The study, led by Gretchen A. McNally and Grace Ji-eun Shin of The Ohio State University Comprehensive Cancer Center, tracked 25 adults with non-Hodgkin lymphoma who were beginning first-line chemotherapy with vincristine-containing CHOP or EPOCH regimens. Rather than conducting an artificial clinical-trial setting, the researchers embedded their measurements directly into routine care, drawing blood and administering questionnaires at the same visits where patients received their infusions. Sixteen participants received CHOP, typically delivered in an outpatient infusion lasting about ninety minutes, while nine received EPOCH, which required roughly five days of inpatient admission with continuous drug infusion over 96 hours. The team followed participants through four to six months of chemotherapy, and a small subset of six survivors for an additional 24 to 42 months after treatment ended.

The centerpiece of the research is neurofilament light chain, or NfL, a structural protein that forms part of the internal scaffolding of nerve axons. When axons are injured, NfL leaks into the surrounding fluid and eventually reaches the bloodstream, where it can be quantified with single-molecule array technology capable of detecting concentrations as low as 0.038 picograms per milliliter. Because NfL rises after virtually any form of neuroaxonal damage, from multiple sclerosis to traumatic brain injury, its interpretation requires careful longitudinal context. In the case of vincristine, the biological logic is compelling: the drug disrupts microtubule assembly and axonal transport, the cellular freight system that keeps long peripheral nerves alive, and the resulting distal axonal injury should release NfL into circulation.

That prediction was borne out dramatically. Using mixed-effects statistical models on log-transformed data, the researchers found that model-estimated plasma NfL rose from a geometric mean of 26.3 picograms per milliliter before the first chemotherapy cycle to 91.6 picograms per milliliter before the second cycle, 107.2 before the fourth, and 94.3 before the sixth. Relative to the pretreatment measurement, these values represent increases of roughly 248 to 307 percent, with every timepoint reaching statistical significance at p less than 0.001. Notably, the rise was not a simple cumulative-dose curve of the kind reported for platinum drugs or paclitaxel; instead, NfL surged early and then plateaued at a sustained elevation, a pattern consistent with prior clinical observations that vincristine neuropathy does not track neatly with total cumulative dose.

Alongside the biomarker measurements, patients completed the EORTC QLQ-CIPN20 questionnaire, a validated 20-item instrument capturing sensory, motor, and autonomic neuropathy symptoms, as well as the Brief Pain Inventory Short Form for pain severity. Clinicians recorded neuropathy grades using the National Cancer Institute’s Common Terminology Criteria for Adverse Events as part of standard toxicity documentation. Patient-reported symptom scores climbed significantly during treatment, with the model-estimated CIPN20 score rising 23.4 percent by the time of the sixth cycle compared with baseline. Clinician-graded CTCAE neuropathy also increased, with grade 1 or higher neuropathy documented in a third of participants by the final on-treatment assessment, though the clinician-recorded changes were smaller and less clearly longitudinal than what patients reported about themselves.

The relationship between the blood biomarker and the lived experience of neuropathy proved more nuanced than a simple one-to-one correspondence. When the researchers pooled all paired NfL and CIPN20 observations across treatment visits, they found a modest but statistically significant association, with a repeated-measures correlation of 0.439 and a p value of 0.020 across 51 paired observations. However, at any single timepoint, the two measures did not correlate reliably, and exploratory lagged analyses of whether early NfL change predicted later symptom worsening were inconclusive in the full dataset. A post hoc sensitivity analysis, excluding one participant with extreme symptom severity, hinted that early NfL concentrations might anticipate later patient-reported symptoms, but the authors are careful to frame this as hypothesis-generating rather than proof of predictive value.

This imperfect alignment between biology and symptom is not a failure of the biomarker; it is a window into the complexity of chemotherapy neurotoxicity. Patient-perceived symptoms may emerge before, during, or after detectable NfL release, reflecting not just structural axonal injury but also altered nerve excitability, immune activation, pain sensitization, and functional adaptation. Emerging mechanistic research suggests that NfL itself may not be a passive bystander: a recent study reported that NfL released after neuronal damage can activate myeloid immune cells and promote neuroinflammation, and preclinical work has implicated astrocyte activation, macrophage infiltration, cytokine release, and NLRP3 inflammasome signaling in vincristine-induced pain. NfL may therefore be best understood as one node in a broader injury-and-immune response rather than a same-day readout of symptom severity.

The study also illuminated the messy realities of real-world oncology that biomarker research must eventually confront. A quarter of participants, six of 25, underwent vincristine dose modifications, including reductions, holds, or discontinuations, for reasons ranging from neuropathy to ileus and constipation. Dose-modified participants showed a numerically larger median rise in NfL, approximately 4.3-fold versus 1.75-fold, but paradoxically reported no median increase in CIPN20 scores, while non-modified participants reported a median increase of six points. These exploratory comparisons did not reach statistical significance, but they suggest that clinical decisions to reduce drug exposure reshape both the biology and the symptom trajectory in ways that any future predictive model must account for. Among the six survivors followed two to three and a half years after treatment, patient-reported neuropathy scores remained elevated, ranging from 20 to 36, and the two participants with the highest long-term symptom burden had also experienced dose modification during therapy.

The authors are candid about the limitations of their pilot. Twenty-five participants from a single center constitute a small cohort with substantial missing data, heterogeneous early sampling because some patients were recruited after treatment had already begun, and no standardized neurologic examination, nerve conduction studies, or quantitative sensory testing to validate the biomarker against objective nerve function. NfL concentrations are also influenced by age, kidney function, body size, and neurological comorbidities, all of which complicate interpretation in a small and demographically diverse sample. The clinician-recorded CTCAE grades came from routine care rather than a standardized assessment protocol, introducing provider-level variability. These constraints mean the study cannot establish NfL as a standalone diagnostic or predictive tool, and the authors explicitly caution that larger studies with standardized pretreatment sampling are needed before NfL can inform routine clinical decisions.

What the study does establish is a methodological foundation and a compelling proof of concept. All prespecified feasibility criteria were met: every enrolled participant contributed early assessment data, retention through the final on-treatment timepoint reached 100 percent among those who remained eligible, and the full sampling schedule fit within the six-month treatment window. Plasma NfL proved sensitive to vincristine exposure, rising early and staying elevated, while patient-reported symptoms rose in parallel without perfectly mirroring the biomarker. The most productive path forward, the researchers argue, is not to replace patient questionnaires or clinician grading with a blood test, but to combine all three into a multimodal monitoring framework in which NfL provides biological context that symptoms alone cannot supply. For a growing population of lymphoma survivors facing persistent neuropathy with few treatment options, the prospect of identifying nerve injury early, before symptoms become disabling, represents a meaningful step toward personalized neuroprotection during cancer therapy.

Subject of Research: Neurofilament light chain as a blood biomarker of vincristine-induced peripheral neuropathy in lymphoma chemotherapy

Article Title: Longitudinal plasma neurofilament light chain and patient-reported outcomes as complementary markers of vincristine-associated peripheral neuropathy in adults with lymphoma: a cohort study

Article References: McNally, G. A., Shin, G. J.-E., Worthen-Chaudhari, L., Schnell, P. M., Flora, L., Krishna, S. S., Voorhees, T., Baiocchi, R., Bond, D., Christian, B., Maddocks, K., Sawalha, Y., & Lustberg, M. B. (2026). Longitudinal plasma neurofilament light chain and patient-reported outcomes as complementary markers of vincristine-associated peripheral neuropathy in adults with lymphoma: a cohort study. Journal of Neurology, 273(10), Article 594. https://doi.org/10.1007/s00415-026-14096-2

Image Credits: AI Generated

DOI: 10.1007/s00415-026-14096-2

Keywords: neurofilament light chain, chemotherapy-induced peripheral neuropathy, vincristine, non-Hodgkin lymphoma, biomarker, patient-reported outcomes, CHOP, EPOCH, neuroaxonal injury, survivorship, CTCAE, longitudinal cohort study

The study, led by Valerio Nicolella and Marcello Moccia of Federico II University of Naples, included 792 people with multiple sclerosis and 302 healthy controls, for a total of 1,094 participants. Crucially, the researchers deliberately selected a clinically stable population: every person with multiple sclerosis included in the analysis had shown no evidence of disease activity in the previous year, with no clinical relapses, no new MRI lesions, and no progression on the Expanded Disability Status Scale, and none went on to develop relapses or radiological activity during follow-up. This design choice matters, because neurofilament light chain levels are known to spike during acute inflammatory attacks. By excluding anyone with recent or subsequent disease activity, the team could focus on what the biomarker says about chronic, slowly accumulating neuroaxonal damage rather than transient inflammatory flares.

Neurofilament light chain is a structural protein of nerve fibres that leaks into the blood when axons are injured. Its concentration rises steadily with age in healthy people, following a non-linear trajectory with a steeper climb after around age fifty, and it also rises in neurological diseases that destroy nerve tissue. Because of this dual sensitivity, the protein must always be interpreted in the context of a person’s age. The Italian team turned that necessity into an opportunity. Using age-specific reference cut-offs derived from large cohorts of healthy individuals, they assigned each participant a so-called NfL-derived age: if the protein level exceeded the threshold for the person’s chronological age band, the individual was classified as biologically older than their birth certificate suggested. The gap between this NfL-derived age and chronological age became the study’s central metric.

The results were striking. Even though absolute neurofilament levels were, on average, higher in the older control group than in the multiple sclerosis group, the relationship between chronological age and NfL-derived age was markedly weaker in the patients. In controls, chronological age strongly predicted the NfL-derived age category, with an odds ratio of 7.23, whereas in people with multiple sclerosis the association was much looser, with an odds ratio of 2.67. When the two groups were compared directly, individuals with multiple sclerosis had a 45 percent higher likelihood of having an NfL-derived age above their chronological age than controls, with an odds ratio of 1.44 and a p-value of 0.007. In other words, disease-related neuroaxonal injury appears to inject variability into what is otherwise a predictable, age-driven trajectory, shifting the biological profile of many patients toward an older brain.

The clinical correlates of this NfL age gap were equally informative. Each one-point increase on the Expanded Disability Status Scale was associated with a higher gap, with a coefficient of 0.147, and patients with a progressive disease course showed a significantly larger gap than those with relapsing disease, with a coefficient of 0.34. Notably, the gap was not associated with disease duration or with the class of disease-modifying treatment a patient was taking, although each additional previous therapy was linked to a slightly larger gap. This pattern suggests that accelerated neuroaxonal ageing in multiple sclerosis tracks disease severity and progression rather than simply the passage of time since diagnosis, echoing earlier MRI-based studies in which the brain age gap estimated from structural scans was higher in patients with more severe disease.

The researchers were careful to position their metric within the broader landscape of biological ageing research. Epigenetic clocks, telomere length, and plasma proteomic signatures capture systemic ageing across the whole organism, whereas neurofilament light chain reflects a brain-specific dimension of biological age. Recent work on plasma proteomics has shown that brain-derived proteins, most prominently neurofilament light chain, trace a distinct ageing trajectory that correlates only weakly with organism-wide ageing markers and with MRI-derived estimates of brain ageing. The Naples team argues that this is a feature rather than a flaw: a blood-based neuroaxonal marker offers a scalable, inexpensive approximation of brain ageing that can be measured in routine clinical practice, without the cost and complexity of brain imaging pipelines or genome-wide methylation assays.

Methodologically, the study was rigorous in its handling of the biomarker’s quirks. Plasma samples were drawn fasting, processed within three hours, and quantified with the Fujirebio Lumipulse immunoassay, with values then converted to SIMOA equivalents using published algorithms so that the age-specific reference cut-offs of Simrén and colleagues could be applied. Because the resulting age categories are ordinal, the researchers converted them into numerical ranks and defined the NfL age gap as the difference between the NfL-derived rank and the chronological rank, a bounded variable ranging from minus three to plus four. Statistical models, including ordered and generalized ordered logistic regression and linear regression adjusted for age and sex, were used to test the associations, and sensitivity analyses restricted to patients without confirmed disability progression yielded consistent results.

The study also has honest limitations that temper the enthusiasm. Controls were older than the patients on average, 50.6 versus 47.4 years, which explains their higher absolute neurofilament levels, although the analysis focused on within-person age gaps rather than raw group comparisons. Information on comorbidities and body mass index was lacking, and both are known to influence circulating neurofilament levels, partly through effects on blood volume and the distribution of the protein outside the nervous system. The broad 18-to-50-year reference band also limits precision, particularly for individuals near its edges, and the reference cut-offs themselves represent high-percentile red-flag thresholds rather than normative ageing curves. Perhaps most importantly, the NfL age gap did not predict the risk of future EDSS progression in Cox models, a null result consistent with emerging evidence that disability accumulation may be tracked more closely by glial markers such as glial fibrillary acidic protein than by neurofilament light chain alone.

Those caveats aside, the implications are considerable. If neurofilament light chain can flag patients whose brains are biologically older than their years, clinicians may gain a tool for risk stratification that goes beyond what chronological age or disability scores alone can offer, potentially informing decisions about when to escalate therapy or when to consider neuroprotective and reparative strategies in clinical trials. The authors emphasize that the metric should be viewed as a dynamic marker of neuroaxonal injury relative to age rather than a definitive brain age clock, and that validation in more diverse populations, including patients with active inflammatory disease, is still needed. Future studies combining neurofilament light chain with GFAP and other emerging biomarkers may refine these ageing models further. For now, the message is provocative and simple: multiple sclerosis leaves a measurable fingerprint of accelerated brain ageing in the blood, and that fingerprint is visible even in patients whose disease appears, by every conventional measure, to be quiet.

Subject of Research: Neurofilament light chain-derived biological brain age and its clinical correlates in multiple sclerosis

Article Title: Evaluating the neurofilament light chain-derived gap between biological and chronological age in multiple sclerosis and controls

Article References: Nicolella, V., Polito, C., Criscuolo, V., Varelli, M., Fasano, S., Fiorenza, M., Pascarella, G., Carotenuto, A., Petracca, M., Lanzillo, R., Castaldo, G., Morra, V. B., Salvatore, F., Terracciano, D., & Moccia, M. (2026). Evaluating the neurofilament light chain-derived gap between biological and chronological age in multiple sclerosis and controls. Journal of Neurology, 273(10), Article 625. https://doi.org/10.1007/s00415-026-14157-6

Image Credits: AI Generated

DOI: 10.1007/s00415-026-14157-6

Keywords: multiple sclerosis, neurofilament light chain, biological age, brain ageing, biomarkers, neuroaxonal injury, disability, progressive disease, plasma biomarkers, neurodegeneration, Journal of Neurology, Evaluating