Chemotherapy-induced polyneuropathy and neurofilaments

Full Text

Introduction

Platinum-based drugs and taxanes are widely used in the treatment of malignant neoplasms, especially solid tumors. However, their use is often associated with chemotherapy-induced polyneuropathy (CIPN), the pathogenesis of which is complex and can be exacerbated by underlying diseases, paraneoplastic processes, and endogenous toxins. This complication can persist and progress even after completion of therapy, significantly worsening patients’ long-term quality of life [1–5].

The diagnosis of CIPN is primarily based on symptom analysis and neurological examination, highlighting the need for objective and quantitative assessment methods. However, nerve conduction study (NCS) can only confirm damage to thick fibers, while invasive skin biopsy remains the gold standard for diagnosing small fiber neuropathy. These methods are labor-intensive and not always applicable in routine oncological practice. This stimulates the search for simpler and more accessible methods for diagnosing and predicting CIPN.

Particular interest lies in biomarkers capable of objectively reflecting the degree of peripheral nervous system damage [6–8], with neurofilaments being the most promising among them. They shape the neuronal cytoskeleton and are expressed exclusively in neurons. Modern neuroscience focuses greatest interest on neurofilament light chain (NfL) and peripherin in peripheral nervous system lesions, as they unlike medium and heavy subunits are decisive in polymerizing peripheral neuron filaments and shaping their structure. During neuronal and axonal damage, they are released into the extracellular space and blood [9].

Recent studies have shown that serum NfL levels increase during chemotherapy (CT) with platinum and taxanes, correlating with neuropathy severity [6–14]. Meanwhile, concentration returns to normal several months after treatment completion, potentially indicating halted damage progression and possible regeneration [15, 16]. However, unlike peripherin, NfL level may increase.

Peripherin, as a type III intermediate filament protein, plays a crucial role in neuronal growth and stability, axonal transport, and myelination. It demonstrates greater specificity in peripheral nervous system lesions, being expressed exclusively in peripheral neurons, and has recently emerged as a specific marker of axonal damage [17, 18]. In the study by S. Keddie et al., peripherin levels exceeded those of NfL in peripheral nervous system lesions [19]. However, its clinical application is limited by detection challenges, necessitating further research [20].

The study aimed to evaluate serum levels of NfL and peripherin biomarkers using ELISA in patients receiving chemotherapy for solid malignant neoplasms.

Materials and Methods

The study was conducted in 2023–2024 in accordance with the ethical principles of the World Medical Association’s Helsinki Declaration and approved by the Independent Ethics Committee of the Immanuel Kant Baltic Federal University Center for Clinical Research (Protocol No. 35 dated October 27, 2022).

The study included 66 patients aged >18 years of age with newly diagnosed solid tumors, scheduled for planned neurotoxic CT using platinum-based and taxane-based treatment regimens. Exclusion criteria were complaints indicating peripheral nerve damage at the time of study inclusion; concomitant diseases and pathological conditions that might induce polyneuropathy (PNP); and the use of medications with potential neurotoxic effects. After CT, 51 patients were followed up; 15 patients withdrew from the study at the second visit and were subsequently excluded from the final analysis.

Platinum-based CT was administered to 35 patients (68.6%): oxaliplatin — 23 patients with gastrointestinal tract (GIT) cancer, cisplatin — 6 with respiratory system cancer and 6 with pelvic cancer; taxane-based chemotherapy — 16 patients (31.4%) (paclitaxel/docetaxel, predominantly for pelvic cancer).

All patients underwent neurological examination assessing neurotoxicity using the National Cancer Institute Common Terminology Criteria for Adverse Events Version 5.0 (NCI-CTCAE) scale and neuropathy severity using the Neuropathy Disability Score (NDS); NCS employing the SRAR (Sural/Radial Ratio) index and sural nerve action potential amplitude as highly sensitive and specific neurophysiological markers [21]; and intraepidermal nerve fiber density evaluation (IENFD) [22]. NfL and peripherin concentrations were measured in blood serum at baseline and at 4.50 ± 1.02 months post-CT. Blood was collected followed by centrifugation at 1000g for 20 minutes at room temperature to obtain serum. The obtained serum was carefully aspirated to avoid cellular elements and aliquoted into 500 μL volumes in low-adhesion Eppendorf tubes to minimize protein loss. Aliquoted samples were immediately frozen at –80°C and stored until analysis. The entire preanalytical processing was strictly controlled, with documentation of time intervals and storage conditions to ensure result reproducibility.

Levels of NfL and peripherin were determined using solid-phase sandwich ELISA, employing Cloud-Clone Corp. test systems on a Bio-Rad 680 microplate photometer (Bio-Rad Laboratories). Optical density was measured at 450 nm.

Statistical processing was conducted using StatTech v. 4.2.8 (Stattech) and GraphPad Prism v. 9.3.1 (Insightful Science) software packages. Quantitative indicators were evaluated for normality of distribution using the Shapiro-Wilk test.

In cases of non-normal distribution, quantitative data were described using the median (Me) and lower and upper quartiles [Q₁; Q₃].

Group comparisons employed the Kruskal–Wallis, Wilcoxon, Pearson’s χ² tests, and ROC analysis. Correlation analysis between two quantitative indicators was performed using Spearman’s rank correlation coefficient (for non-normally distributed indicators). A predictive model characterizing the dependence of a quantitative variable on factors was developed using linear regression. Differences were considered significant at p < 0.05.

Results

A total of 51 patients completed the study: 23 with GIT cancer, 22 with pelvic cancer, and 6 with respiratory system cancer; the majority were women (n = 33; 65%). The median age was 60 years, with 28 patients (55%) over 60 years old. The average disease duration from diagnosis to study inclusion was 6.06 months.

CIPN developed in 36 cases (70%). The neurological presentation corresponded to classic distal symmetric polyneuropathy. Sensory disturbances were the predominant clinical component, documented in 36 patients (71%). Motor deficits were rare (n = 3; 6%), manifesting as diminished or absent Achilles and brachioradial reflexes. Autonomic dysfunction symptoms (e.g., sweating abnormalities) were observed in 12 patients (24%).

Patients were stratified by neurotoxicity grade using the NCI-CTCAE scale according to primary tumor site (Table 1).

 

Table 1. Assessment of the severity of neurotoxicity after chemotherapy according to the NCI-CTCAE scale depending on cancer site, n (%)

Neurotoxicity severity	Number of patients post-CT (n = 51)
	GIT cancer	pelvic cancer	respiratory system cancer
Total	23	22	6
No manifestations	2 (8.7)	6 (27.3)	1 (16.7)
1	12 (52.2)	7 (31.8)	2 (33.3)
2	9 (39.1)	9 (40.9)	3 (50.0)
3	–	–	–
4	–	–	–

 

The presented data indicate that the groups predominantly comprised patients with grade 1–2 NCI-CTCAE neurotoxicity. Nine patients showed no signs of neurotoxicity; patients with pelvic cancer prevailed. The highest frequency of neurotoxicity (any grade) was observed in GIT cancer (91.3%), and the lowest in the pelvic cancer (72.7%). No cases of grade 3–4 severity were recorded. Analysis of PNP severity according to the NDS scale, depending on tumor site and corresponding therapy, is presented in Table 2.

 

Table 2. Severity of PNP after chemotherapy according to the NDS scale depending on cancer site, n (%)

Neuropathy severity	Number of patients post-CT (n = 51)
	GIT cancer	pelvic cancer	respiratory system cancer
No (0–4 points)	6 (26.1)	2 (33.3)	6 (27.3)	p = 0.535
Moderate (5–13 points)	12 (52.2)	3 (50.0)	15 (68.2)
Moderate (5–13 points)	5 (21.7)	1 (16.7)	1 (4.5)

 

PNP was mostly moderate in all groups, particularly in pelvic cancer (68.2%). Severe PNP was more frequent in GIT tumors, though this was not statistically significant.

Serum analysis before and after CT revealed undetectable peripherin levels. The median serum NfL level increased significantly from 94.22 [33.24; 152.36] pg/mL before CT to 143.98 [87.91; 222.47] pg/mL after treatment (p < 0.0003; Fig. 1).

 

Fig. 1. Distribution of NfL before and after CT.

 

NfL level changes showed dependence on the patient’s sex: pre-CT levels were substantially lower in men (< 18.93 pg/mL) compared to women (4.6-fold difference), but post-treatment analysis revealed a marked increase to 211.92 pg/mL specifically in male patients (p < 0.001; Fig. 2).

 

Fig. 2. Changes in Nfl concentrations depending on the sex.

 

Analysis of NfL level changes depending on the tumor site showed the most pronounced increase in NfL (~5-fold) specifically in patients with gastrointestinal cancer (p = 0.001), unlike other sites (Table 3).

 

Table 3. Changes in NfL levels (pg/mL) before and after CT depending on cancer site, median [Q₁; Q₃]

Cancer site	Before CT	After CT	р*
GIT (n = 23)	41.13 [14.33; 87.53]	200.16 [107.60; 292.54]	< 0.001
Respiratory system (n = 6)	0.00 [0.00; 1.41]	133.43 [89.37; 139.66]	0.250
Pelvis (n = 22)	88.73 [40.44; 142.31]	112.40 [84.54; 178.70]	0.229

 

To analyze the sensitivity and specificity of NfL as a diagnostic marker for CIPN, ROC analysis was performed (Fig. 3). The sensitivity and specificity of the obtained predictive model were 90.9% and 35.7%, respectively. The NfL threshold after CT at the cut-off point using Youden’s index was 95.37 pg/mL (AUC = 0.526; 95% CI 0.321–0.731; p = 0.803).

 

Fig. 3. ROC analysis of sensitivity and specificity of NfL in CIPN.

 

No significant association was found between post-chemotherapy NfL levels and prediction of CIPN. However, in CIPN patients (n = 36), serum NfL concentrations differed significantly (p = 0.001) before versus after chemotherapy: 56.64 [13.01; 120.82] vs. 133.23 [68.58; 266.19], respectively.

No statistical relationship was observed between NfL levels and SRAR index (r = –0.259; p = 0.112) or gastrocnemius nerve action potential amplitude (r = –0.163; p = 0.322), but a negative correlation was found with IENFD (r = –0.416; p = 0.012; Fig. 4).

 

Fig. 4. Negative correlation between NfL concentration and IENFD.

 

Discussion

Results of a prospective study showed a significant increase in serum NfL levels after chemotherapy (p < 0.0003); the median level was 143.98 pg/mL, particularly in male patients (p < 0.001) and in those with GIT cancer treated with platinum-based regimens (p < 0.001). Our data are generally consistent with the results of the study by S.-H. Kim et al., where oxaliplatin was also used in patients with GIT cancer [10]. The observed discrepancy of higher NfL levels in their study (mean 373.4 pg/mL) can be explained by the lack of strict selection criteria, inclusion of patients with risk factors for peripheral neuropathy, and the use of a more sensitive analytical method on the Simoa platform.

Ultrasensitive single molecule array technology has enabled the detection of lower biomarker concentrations in serum compared to ELISA. However, this method has higher costs and limited availability in most medical institutions. The null result for peripherin in our study may be explained by concentrations below the detection threshold of ELISA using digital ELISA, unlike in previous studies [19], or by methodological limitations, for example, the selected antibodies in ELISA might not have captured serum peripherin.

Our data did not confirm changes in NfL levels in female patients with pelvic cancer, which contradicts the results of other studies demonstrating the effect of taxane therapy on this indicator [6, 11]. This discrepancy may be associated not only with different research methodologies but also with the study design, specifically the time points, as well as the distinct mechanism of action of the neurotoxic agent, since platinum-based drugs primarily affect the neuron itself, while taxanes target microtubules [2, 23].

We also found no correlation with NCS data, unlike other studies [10], which may be explained by the absence of a more pronounced degree of neurotoxicity according to the NCI-CTCAE scale and lower NfL concentrations in patients with grade 0–2, compared to previous research. NfL is a good screening marker for neuropathy due to its high sensitivity (85.7%), but it has low specificity (45.0%) and an AUC close to 0.5. Measuring NfL levels for severity monitoring may offer a significant advantage over NCS. Blood is frequently tested in cancer patients undergoing chemotherapy, unlike NCS, which is challenging to implement in daily practice, especially since electrophysiological changes may lag behind clinical manifestations and identify established axonal damage, whereas NfL levels can dynamically reflect this process. Thus, serum NfL can serve as a diagnostic biomarker for CIPN, as its significant association with this condition has been confirmed both in our study (p = 0.001) and in other authors’ work. However, more extensive studies are required to evaluate its prognostic value.

Our previous data demonstrated a significant reduction in IENFD after CT in patients with GIT cancer [22]. In the present study, we found that this reduction shows a significant negative correlation with increased NfL concentration. This finding is well explained by the cytotoxic mechanism of chemotherapeutic agents, which leads to nerve fiber degeneration. However, the absence of such correlation between NfL and IENFD in patients with idiopathic small fiber neuropathy, as demonstrated in the study by R. Baka et al. [24], indicates a distinct pathophysiology. This contradiction underscores that the relationship between structural IENFD and biochemical (NfL) markers of nerve damage is not universal and depends on the etiology and mechanisms of neuropathy.

Our study had several limitations: small sample size (n = 51); lack of a control group; short follow-up period with two control points; absence of long-term data on NfL changes after CT completion.

Conclusion

Post-CT NfL is a promising, non-invasive, yet insufficiently specific CIPN biomarker. Its clinical application requires further investigation alongside clinical assessment and other biomarkers. Monitoring NfL levels during CT enables quantitative serial measurements with minimal patient burden and real-time tracking of ongoing neuronal damage, facilitating early CIPN prediction algorithms for personalized therapy. Extensive validation studies are needed to confirm the value of these data for both clinical trials and general practice. ELISA is less sensitive to changes in NfL and peripherin concentrations, while the implementation of highly sensitive analytical methods such as the Simoa platform immunoassay opens new avenues for its application.

About the authors

Olga A. Tikhonova

Immanuel Kant Baltic Federal University

Author for correspondence.
Email: annaly-nevrologii@neurology.ru
ORCID iD: 0000-0002-1796-0193

Cand. Sci. (Med.), senior lecturer, Department of psychiatry and neurosciences

Russian Federation, Kaliningrad

Evgeniia S. Druzhinina

Pirogov Russian National Research Medical University

Email: annaly-nevrologii@neurology.ru
ORCID iD: 0000-0002-1004-992X

Cand. Sci. (Med.), Associate Professor, Department of neurology neurosurgery and medical genetics named after acad. L.O. Badalyan, Institute of Neurosciences and Neurotechnology

Russian Federation, Moscow

Dmitry S. Druzhinin

Sechenov First Moscow State Medical University (Sechenov University)

Email: annaly-nevrologii@neurology.ru
ORCID iD: 0000-0002-6244-0867

Dr. Sci. (Med.), deputy chief physician for research and innovation, University Clinical Hospital No. 3

Russian Federation, Moscow

References

Supplementary files