NfL is a neuronal cytoskeletal protein released with axonal damage (10). NfL was initially described as a marker of neurodegenerative diseases (11), and later of inherited neuropathies, and CIPN in animal models (10, 12).

Recently NfL has been studied in adults with CIPN. A prospective study of 43 patients receiving oxaliplatin evaluated serum NfL and CIPN severity by nerve conduction studies throughout therapy (13). Mean serum NfL levels increased over time, with significant differences in serum NfL between low grade (0–2) CIPN vs. high grade (≥3) CIPN at 6 months, and a cut-off of 195 pg/ml being 80% sensitive and 86.2% specific to identify high grade CIPN (13).

BDNF is a protein that promotes neuronal survival (14, 15). BDNF was associated with CIPN in 25 bortezomib-treated multiple myeloma patients evaluated for CIPN using the reduced Total Neuropathy Score (TNS-r) before and during therapy (16). Eight participants who developed CIPN had lower mean BDNF levels (2.16 ± 0.72 vs. 4.62 ± 0.61 ng/ml, p = 0.007), and were more likely to have a reduction from baseline BDNF (−1.67 ± 0.67 vs. 0.41 ± 0.71, p = 0.02) than those without CIPN (16). Similarly in 91 multiple myeloma patients treated with bortezomib or thalidomide lower BDNF levels during treatment were associated with CIPN by Common Terminology Criteria for Adverse Events (CTCAE), and a cut-point of 9.11 ng/ml was 76% sensitive and 71.4% specific to identify CIPN (17). Correlation between BDNF levels and CIPN by TNS-r was also reported in 22 non-Hodgkin lymphoma patients treated with vincristine (18).

Single nucleotide polymorphisms (SNPs) in BDNF may also confer increased risk for CIPN in individuals with Met/Met or Val/Met vs. Val/Val genotype (17–20). This association has been shown in bortezomib (16), and taxane-treated patients (20). A study of 35 breast cancer patients treated with taxanes, however, found the risk may be due to increased prevalence of baseline neuropathy, with no difference in prevalence of CIPN between genotypes when excluding patients with baseline neuropathy (21). There was also no association between the met-BDNF genotype and CIPN in 22 non-Hodgkin lymphoma patients treated with vincristine (18).

NGF is a protein that also regulates neuronal survival (22). In 23 cancer patients receiving taxanes or platinums NGF levels decreased after four to six cycles of chemotherapy, and decline was associated with severity of CIPN by nerve conduction studies (23). Similarly in 129 plasma samples from 34 women treated for cervical cancer with paclitaxel and cisplatin, decrease in NGF from baseline was associated with CIPN severity by TNS (24). A prospective study of 45 patients with hematologic malignancies treated with bortezomib, thalidomide, or vincristine, also found there was a significant decrease in NGF in participants who developed CIPN symptoms, whereas there was no change in NGF in participants who did not develop CIPN symptoms (25). In contrast, in a study specifically evaluating neuropathic pain in 60 patients treated with platinum or taxane therapy, the 13 patients who developed painful neuropathy had higher NGF levels post-therapy than those without neuropathic pain (26).

Osteopontin is a glycoprotein involved in inflammatory pathways and has been implicated in cancer progression (27). It has been described as a marker of inflammation in other neurologic conditions included multiple sclerosis (28), and has also been implicated in neuronal repair (29). In a study of 50 breast cancer patients treated with taxanes evaluated by the TNS-r lower baseline levels of osteopontin were associated with developing moderate or severe CIPN, and baseline osteopontin levels were inversely associated with the magnitude of change in nerve conduction over time (30).

The immune system has increasingly been implicated in CIPN in animal models (31, 32), but few studies examine the association of immune markers with CIPN in humans. In 67 breast cancer patients treated with taxanes there was a higher peripheral blood neutrophil-to-lymphocyte ratio in patients with CIPN than those without CIPN (33). In an analysis of cytokines in 55 breast cancer patients treated with taxane and platinum chemotherapy, high IFN-γ, IL-1β, and IL-8 and low IL-10 and IL-6 were associated with CIPN symptoms (34).

A recent approach to biomarker discovery in CIPN has included microRNA and exosome sequencing. In a preclinical model, miRNA-124 was associated with CIPN in mice treated with paclitaxel (35). MicroRNA may be a promising marker in humans as well, in cross-sectional analysis of microRNA expression in 84 breast cancer patients treated with paclitaxel, 15 microRNAs were identified with a significant fold change between CIPN and non-CIPN groups, and miRNA-451a was the most discriminatory (36). Mass spectrometry-based proteomic technology is another novel approach to biomarker discovery used to identify protein signatures associated with CIPN. In a study of 17 patients with breast cancer treated with taxanes, 12 protein signatures identified patients who developed CIPN (37).

Taxanes and vinca alkaloids interfere with microtubule function and may lead to CIPN pathogenesis (1, 73), therefore genes encoding microtubule function have been studied as predictors of CIPN sensitivity. An SNP in CEP72 (rs924607) is associated with CIPN in children and adults (38–41). In 48 adults with acute lymphoblastic leukemia (ALL) receiving vincristine, 75% with the TT genotype developed CIPN vs. 44% with CC or CT genotypes (39). In a combined sample of pediatric ALL patients treated with vincristine in two large therapeutic trials, the TT genotype was also associated with an increased risk for CIPN (38). This finding was replicated when measuring CIPN with nerve conduction studies (41), and in a separate cohort of pediatric ALL patients (40). However, other studies evaluating CEP72 alterations did not find associations with CIPN in cohorts of Spanish and Arab patients treated with vincrisitne (74, 75). Additionally, there was no association of CEP72 alterations with CIPN in patients treated with taxanes (43).

A polymorphism in TUBB2A, encoding tubulin, was associated with CIPN in 1,303 European patients treated with paclitaxel (42). However this finding has not been replicated in other studies of taxanes and vinca alkaloids (43, 76). Individual polymorphisms in MAPT (43, 76) and GSK3B (43) have not been associated with CIPN in patients treated with taxanes or vinca alkaloids, however additive polymorphisms in MAPT and GSK3B were associated with patient and clinician reported CIPN in 454 ovarian cancer patients treated with paclitaxel and carboplatin (43). SNPs in cytoskeletal protein genes, ACTG1 and CAPG, have also been associated with CIPN in pediatric ALL patients treated with vincristine (44).

Disturbance in neuronal function through ion channels may also contribute to CIPN, and alterations in these genes have been associated with CIPN sensitivity (1, 73, 77). In 186 Japanese breast and ovarian cancer patients treated with taxanes a SNP in SCN9A, encoding voltage-gated sodium channels, was associated with developing ≥grade 2 CIPN, and predicted CIPN persistence post-treatment (56). In 94 Spanish patients with gastrointestinal cancer treated with oxaliplatin another polymorphism in SCN9A was associated with a lower risk of acute CIPN by neurologic evaluation (57), and in 228 South Indian gastrointestinal cancer patients treated with oxaliplatin it was associated with increased incidence of chronic CIPN (58). In 200 patients with colorectal cancer treated with platinums, polymorphisms in SCN4A and SCN10A that encode voltage-gated sodium channels were associated with CIPN risk and severity (59). SCN10A has also been associated with chronic CIPN (58). Associations of mutations in voltage-gated potassium channels with CIPN have not been identified (78).

Genes implicated in inherited neuropathies have also been examined in relation to CIPN. SBF2, associated with Charcot-Marie-Tooth disease, was associated with CIPN in 213 African American patients treated with paclitaxel (61). However, another prospective study of 58 paclitaxel-treated patients found FZD3 was associated with CIPN, but not SBF2 (61). In a study of 219 breast cancer patients treated with taxanes, FGD4 was associated with CIPN (46). In a large prospective study of 855 patients of European Ancestry receiving paclitaxel, another polymorphism in FGD4 was associated with patient-reported sensory CIPN, which was replicated in two additional cohorts (62). In the replication cohorts a different polymorphism in FZD3 was also associated with sensory CIPN (62). ARHGEF10 was associated with CIPN in a prospective study of 269 cancer patients treated in Alliance N08C1 that analyzed blood samples for 49 Charcot-Marie-Tooth genes (63). These findings were confirmed in 138 patients treated with paclitaxel in Alliance N08CA (64).

There is a growing body of literature suggesting inflammation contributes to CIPN (73, 79, 80), and genetic alterations in inflammatory pathways have been studied in association with CIPN. In 3,431 breast cancer patients treated with paclitaxel a SNP in FCAMR that encodes the FC receptor, trended toward a significant association with CIPN (60). In 139 patients treated with bortezomib, variations in genes regulating immune function, CTLA4 and CTSS, were associated with time to onset of CIPN, with a similar trend for CTLA4 in a validation cohort (55). Bortezomib-neuropathy has also been associated with alterations in IL17RD, IL10RA, and genes in the NF-KB signaling pathway in 646 patients with multiple myeloma (52). Late-onset bortezomib-neuropathy was associated with polymorphisms in other genes in inflammatory pathways, MBL2 and PPARD in 186 myeloma patients (53). In a meta-analysis of pediatric patients treated with vincristine for ALL from two large clinical trials rs7963521, associated with coding of the protein chemerin implicated in inflammatory pathways (66), was associated with CIPN.

Polymorphisms in genes involved in chemotherapy metabolism may also increase CIPN sensitivity. In 107 children treated for ALL with vincristine CYP3A5 polymorphisms were associated with CIPN (45). SNPs in CYP2C8 and CYP3A4 were associated with ≥grade 2 CIPN in two studies of breast cancer patients treated with taxanes (46, 47). In 79 breast cancer patients treated with taxanes SNPs in NR1I3 and UGT2B7 involved in drug metabolism were also associated with CIPN (48). In 1,303 patients treated with taxanes, additional SNPs in genes involved in taxane metabolism including, CYP2C8^*4 and CYP1B1^*3, were associated with CIPN, as were, ABCB1 and SLCO1B1, involved in drug transport (42). In multiple myeloma patients treated with bortezomib, alterations in PSMB1, encoding drug binding proteins (55), and ABCC1 and ABCC6, encoding drug transport, were also associated with CIPN (52). However, in a separate study of 369 multiple myeloma patients ABCC1 polymorphisms were not associated with bortezomib-neuropathy, but were associated with vincristine-neuropathy, as was DPYD responsible for drug excertion (53). ABCC1 polymorphisms have also been associated with CIPN in pediatric ALL patients treated with vincristine (40, 50). Alterations in, ABCB1, encoding drug transport, has also been widely associated with CIPN in patients treated with vincristine and taxanes (44, 46, 49–51), and alterations in ABCC2 are associated with CIPN in children treated with vincristine (50). An alteration in ABCG2, involved in oxalate metabolism, was associated with oxaliplatin-induced neuropathy in 206 colon cancer patients (54).

Genes involved in nervous system development and function, and in cellular repair pathways, have also been associated with CIPN.

Alterations in genes encoding ephrin receptors (EPHA4, EPHA5, EPHA6, EPHA8), a family of tyrosine kinase receptors involved in neural development, are associated with CIPN in patients treated with taxanes (62, 69–71, 81). An SNP in TAC1, encoding neuronal signaling hormones, was associated with CIPN in colon cancer patients treated with oxaliplatin (65). A polymorphisms in COCH, encoding cochlin involved in vestibular function and hearing loss, was associated with CIPN in a study of children with ALL treated with vincristine (82). Alterations in SOX10, involved in neuronal development, and GPX7 were associated with CIPN in 107 survivors of gynecologic cancers treated with taxane or platinum (67). Polymorphisms in genes involved in nervous system function, NFATC1, NFATC4, and EDN1 were associated with CIPN in 646 myeloma patients treated with bortezomib (52), as were TCF4, DYNC1I1, and GJE1 in 139 myeloma patients treated with bortezmib (55). GNGT1 encodes a protein in photoreceptors and has been associated with taxane-CIPN (68, 83).

Alterations in genes associated with DNA repair are also associated with CIPN. SNPs in CCNH, encoding cyclin H, involved in cell cycle progression and DNA repair (84), were associated with CIPN in 206 colon cancer patients (54), and in 228 gastrointestinal cancer patients treated with oxaliplatin (72). In myeloma patients treated with bortezomib, ERCC4 and ERCC3 involved in DNA repair were associated with CIPN (53). In a study of 680 testicular cancer survivors treated with cisplatin, lower expression of DNA repair gene RPRD1B, was associated with an increased risk of CIPN, which was replicated in two independent datasets (85).

In 24 patients who received 12 weekly 3 or 1 h infusions of paclitaxel, longer duration of paclitaxel concentration >0.05 μM was associated with developing CIPN (86). In another prospective evaluation of 60 breast cancer patients receiving weekly paclitaxel infusions neither peak plasma concentration nor time above concentration of 0.05 μM were associated with CIPN, but were associated with increased toxicity-related treatment disruptions (87). An early study of lung cancer patients treated with paclitaxel also found no association between plasma concentration and neuromuscular or neurosensory outcomes (88).

An early study of pharmacokinetics in 54 children treated with vincristine did not find any association between vincristine clearance and neurotixicy (89). In a subsequent study assessing pharmacokinetics of vincristine in 74 pediatric patients, lower vincristine metabolite concentrations were associated with increased CIPN severity (45). Another study assessed vincristine pharmacokinetics in 35 patients receiving vincristine via push or 1 h infusions and found intercompartment clearance of vincristine was associated with an increased risk of CIPN, however other pharmacokinetic measures including maximum concentration were not associated with an increased CIPN risk (90).