The investigation included 70 people with genetically confirmed 5q SMA recruited from centres in Germany and Belgium, together with comparison groups of 31 patients with ALS, 31 with Alzheimer’s disease and 31 disease controls who had no evidence of neurodegeneration based on clinical evaluation, cerebrospinal fluid analysis and neuroimaging. SMA participants underwent repeated blood sampling and clinical assessment. The main follow-up points were 9–16 months and 22–26 months after baseline, while an exploratory cohort of eight Halle patients was sampled at approximately 50–58 months but had no corresponding baseline data. All 70 patients with SMA received nusinersen or risdiplam at follow-up: 45 received nusinersen and 25 received risdiplam. Because the study was retrospective, relied on available observations at each time point and had incomplete sampling at some visits, its longitudinal findings were considered exploratory rather than confirmatory.

Serum p-tau181 and p-tau217 were quantified with single-molecule array immunoassays, while neurofilament light chain was measured with a separate automated assay. Group differences were evaluated with multiple linear regression adjusted for age and sex, using SMA as the reference category. Concentrations of p-tau181 differed across the diagnostic groups in the overall model, with p less than 0.001. The adjusted coefficient for disease controls was −0.48, with a 95% confidence interval of −0.81 to −0.15, indicating lower p-tau181 in controls than in SMA. Levels were higher in ALS, with a coefficient of 0.99 and 95% confidence interval of 0.58 to 1.39, and in Alzheimer’s disease, with a coefficient of 0.56 and 95% confidence interval of 0.11 to 1.00. For p-tau217, no difference was detected between SMA and disease controls: the coefficient was −0.09, with a 95% confidence interval of −0.22 to 0.05. Levels of p-tau217 were higher in both ALS and Alzheimer’s disease than in SMA. The results show that the two phosphorylated tau species did not behave as interchangeable measures across these conditions.

Centre-specific analyses added nuance to the pooled comparisons. In the Ulm cohort, p-tau181 was lower in disease controls than in SMA, with a coefficient of −0.52 and 95% confidence interval of −0.73 to −0.31, but no clear difference was detected between SMA and either ALS or Alzheimer’s disease. In Leuven, the SMA-control difference was not detected, with a coefficient of −0.20 and 95% confidence interval of −0.56 to 0.17, while p-tau181 was higher in ALS and Alzheimer’s disease than in SMA. The Halle baseline group contained only two patients and was not statistically tested. For p-tau217, neither Ulm nor Leuven showed a clear SMA-control difference, although both cohorts had higher concentrations in ALS and Alzheimer’s disease than in SMA. These variations mean the overall diagnostic-group pattern should not be interpreted as a uniform, validated threshold for distinguishing SMA from other conditions.

Within-person comparisons produced the clearest longitudinal result. Among 56 patients with measurements at both baseline and follow-up 2, p-tau181 decreased by a median of 1.80 pg/mL, with an interquartile range of −5.90 to 2.10 pg/mL; the paired Wilcoxon signed-rank test gave p = 0.017. No relevant change was detected between baseline and follow-up 1 among 29 paired patients, when the median difference was 0.17 pg/mL, or between follow-up 1 and follow-up 2 among 28 paired patients, when it was 0.00 pg/mL. P-tau217 showed no relevant change at any interval. An unpaired analysis using all available samples at each time point did not detect significant differences among baseline, follow-up 1 and follow-up 2, showing how the conclusion depended on which patients could be paired. The eight-patient exploratory group had p-tau values within the upper range of earlier observations, but its size and lack of baseline samples prevented statistical testing or paired analysis.

Linear mixed-effects models, with repeated observations accounted for and estimates adjusted for age, sex and other applicable variables, linked both p-tau species to several clinical measures. Higher p-tau181 and p-tau217 concentrations were associated with higher Hammersmith Functional Motor Scale Expanded scores, higher Revised Upper Limb Module scores and higher forced vital capacity as a percentage of the predicted value; all corresponding false-discovery-rate-adjusted p values were below 0.001. In the Leuven risdiplam-treated subgroup, where the Motor Function Measure 32 was available, both biomarkers were also positively associated with total and domain scores, although those analyses were exploratory because they were restricted to one centre and a limited sample. Concentrations were lower with longer disease duration, and ambulatory participants had higher levels than non-ambulatory patients. Thus, higher circulating p-tau was linked to better preserved motor and respiratory function, rather than simply indicating more severe impairment.

Creatine kinase was positively associated with both p-tau181 and p-tau217, with false-discovery-rate-adjusted p values below 0.001, whereas neither biomarker showed a significant association with neurofilament light chain after correction for multiple testing. The authors considered this pattern consistent with the possibility that peripheral p-tau in SMA is more closely related to denervated or remodelling muscle than to ongoing neuroaxonal injury. They also proposed that p-tau181 and p-tau217 may capture different aspects of motor-neuron and muscle pathology. These interpretations remain mechanistic hypotheses, however, because the study did not directly measure tau release from tissue, analyse p-tau in cerebrospinal fluid alongside serum, or use imaging and electrophysiology to locate its biological source. The absence of an association with neurofilament light chain should therefore be viewed as an observed result within this cohort, not direct proof of a peripheral origin.

Exploratory multivariable regression models suggested a possible prognostic association while showing substantial uncertainty. Higher baseline p-tau181 was associated with a greater decline in the Hammersmith Functional Motor Scale Expanded by follow-up 2, with a coefficient of 1.79 and 95% confidence interval of 0.59 to 2.99. The corresponding estimate for p-tau217 was 1.60, with a 95% confidence interval of −0.13 to 3.33, indicating a less certain association. Neither biomarker was clearly associated with change in the Revised Upper Limb Module; in those exploratory models, male sex and older age accounted for more of the observed variation. Motor scores remained broadly stable during follow-up, and ceiling effects in adult SMA may have limited the sensitivity of upper-limb analyses. The p-tau181 association was therefore characterized as preliminary and hypothesis-generating, and it requires independent confirmation before it can be regarded as a prognostic marker.

The findings also raise questions about the interpretation of blood p-tau measurements in people with motor neuron disorders. Because p-tau181 was higher in SMA and ALS than in non-neurodegenerative controls, elevated circulating p-tau may occur without concurrent Alzheimer-type pathology. The comparison is strengthened by the study’s exclusion of ALS participants with positive or unknown amyloid-tau-neurodegeneration status and by the cerebrospinal-fluid biomarker criteria used to define the Alzheimer’s disease group. At the same time, the retrospective design, modest sample size, few time points, advanced disease stage in many participants, missing body-mass-index data and absence of paired cerebrospinal-fluid measurements limit how broadly the results can be generalized. Future studies involving earlier-stage SMA, more consistent serial sampling, cerebrospinal fluid and serum collected together, and imaging or electrophysiological measures would be needed to determine whether p-tau181 adds reliable value to clinical assessment or whether it mainly provides information about muscle-related processes in adult SMA.

Subject of Research: Serum phosphorylated tau biomarkers and clinical measures in 5q spinal muscular atrophy

Article Title: Serum p-tau181 and p-tau217 levels in adolescents and adults with 5q spinal muscular atrophy

Article References: Stapf, C. D., Uzelac, Z., Iterbeke, L., Guthknecht, A., Barba, L., Strube, D., Kölsch, A., Kleeberg, K., Kendzierski, T., Roos, A., Steinacker, P., Abu-Rumeileh, S., Claeys, K. G., Weishaupt, J., Mensch, A., & Otto, M. (2026). Serum p-tau181 and p-tau217 levels in adolescents and adults with 5q spinal muscular atrophy. Journal of Neurology, 273(11), Article 672. https://doi.org/10.1007/s00415-026-14194-1

Image Credits: AI Generated

DOI: 10.1007/s00415-026-14194-1

Keywords: spinal muscular atrophy, p-tau181, p-tau217, blood biomarkers, motor function, muscle turnover, denervation, creatine kinase, longitudinal study, tau protein, Serum, levels