospectively analyzed the clinical data of patients presenting with UMN syndrome. At the first evaluation, when available, serum biomarkers were measured using ultrasensitive single molecule array.

Results

The study population included 55 patients with PLS and 50 patients with ALS. Patients with PLS presented a longer time to first neurologic evaluation (PLS: 35.0 months, interquartile range [IQR] 17.0–38.0 months; ALS: 12.5 months, IQR 7.0–21.3 months; p < 0.01) and lower levels of neurofilament light chain (NfL) (PLS: 81.8 pg/mL, IQR 38.4–111.1 pg/mL; ALS: 155.9 pg/mL, IQR 85.1–366.4 pg/mL; p = 0.01). Two patients with PLS and 3 patients with ALS carried the C9orf72 expansion. NfL resulted an independent predictor of final diagnosis (odds ratio 1.01, 95% CI 1.00–1.02; p = 0.04) and an independent prognostic factor (hazard ratio 1.01, 95% CI 1.00–1.01; p < 0.01).

Discussion

NfL might help to differentiate patients with PLS from patients with ALS and to predict prognosis in patients with UMN syndrome.

Motor neuron diseases (MNDs) comprise a heterogeneous group of disorders affecting the upper (UMN) and/or the lower motor neuron (LMN).^1,2 The presentation of patients with primary (PLS) and amyotrophic lateral sclerosis (ALS) may be similar, and the differentiation between the 2 requires up to 4 years of follow-up.^3,4 Previous studies have consistently demonstrated a negative prognostic role for increased levels of neurofilament light chain (NfL) in MND; while a putative role for glial fibrillary acid protein (GFAP) and ubiquitin carboxyl terminal hydrolase isozyme L1 (UCHL1) in distinguishing prognostic subgroups has recently been suggested.⁵ Because PLS and ALS exhibit different disease courses,⁶ potentially underpinned by different pathologic mechanisms,⁷ the aim of the present study is to evaluate the clinical features and the diagnostic performance of an integrated panel of serum biomarkers to differentiate PLS from ALS in patients presenting with UMN syndrome.

Methods

We retrospectively examined the first neurologic evaluation of 848 patients with MND performed at our site between January 2003 and December 2021, including in the final cohort patients presenting with UMN syndrome. Patients with PLS and ALS received a final diagnosis according to contemporary diagnostic criteria³ and revised El Escorial criteria, respectively.¹ For each patient, we collected demographic and clinical data, neurophysiologic assessment, MRI abnormalities, and results of genetic screening for C9orf72 hexanucleotide repeat expansion (HRE). Patients were regularly followed to collect information concerning noninvasive mechanical ventilation (NIMV), percutaneous endoscopic gastrostomy (PEG), tracheostomy, death, and time to Milano-Torino staging (MiToS) stage 1 scale, which records changes in functional domains.¹ The study was approved by the San Raffaele Scientific Institute Ethical Committee (Milan, Italy).

Available serum samples were collected at the first evaluation, and single-protein array technology (Simoa; Quanterix Corp) was used to quantify serum GFAP, UCHL1, NfL, and total TAU (tTAU) levels (pg/mL).⁵

Comparison among groups was performed using the χ² test, Mann-Whitney U test, and Kruskal-Wallis test as appropriate for variable type and distribution. Receiver operating characteristic (ROC) curve analysis was performed to calculate the area under the curve (AUC) of serum biomarkers. To assess the predictor role on final diagnosis of the variables differring between groups, a univariate logistic regression analysis was performed, followed by a backward stepwise binary logistic regression analysis to assess the best model for predicting the MND phenotype. Time to MiToS stage 1 was assessed using Kaplan-Meier curves and the Mantel-Cox log-rank test. Finally, a Cox multivariate regression analysis was performed, corrected for well-known prognostic factors,¹ to estimate the proportional hazard ratio of each variable. A full description of the procedure is detailed in eMethods (links.lww.com/WNL/C686).

Results

The final data set consisted of 105 patients presenting with UMN syndrome and included the following: PLS: n = 55 (52.3%); pyramidal ALS: n = 26 (23.8%); classic ALS: n = 4 (3.8%); bulbar ALS: n = 18 (17.1%); and FTD-ALS: n = 2 (1.9%).⁴ Clinical characteristics of patients with PLS and ALS are summarized in Table 1. No difference emerged in sex, age at onset, and MND familiality. There were 3/49 patients with ALS and 2/50 patients with PLS harboring the C9orf72 HRE. Patients with PLS exhibited a significantly longer mean time to the first visit (p < 0.01). At evaluation, the Revised Amyotrophic Lateral Sclerosis Functional Rating Scale score was higher in PLS (p < 0.01), whereas the disability progression rate was lower in patients with PLS (p < 0.01). Moreover, patients with PLS showed lower rates of EMG abnormalities (p < 0.01) and were less likely to develop dementia (p = 0.03). Patients with PLS exhibited a reduced proportion of PEG positioning and NIMV utilization (p < 0.01).

Table 1.

PLS and ALS Patients' Demographics, Clinical and Neurophysiologic Features and NfL Values

Serum biomarkers were tested on 50 patients (PLS: n = 28; ALS: n = 22). The mean concentrations of each biomarker are shown in eTable 1 (links.lww.com/WNL/C686). We observed lower mean concentrations of NfL in patients with PLS (p = 0.01; Table 1, Figure 1A), although no differences emerged when comparing patients with early-onset and mildly affected ALS with long-duration, severely affected PLS (Figure 1B; eFigure 1) and in UCHL1, GFAP, and tTAU mean levels. In a subgroup of patients with PLS, NfL levels did not show statistically significant changes over time (eFigure 2).

Figure 1. Serum NfL Differs in Patients With PLS and ALS.

(A) Serum NfL levels in patients with PLS and ALS. (B) 3D scatter plot figure describing the relationship between NfL levels, disease duration, and functional impairment measured with the ALSFRS-R score in patients with PLS (dots) and patients with ALS (triangles). Although NfL mean values are lower in patients with PLS compared with patients with ALS (A and B), NfL levels were not able to discriminate between patients with early-onset, mildly affected ALS (disease duration >12 months; ALSFRS-R score >42) and long-duration, severely affected PLS (disease duration >12 months; ALSFRS-R score <36) (PLS: 86.6, IQR 22.8–235.9; ALS: 104.4, IQR 80.6–156.1; Mann-Whitney test: p = 0.694). (C) Kaplan-Meier curves of patients with MND presenting with UMN syndrome estimate time to MiToS stage 1 grouped according to the median value of NfL: high levels (above median value: 24.00 months, 95% CI 17.3–30.7, straight line; events: 15) and low levels (below median value: 81.0 months, 95% CI 67.6–94.4, broken line; events: 8). ALS = amyotrophic lateral sclerosis; ALSFRS-R = Revised Amyotrophic Lateral Sclerosis Functional Rating Scale; IQR = interquartile disease; MiToS = Milano-Torino staging; MND = motor neuron disease; NfL = neurofilament light chain; PLS = primary lateral sclerosis; UMN = upper motor neuron.

NfL displayed a good diagnostic performance to discriminate PLS from ALS (AUC 0.768, 95% CI 0.633–0.903), as opposed to GFAP, UCHL1, and tTAU (eFigure 3, links.lww.com/WNL/C686). Because NfL levels below 250 pg/mL offered no clear predictive value (Figure 1, A and B), to select the features better predicting the phenotypic evolution, a univariate (eTable 2) and a multivariate binary logistic regression analysis was performed: our final model (Nagelkerke R² = 0.645) confirmed that NfL serum levels, together with EMG abnormalities and time from symptoms onset to first visit, are independent predictors of diagnosis (eTable 2).

In univariate analysis, time to MiToS stage 1 was different between patients with PLS and ALS (p < 0.01; eFigure 4, links.lww.com/WNL/C686). Moreover, we noticed a significant stratification in time to MiToS stage 1 by dividing the entire cohort for median NfL levels (p < 0.01; Figure 1C) but not for GFAP, tTAU, and UCHL1 (eFigure 5). Multivariate Cox regression models confirmed the independent role of NfL on time to MiToS stage 1 (eTables 3 and 4).

Discussion

To date, the differential diagnosis between PLS and ALS at disease onset remains challenging, although it entails relevant consequences for prognosis and management.^2,6 Few studies have investigated the distinctive features between patients with PLS and ALS⁸ and the diagnostic performance of wet biomarkers.^9,10 In this study, we highlight the clinical and neurophysiologic features and serum biomarkers useful for reaching an early differential diagnosis.

In our cohort, patients with PLS showed at the first neurologic examination a lower rate of bulbar onset, a longer time to first neurologic evaluation, and a slower disability progression rate compared with patients with ALS; these findings are in line with previous reports.^3,6,11 At follow-up, patients with PLS showed a lower rate of PEG and NIMV utilization.⁶ Dementia was less frequent in patients with PLS, which might be indirectly related to the high number of patients with bulbar-onset ALS in our cohort.⁴ The overall frequency of dementia in PLS is in line with previous studies.¹² Of interest, 2 patients with PLS carried the C9orf72 HRE. Although this association is not novel, the percentage we observed is, unexpectedly, similar to patients with ALS,³ which might be related to the relatively small sample size or patient selection criteria.

NfL levels were significantly lower in patients with PLS, consistently with a slower pace of neurodegeneration.⁵ Intriguingly, the apparent decline in NfL levels after year 4 in patients with PLS might point to a time window during which degeneration may be most active. Indeed, previous studies highlighted that NfL concentration is higher in UMN phenotypes exhibiting a fast disease progression.⁵ Although serum NfL displayed a good diagnostic performance in differentiating patients with PLS from patients with ALS, it is important to note that although NfL levels >250 pg/mL provided a high positive predictive value for ALS, lower values should be more carefully interpreted and considered together with EMG abnormalities and other clinical features such as latency to first visit. The finding of low-grade, nonprogressive electromyographic abnormalities in few muscles is not uncommon in PLS³ and might be related to TDP-43 pathology in LMN or trans-synaptic neurodegeneration.¹³ Nevertheless, EMG alone may lack sensitivity in cases of bulbar-onset ALS with prevalent UMN signs¹⁴ and may lead to patients' misclassification.

Univariate and multivariate analysis confirmed that NfL levels at disease onset are strong predictors of prognosis, enhancing the importance of testing serum NfL,⁵ and consistently with previous studies on neurofilament heavy chain.^9,10 Compared with studies focused on NfL in CSF, these differing results may reflect the larger cohort sizes or the type of biomarker assay.¹⁵

Our study benefits from a large series of patients with PLS.^8,11 However, we recognize as study limitations the retrospective nature of this study, the lack of a validation and a larger longitudinal cohorts and of neuroimaging biomarkers.² Further studies are required to confirm our findings.

In conclusion, our study supports that serum NfL may be helpful to early differentiate patients with PLS from patients with ALS, confirming their prognostic role. The results of our study may prove essential in order to shorten the diagnostic delay, assist the neurologist in patient management and help a correct stratification of patients in clinical trials.

Appendix. Authors

Study Funding

This work was supported by the Giovanni Marazzina Foundation.

Disclosure

M. Filippi is Editor-in-Chief of the Journal of Neurology and Associate Editor of Neurological Sciences; received compensation for consulting services and/or speaking activities from Bayer, Biogen Idec, Merck-Serono, Novartis, Roche, Sanofi Genzyme, Takeda, and Teva Pharmaceutical Industries; and receives research support from Biogen Idec, Merck-Serono, Novartis, Teva Pharmaceutical Industries, Roche, Italian Ministry of Health, Fondazione Italiana Sclerosi Multipla, and ARiSLA (Fondazione Italiana di Ricerca per la SLA). All other authors report no competing interests related to this paper. Go to Neurology.org/N for full disclosures.