Chemotherapy-Induced Peripheral Neuropathy: Epidemiology, Pathomechanisms and Treatment

The most recent estimation for all-cause cancer incidence is 18.1 million new cases per year [1]. With more effective targeted cancer treatments, long-term cancer survival is increasing in high-income countries [2], as evidenced by the 27% drop in the overall cancer death rate in the United States between 1991 and 2016 [3, 4]. However, chemotherapy-induced peripheral neuropathy (CIPN) is a common and challenging complication of several frequently administered antineoplastic agents [5]. The development of CIPN may result in prolonged infusion times, dose reduction or premature cessation of chemotherapy [6‐8], which may negatively impact both treatment efficacy and patient survival [9, 10]. A meta-analysis of randomised controlled trials and cohort studies showed that around half of all patients develop CIPN during treatment [10].

There is currently no gold standard for the assessment of CIPN, with a variety of clinical tools utilised in studies with heterogeneous primary outcome measures [11‐21]. Indeed, subclinical nerve damage and motor involvement are poorly defined when using current standardised clinical instruments [15]. Accurate comparisons of the prevalence, incidence, prevention and treatment of CIPN are therefore problematic (Table 1). Additionally, there are considerable disparities in patient- and clinician-reported neurotoxicity. For example, in the ICON7 trial, clinicians reported CIPN in 28% of patients, while 67% of patients reported ‘quite a bit’ or ‘very much’ tingling or numbness, with poor agreement between patients and clinicians (κ = 0.236, 95% confidence interval, 0.177–0.296, p < 0.001) [22].

Chemotherapeutic agents result in neurotoxicity through a variety of mechanisms, culminating in a predominantly symmetrical sensory or sensorimotor, length-dependent neuropathy and autonomic dysfunction [23‐28]. Neuropathic syndromes specific to chemotherapeutic agents can be observed, each with their own presentation and natural history [29‐33] (Table 2). CIPN can develop, or continue to worsen, several months after treatment has stopped, in a phenomenon termed “coasting”. The prevalence of CIPN one month after finishing chemotherapy approaches 68%, and persists in approximately one third of patients beyond 6 months [10]. Risk factors for CIPN include the agent used, cumulative dose, number of cycles, treatment duration, combination therapies, genetic predisposition, age, existing nerve damage, severity of acute symptoms and chronic alcohol consumption amongst others [15]. The ageing population and more efficacious chemotherapeutic regimens will continue to increase cancer cure rates and long-term cancer survival [34], together with CIPN [35]. It is therefore imperative to develop effective strategies for the early identification with prevention and more efficacious management of CIPN.

Electronic database searches were undertaken in EMBASE, PubMed, OVID and Cochrane CENTRAL to identify included articles. The reference lists of relevant articles were searched, and in addition, studies were identified by authors with expertise in CIPN. Studies published from initial curation of the electronic database to March 2021 were identified, and those felt not relevant by authors were excluded with the guidance of the senior author (U.A.). This article is based on previously conducted studies and does not contain any new studies with human participants or animals performed by any of the authors.

Balancing the risk of different manifestations of chemotoxicity and the potential benefit of reduced disease burden/remission is a demanding aspect of oncological practice. For instance, there is an increased risk of CIPN with oxaliplatin compared to cisplatin, but the risk of thromboembolism is greater for cisplatin than for oxaliplatin [36], with a small survival benefit with oxaliplatin compared to cisplatin. Accordingly, there are many options to try to limit CIPN by reducing the number of doses and cumulative toxicity, especially in older individuals or those more likely to have pre-existing neuropathy such as diabetes [37]. Patients are less likely to continue chemotherapy if they develop serious complications. In a population of older patients in a non-curative setting, lower doses of oxaliplatin and capecitabine were better tolerated, resulting in patients receiving a greater number of cycles and a small survival benefit [37]. Al-Batran et al. [38] reported that the rate of CIPN in patients treated with epirubicin, cisplatin and fluoropyrimidine was half that observed in patients treated with oxaliplatin, docetaxel and fluoropyrimidine. Similarly, Cunningham et al. [36] reported that the rate of thromboembolism doubled in patients treated with epirubicin, cisplatin and fluoropyrimidine [38]. Ultimately, there are trade-offs when treating patients with cancer, especially those with limited therapeutic options, worse prognosis or pre-existing conditions that may predispose them to chemotherapy-related complications.

Platinum-based chemotherapeutics (oxaliplatin, cisplatin and carboplatin) are used in the treatment of solid tumours of the gut, bladder, testes, ovary, uterus, lung, head and neck [39, 40]. Platinum chemotherapeutic agents have the highest prevalence rates of CIPN, affecting ~ 70% of patients, often complicated by coasting [29, 41]. The main anatomical structure injured by platinum agents is the dorsal root ganglion, and manifests as a sensory neuropathy with prominent pain accompanied by cold-induced allodynia and muscle cramps due to peripheral nerve hyperexcitability or neuromyotonia. Acute oxaliplatin-induced peripheral neuropathy (OIPN) can result in prolonged infusion times (~ 22%), dose reduction (15–43%) and treatment cessation (6–21.4%) [42‐46]. A systematic analysis of studies including 6211 participants undergoing oxaliplatin treatment found acute OIPN with an incidence of 4–98% [42]. The wide range of incidence may be attributed to heterogeneous dosing regimens, drug combinations, dosing intervals and screening instruments used to identify acute OIPN [42]. A longitudinal study following 346 participants undergoing FOLFOX chemotherapy demonstrated a 3-day peak in acute OIPN, with sensory symptoms including cold-induced hypersensitivity (71%), sensitivity to swallowing cold food and drink (71%), throat discomfort (63%) and muscle cramps (42%) [25]. Symptoms often persist between treatments and increase in severity with subsequent doses [25, 26]. The initial severity of acute OIPN also predicts progression to chronic sensory OIPN [25, 47], which can be identified in 84% of patients after 25 months, with long-term impact on functionality and quality of life.

The dorsal root ganglion (DRG) is particularly susceptible to chemotherapeutic agents, as it lies outside the central nervous system and is not protected by the blood–brain barrier [48]. In an animal model of CIPN, the accumulation of oxaliplatin in DRG neurons was associated with intracellular overexpression of Octn1/2 and Mate1 transporters [49]. Oxaliplatin also interferes with DNA cross-linking, resulting in direct neurotoxicity [50] and early p38 and ERK1/2 activation, reduced mitochondrial respiration, increased oxidative stress and dose-dependent apoptosis of DRG neurons [51]. Cell culture studies have shown greater neuronal cell body atrophy and apoptosis when exposed to oxaliplatin compared to both paclitaxel and controls, promoting a sensory neuronopathy (neuronal cell body) as opposed to an axonopathy that is phenotypic of other chemotherapeutic agents. OIPN also correlates with mitochondrial morphological artefacts, decreased adenosine triphosphate generation and depressed respiration rates in mitochondrial complexes I and II [52‐54] 55. Indeed, platinum agents and their metabolites form adducts with mitochondrial DNA (mtDNA), disrupting replication and transcription, with a reduction in neuronal cell body mitochondrial populations [56]. Oxidative stress leads to oxidation of intracellular moieties of neurons, diminishing neuronal energy status and increasing apoptosis [55, 57‐61]. Reduction of oxidative stress with phenyl N-tert-butylnitrone has been shown to decrease oxaliplatin-induced mechanical hyperalgesia and cold allodynia [62, 63]. Oxaliplatin also interacts with voltage-gated potassium channels (VGKC) expressed on peripheral motor neurons and is implicated in the acute phase of OIPN in which patients exhibit nerve hyperexcitability, prolonged depolarisation, increased neurotransmission and muscle contraction similar to that seen in neuromyotonia [44]. Notably, the excitability of Aδ and C-type fibres of the maxillary branch of the trigeminal nerve are controlled by VGKCs. Further, VGKC isotype 4.3 channels had slower deactivation after administration of oxaliplatin, and this may underlie cold-induced orofacial allodynia [64]. Intramuscular injections at the base of the tail of mice with oxaliplatin were shown to cause acute transient dose-dependent changes in excitability of both motor and sensory axons and evoked ectopic activity in these fibres [65]. Moreover, mathematical modelling indicates that oxaliplatin causes slow inactivation of voltage-gated sodium (NaV) channels and reduces the resting membrane potential of nerve fibres through the reduction of fast potassium conductance in the acute phase of OIPN [65]. Indeed, in preclinical studies, NaV channel-blocking drugs such as topiramate have recently been shown to have a neuroprotective effect in the prevention of both the acute and chronic phase of OIPN, with no interactions with the antineoplastic activity of oxaliplatin [66].

Cold hyperalgesia is a major feature of OIPN and is thought to be driven by TRPA1 and p38 MAPK activation in DRG neurons and increased activity of NaV channel isoforms NaV1.6 and NaV1.9 in nociceptive subpopulations of peripheral and DRG neurons. Further, there is a potential role played by transient receptor potential melastatin 8 (TRPM8) in acute cold-induced allodynia [67]. Altered expression of pain receptor-associated TREK-1, TRAAK, TRPA1 NaV channel isoforms and hyperpolarisation-activated cyclic nucleotide–gated (HCN1) channels in sensory neurons contribute to the prominent neuropathic pain associated with this condition [68]. Oxalate chelates Ca²⁺ ions, contributing to neuronal excitability and increasing spontaneous pain signalling [69]. There is also increased expression of pro-inflammatory cytokines including tumour necrosis factor-α (TNF-α) and interleukin (IL)-1β and decreased expression of the neuroprotective cytokines IL-10 and IL-4 [70‐72] through the activation of astrocytes by platinum-based chemotherapy agents. A summary figure of these processes is shown in Fig. 1.

The taxanes (paclitaxel, docetaxel and cabazitaxel) are currently first-line treatments for breast, ovarian, lung, bladder, prostate and other solid tumour cancers [34, 73, 74]. Taxane-induced peripheral neuropathy (TIPN) is the most common non-haematological adverse event of treatment, which may result in dose reduction and cessation of treatment, impacting patient survival [75]. TIPN primarily causes an acute, length-dependent distal sensory neuropathy, accompanied by neuropathic pain, which may progress proximally in more severe cases. Aβ fibres and to a lesser extent Aδ and C-fibres are affected in a glove-and-stocking distribution [31, 76, 77]. Patients report tingling, numbness, paraesthesia, neuropathic pain, cold-induced dysaesthesia and muscle cramps [26, 78], which typically worsen with treatment and gradually improve with cessation [26], although 31–44% of patients treated with docetaxel or paclitaxel report symptoms after up to 6 years of follow-up [79‐81]. TIPN incidence in non-small cell lung cancer (NSCLC) in phase III trials occurred in 13–62% of patients [82]. Severe TIPN (FACT-Lung grade ≥ 3) occurred in 21–40% of patients, with worse outcomes after receiving paclitaxel as opposed to docetaxel-based chemotherapy regimens [83]. Thus, docetaxel is generally considered to be less neurotoxic than paclitaxel [84].

Studies have identified an increase in the incidence of abnormal axonal mitochondria in C-fibres when compared to controls after ~ 1 month of paclitaxel treatment [85]. Paclitaxel interacts with the mitochondrial permeability transition pore, leading to mitochondrial dysfunction, decreased mitochondrial respiration and disruption of neuronal ATP generation [86, 87], with disruption of the axonal microtubule network [88]. Taxane treatment of rat DRG neuronal stem cells increased ROS production and oxidative stress, simultaneously decreasing mitochondrial metabolic activity, membrane potential and antioxidant bioavailability [62, 89]. Similarly, taxane treatment of rat and human DRG neurons lowered the resting and threshold membrane potential and increased the frequency of ectopic spontaneous activity [90]. In experimental models, paclitaxel increased the expression of voltage-gated calcium channels (Ca_v) 3.2 and calcium current amplitude and decreased the excitability threshold of dorsal root sensory neurons, which when inhibited decreased mechanical hypersensitivity [87, 91, 92]. Further, toll-like receptor (TLR) 4 is also upregulated, resulting in increased intracellular calcium mediated by the co-located protein Ca_v3.2. Moreover, paclitaxel increases the expression of Nav 1.7 channels in a dose-dependent manner in human DRG neurons in culture, leading to increased ectopic spontaneous activity [92, 93]. Notably, paclitaxel can bind and activate TLR4 on macrophages, engaging signalling pathways that lead to increased gene expression and release of nuclear factor kappa B (NF-kB), initiating inflammatory and cytokine cascades [94]. TLR4, MyD88 and ERK1/2 expression is increased in IB4⁻ and CGRP⁺ DRG neurons [94‐96]. Inflammatory mediators IL-6, IL-8, IL-10, monocyte chemoattractant protein-1 (MCP-1) and activated Langerhans cells are upregulated, where they propagate the further release of pro-inflammatory cytokines [97‐99]. Furthermore, there is increased expression of stress and inflammatory markers in Schwann cells and lumbar DRG neurons [100, 101]. Activation and migration of CD68⁺ macrophages, CD8⁺ T cells and CD11b⁺ leucocytes towards the DRG and peripheral nerves has been identified [99‐101]. Thus, sensitisation of C-fibres, net energy loss, neuroinflammation and hyperexcitability contribute to paclitaxel-induced peripheral neuropathy. The hypotheses of the pathomechanism of TIPN is summarised in Fig. 2.

Vinca alkaloids are natural (vincristine and vinblastine) and semi-synthetic (vindesine and vinorelbine) chemotherapeutics derived from the periwinkle plant and are used either alone or in combination therapy to treat haematological malignancies, testicular cancer, myeloma, sarcoma, non-small cell lung cancer and tumours of the kidney, liver, lung, brain and breast [102]. Vincristine is arguably the most neurotoxic vinca alkaloid, with a majority of patients developing vincristine-induced neuropathy (VIPN) [10, 103], the severity of which is dose-dependent [104]105. Genetic polymorphisms in genes associated with Charcot-Marie-Tooth (CMT) disease appear to increase the risk of VIPN [106]. The incidence of VIPN or vinorelbine-induced neuropathy leading to sensory neuropathy is ~ 20%, with motor impairment in 17.5% of adult patients [107, 108]. The most common presentation of VIPN is a length-dependent sensory neuropathy, with significant motor impairment and occasional cranial nerve involvement [107]. Surprisingly, 91% of patients reported continuing symptoms 12 months after cessation of treatment [109], and there is evidence for long-term distal sensory [107, 110] and motor deficits in vincristine-treated cancer survivors [30].

Anterograde transport of organelles and membrane proteins and retrograde transport of signalling molecules depends on microtubule-based transport [88]. Vinca alkaloids interfere with and disrupt microtubule assembly and mitotic spindle formation [111, 112]. They also increase the stability of microtubules, which impacts negatively on the ability of the cell to dynamically alter the structure of the cytoskeleton affecting axonal transport [88] 113. Additionally, vincristine is mitotoxic and can impair the mitochondrial electron transport chain, resulting in defective energy production [114]. Axonal degeneration requires both sterile alpha and TIR motif-containing proteins SARM1 and MAPK, and the deletion of SARM1 protects mice from developing VIPN [115]. Other intracellular targets include the NF-E2-related factor and haeme oxygenase 1/carbon monoxide system (Nrf2/HO-1/CO) which modulates the expression of connexin 43 (Cx43), protecting against nerve damage and reducing vincristine-induced neuroinflammation [116]. Increased expression of inflammatory markers TNF-α and IL-1β and increased expression of TRPA1 were recently identified in models of VIPN [117]. Moreover, mRNA gene ontology has identified the inflammatory role of vincristine on microglia and upregulation of pro-inflammatory genes including frizzled-related protein 2 (SFRP2) and C-X-C motif chemokines (CXCL) 10 and 9 [118]. The current available data on the hypothesised mechanism of VIPN is shown in Fig. 3A.

Thalidomide is a US Food and Drug Administration (FDA)-approved treatment for multiple myeloma (MM) [119]. Patients treated with thalidomide for MM, glioblastoma, renal cell carcinoma, colorectal and lung, melanoma, and breast and prostate cancer can develop thalidomide-induced peripheral neuropathy (ThiPN) [32, 102, 120‐122]. Symptoms include symmetrical numbness, tingling, burning pain and sensitivity to touch and heat, with hyperaesthesia, hypoaesthesia and paraesthesia in a glove-and-stocking distribution [32] with tremor, muscle cramps, distal muscle weakness, areflexia, loss of proprioception, gait ataxia and/or a lack of coordination [32, 123‐128]. The incidence of ThiPN ranges from 11 to 75% and is dependent on dose [120, 122, 129‐134] and duration of exposure [125]. As such, the results of phase I studies giving thalidomide to the maximum tolerable dose are not representative of patients who are receiving this medication over a longer duration. Peripheral neuropathy induced by the thalidomide analogues lenalidomide and pomalidomide are less severe and occur at a lower incidence [135‐137], making them the agents of choice in those with pre-existing neuropathy. MM is currently incurable and requires long durations of exposure to thalidomide and its analogues, which results in accumulative chemotoxicity [138]. This is especially relevant as the 5-year relative survival rate of M has increased in recent years [139]. Barlogie et al. [140] reported that 90% of participants with a National Cancer Institute Common Terminology Criteria for Adverse Events (NCI-CTCAE) ≥ 2 grade ThiPN improved to a grade ≤ 2 within 3–4 months of thalidomide dose attenuation. However, complete clinical recovery is limited to approximately one quarter of patients [124, 128, 141‐143].

The exact pathomechanism of thalidomide is yet to be fully elucidated, but antiangiogenic properties [144] may lead to hypoxia of small nerve fibres [145]. Additionally, the immunomodulatory action of thalidomide inhibits basic fibroblast growth factor (b-FGF), vascular endothelial growth factor (VEGF), TNF-α and NF-kB and dysregulates neurotrophins; the signalling molecules responsible for the proliferation, survival and function of neurons are shown in Fig. 3B [146]. Further, preclinical ThiPN animal models demonstrate improvement of NCS following the injection of VEGF [147].

The proteasome inhibitors bortezomib, ixazomib and carfilzomib are FDA-approved treatments for MM [148, 149] and are used in the treatment of progressive, relapsed or refractory MM and mantle cell lymphoma [150, 151]. Bortezomib-induced peripheral neuropathy (BIPN) is a distal, symmetrical, length-dependent axonal sensorimotor neuropathy characterised by mild to moderate sensory loss, mild to severe neuropathic pain and mild motor weakness of the distal lower extremities [[33, 152]. Phase II trials have identified a BIPN incidence of 31–37%, with grade ≥ 2 neuropathy present in 28% of participants [153‐155]. Although ixazomib and carfilzomib have a lower incidence of CIPN [156‐158], long-term treatment [159] with the addition of other chemotherapeutic agents [160] is required to maintain remission.

Bortezomib initiates apoptosis through the release of intracellular Ca²⁺ in the endoplasmic reticulum, leading to activation of caspase, a protease enzyme essential for programmed cell death [161]. A study showed vacuolation of DRG-associated mitochondria [162], although these findings could not be replicated [163]. Bortezomib treatment increased the number of swollen and vacuolated mitochondria in A-fibres and C-fibres compared to controls, and mitochondrial respiration and adenosine triphosphate production were reduced, indicating cumulative energy failure as a pathogenic mechanism of BIPN [164]. In a recent study, bortezomib exhibited neurotoxicity in PC12 neuroblastoma cells through the induction of apoptosis which was ameliorated with antioxidants, implicating oxidative stress in the pathogenesis of BIPN [165]. Ultrastructural features of myelin sheath degeneration of large nerve fibres and axonal degeneration of C- fibres have been identified [162, 163]. Inhibition of NF-kB and TNF-α attenuates the severity of BIPN in preclinical models [166, 167]. Indeed, bortezomib treatment increases the expression of GATA-binding protein (GATA3), a transcription factor implicated in the regulation of inflammatory signalling cascades and upregulation of the T-cell chemoattractant chemokine C–C motif ligand 21 (CCL21) in dorsal horn neurons, which when silenced attenuates mechanical allodynia in Sprague Dawley rats [168]. The current hypothesis for the pathomechanism of BIPN is summarised in Fig. 3C.

Electrodiagnostic methods are considered the reference standard for the functional assessment of large sensory and motor fibres which drive paraesthesia, numbness and weakness seen in people with CIPN. Although sensory testing used in composite scoring systems is often deployed in the clinical setting, a rigorous, lengthy battery of standardised sensory tests is required to reliably identify a patient’s sensory phenotype. Further, these tests are subjective and cannot discriminate between a central or peripheral disease process of the somatosensory nervous system, and benefit from the addition of a structural measure of peripheral nerve fibres. In light of this, we include an overview of self-reported outcome measures, composite scoring systems, functional tests of large fibres, structural measure of small fibres such as skin biopsy, and highlight the novel, reiterative method of corneal confocal microscopy. This method is of particular interest, as the early detection of CIPN may enable health care professionals to determine subclinical nerve damage and assist in changes to dosing strategies before the neuropathy becomes irreversible. In this section we highlight the methods used to quantify CIPN in both clinical and research settings.

The methods used in both clinical trials and medical practice to identify and grade the severity of CIPN can be broadly separated into instruments which utilise patient-reported outcomes, composite scoring systems with a functional assessment component, and quality-of-life tools [169]. Most commonly used is the clinician-led patient-reported tool, National Cancer Institute-Common Terminology Criteria for Adverse Events (NCI-CTCAE), although other instruments such as the Eastern Cooperative Oncology Group (ECOG) criteria and the World Health Organization (WHO) neurotoxicity scale [170] are also used. The latest version of the NCI-CTCAE (version 5.0) (Table 3) grades both motor and sensory neuropathy according to asymptomatic (grade 1), moderate (grade 2), severe (grade 3) or life-threatening (grade 4) neurotoxicity. Composite scoring systems such as the Total Neuropathy Score (TNS) use patient-reported symptoms, physical examination, vibration perception threshold and nerve conduction studies to grade CIPN, although there are versions which omit vibration perception threshold (TNSr) and nerve conduction studies (TNSc) [169]. Further, the TNS clinical (TNSc) and nurse-administered TNS (TNSn) have been shown to correlate well with the emergence of sensory and motor symptoms after the completion of chemotherapy, identifying 88% of participants who developed CIPN [171].

Functional assessments are self-reported questionnaires measuring both the quality of life and symptoms specific to how neurotoxicity impairs activity. These measures are often tailored to the primary cancer such as the Functional Assessment of Cancer Therapy/Gynecologic Oncology Group-neurotoxicity (FACT/COG-Ntx) tool, which has been shown to correlate well with the TNSc and TNSn [171]. Other examples of functional assessments include the European Organisation for Research and Treatment of Cancer (EORTC) and the chemotherapy-induced peripheral neuropathy questionnaire (CIPN20). These instruments have been reviewed extensively by Cavaletti et al. [170] and Park et al. [169].

NCS provide an objective measure of large fibre function and are considered the reference standard for the diagnosis of large fibre involvement in CIPN [172]. Peripheral nerve demyelination is accompanied by conduction slowing and latency prolongation, and axonal loss is accompanied by a reduction in amplitude.

In patients treated with paclitaxel and oxaliplatin, NCS can be used to confirm a symmetric, length-dependent, predominantly sensory distal neuropathy [173‐177]. However, in acute OIPN there are rarely significant changes in NCS, although motor axons can develop increased refractoriness resulting in repetitive motor discharges [173, 178, 179]. Further, a change in sensory excitability in acute OIPN predicts the development of chronic OIPN, a purely sensory neuropathy with a reduction in sensory sural nerve action potential (SNAP) and nerve conduction velocity (SNCV) without motor NCS involvement [180]. In a longitudinal study of ten participants, the phenomenon of coasting was evidenced by worsening median and sural sensory amplitudes at least 3 months after completing oxaliplatin-based chemotherapy [181]. The typical presentation of TIPN is that of a predominantly distal sensory axonal neuropathy with some motor involvement [172, 182]. A recent longitudinal study identified significantly reduced SNAP amplitudes predominantly in the upper limbs, but to a lesser extent in the lower limbs, 12 months after completion of taxane chemotherapy, arguing for a non-length-dependent effect [15]. Both acute and chronic thalidomide neurotoxicity are characterised by attenuation of median, radial and sural SNAPs and compound muscle action potentials (cMAPs) of the peroneal and tibial nerves [126]. NCS in patients with BIPN largely indicate a predominantly small fibre sensory axonal neuropathy, with less frequent motor neuropathy [129]. Bortezomib and thalidomide combination therapy is associated with a significant reduction in sural SNAP, peroneal motor nerve action potential (PMNAP) and peroneal motor nerve conduction velocity (PMNCV) [33, 127]. NCS in people with VIPN is characterised by a distal sensorimotor axonal neuropathy and motor involvement [172], with prolongation of distal latencies but preserved conduction velocities [183]. Furthermore, NCS parameters may deteriorate before or after the development of neuropathic symptoms [47, 127, 129, 172, 184‐187].

QST provides an extensively validated mechanism-based and symptom-orientated approach to neuropathic pain. The loss of nerve fibre sensitivity or deafferentation can be detected using quantitative sensory testing for different nerve fibre populations. The loss of Aβ-fibre sensitivity is indicated by impaired vibration perception, light touch or mechanical detection thresholds. C-fibre dysfunction is reflected by abnormal heat detection and heat pain thresholds, whilst Aδ-fibre dysfunction is indicated by abnormal thresholds to pinprick stimuli, mechanical pain and cold detection [188‐190]. The majority of patients with CIPN from a range of drugs exhibit reduced or absent pinprick and vibration perception thresholds and impaired proprioception [191]. Early impairment of vibration detection and cold detection thresholds have been identified from week 12 of treatment with oxaliplatin, with an increase in mechanical detection thresholds 6 months after finishing treatment [180, 192]. Cold pain threshold can be used to dichotomise participants with acute OIPN and change over time [193]. People with TIPN exhibit diminished tactile perception in the upper and lower extremities, with worsening VPT in the lower limbs [194]. Participants with VIPN and BIPN exhibit widespread abnormalities in touch detection, pinprick detection and heat detection thresholds both within and outside self-reported areas of involvement [195, 196].

The accepted gold standard for diagnosing small fibre pathology is skin biopsy [197, 198]. Normative age- and sex-related values for intraepidermal nerve fibre density have been published for clinical use [199]. In preclinical models of paclitaxel- and vincristine-induced peripheral neuropathy, there is a significant reduction in intraepidermal nerve fibres [91, 98]. Indeed, a significant decrease in intraepidermal nerve fibre density (IENFD) at the distal leg was identified in eight patients 6 months after oxaliplatin treatment had been stopped [200]. Notably, a recent study found a significant time-dependent decrease in IENFD 6 months after treatment had been stopped [180]. In patients with BIPN, whilst epidermal nerve density did not differ, there was a reduction in subepidermal nerve fibre density [201]. Further work is needed to characterise the differential effect of different chemotherapy drugs on small nerve fibres in the skin.

Corneal confocal microscopy (CCM) is a non-invasive, reiterative ophthalmic imaging technique that detects small nerve fibre abnormalities in the subbasal nerve plexus in a range of peripheral neuropathies [202‐210]. A large body of published data shows that CCM has good diagnostic [211] and prognostic [212] utility in diabetic neuropathy. Recently, CCM has been proposed to have utility in the diagnosis and follow-up of patients with CIPN [213].

In an early study of 15 patients with colorectal cancer treated with oxaliplatin, 10 patients developed a significant worsening of TNSc and 8 patients developed NCV evidence of a sensory axonal neuropathy [214]. CCM demonstrated a significant abnormality in 10/15 patients characterised by a reduction in corneal nerve fibre density (40%) and length (37%). Interestingly, after receiving the final cycle of chemotherapy, two patients with normal clinical and neurophysiological findings had evidence of severe corneal nerve loss, and 3 weeks later they developed neuropathic symptoms, indicative of coasting [214]. In 21 patients with gastro-oesophageal cancer without neuropathic symptoms there was evidence of corneal nerve loss which correlated with the stage of cancer. After three cycles of platinum-based chemotherapy, 61.5% of patients developed grade 1 symptomatic paraesthesia on CTCAE criteria; however, all patients except those with metastatic liver disease showed an increase in corneal nerve fibre length [205]. CCM has also shown a significant reduction in corneal nerve fibre density, length and beading in patients with MM undergoing treatment with bortezomib, despite clinically evident neuropathy being present in only 38.5% of patients [215]. More recently, of 63 patients who had received docetaxel for breast cancer (n = 28) or oxaliplatin for colorectal cancer (n = 35) 5 years prior to detailed neuropathy phenotyping, 41.3% still had evidence of CIPN, of whom 58% had pure large fibre neuropathy based on NCS [216]. Detailed QST revealed increased cold, warm, mechanical and vibration detection thresholds with no evidence of pinprick hyperalgesia or dynamic mechanical, cold or warm allodynia. CCM demonstrated no significant difference in corneal nerve fiber length, density or branch density between controls and patients with CIPN with and without small fibre neuropathy [216]. In a study comparing CCM in different peripheral neuropathies, patients with CIPN had evidence of corneal nerve fibre loss in a distinct pattern based on the corneal nerve fractal dimension, which differed from patients with diabetic neuropathy or chronic inflammatory demyelinating neuropathy [217]. A study of 70 patients with breast, colorectal, upper gastrointestinal and gynaecological cancer having received either paclitaxel (n = 40) or oxaliplatin (n = 30) within the past 3 to 24 months showed evidence of a significant reduction in the corneal nerve fibre and inferior whorl lengths [218]. Furthermore, corneal nerve fiber length, inferior whorl length, average nerve fiber length and corneal nerve fiber density were significantly lower in patients with neuropathy compared to those without neuropathy based on the correlation of TNSr and inferior whorl length with hand dexterity [218]. These data suggest that CCM may have diagnostic and prognostic value in CIPN.

A large meta-analysis of 13,683 people with CIPN estimated the prevalence of neuropathic pain to be as high as 40% [219]. A recent international study of 2003 patients with CIPN has found a similar prevalence of neuropathic pain, which significantly impacted upon quality of life and daily functioning [220]. CIPN is predominantly a sensory neuropathy, as summarised in Table 2, with pain being the most bothersome symptom [221]. Indeed, the symptom burden of CIPN including sensory disturbances and neuropathic pain profoundly impacts on the quality of life of survivors of cancer [84, 191, 222‐227]. CIPN also affects functionality and the capacity to work both during and after treatment, fuelling unemployment and loss of working time [228]. Moreover, a recent US administrative claims analysis by Song et al. [229] found that individuals with painful CIPN incur a significant economic burden driven by costs of analgesic drug prescriptions, increased rates of hospitalisation, emergency department visits and outpatient hospital visits compared to participants treated for cancer who did not develop CIPN. Pike et al. [230] showed that painful CIPN was associated with higher average costs of $17,344 compared to patients without CIPN. Notably, oxaliplatin- or paclitaxel-based chemotherapy regimens are more likely to result in neuropathic pain, and the pain associated with OIPN and TIPN is more severe and protracted [15]. The treatment of chronic neuropathic pain is often inadequate and may be poorly tolerated [231].

A Cochrane systematic review of interventions and an expert group systematic review by the American Society of Clinical Oncology (ASCO) recommended against the use of a range of interventions (acupuncture, cryotherapy, exercise therapy or ganglioside-monosialic acid (GM-1), retinoic acid, amifostine, amitriptyline, calcium magnesium infusion (Ca/Mg), calmangafodipir, cannabinoids, carbamazepine, l-carnosine, diethyldithiocarbamate (DDTC), gabapentin, pregabalin, glutamate, glutathione, goshajinkigan (GJG), metformin minocycline, N-acetylcysteine, nimodipine, omega-3 fatty acids, ORG 2766, oxcarbazepine, recombinant human leukemia inhibitory factor, venlafaxine, vitamin B or vitamin E) in CIPN [232, 233]. Moreover, acetyl-l-carnitine is strongly advised against due to high-quality evidence indicating worsening neuropathy [232, 234].

The ACTTION [Analgesic, Anesthetic, and Addiction Clinical Trial Translations, Innovations, Opportunities, and Networks]/CONCEPPT [Clinical Endpoints and Procedures for Peripheral Neuropathy Trials] consortia developed recommendations for CIPN prevention studies [235]. These included the selection of outcome measures and endpoints, eligibility criteria, potential effects of the investigational therapy on the efficacy of chemotherapy and accurate sample size estimation [235]. Summaries of studies evaluating putative preventative therapies are detailed in Table 4.

Nutraceuticals as neuroprotective agents have not yielded strong evidence for the prevention of neurotoxicity. For instance, α-lipoic acid [236], OPERA [237], curcumin [238] and Neuronorm [239] have shown no benefit in randomised controlled trials despite positive findings in preclinical studies. Similarly, despite positive pilot studies, a large phase III trial did not demonstrate a significant neuroprotective effect of vitamin E and glutathione supplementation [240‐243, 244, 245] [246]. However, in two randomised, standard-of-care-controlled trials and a smaller non-randomised standard-of-care-controlled trial, glutamine was associated with reduced incidence and severity of dysaesthesias, nerve conduction impairment and interference with daily functioning [247‐249]. Amifostine demonstrated a clinically meaningful benefit for the prevention of sensory and auditory CIPN but was associated with worsening nausea and vomiting [250‐252]. Patients administered diethyldithiocarbamate (DDTC), with lower cumulative doses of cisplatin, were more likely to withdraw from treatment due to CisPN-related adverse events [253]. Similarly, the hexapeptide analogue of ACTH, ORG 2766, increased the incidence of CIPN in a smaller cohort study [252]. Caution is advised with nutraceuticals and supplements with unproven efficacy.

Retrospective studies of patients with advanced colorectal cancer treated with oxaliplatin found that calcium and magnesium infusion (Ca/Mg) significantly reduced the incidence of all-grade OIPN compared to 5-fluorouracil and leucovorin [254, 255]. Notably, a meta-analysis found that Ca/Mg treatment reduced the incidence of severe chronic OIPN (grade ≥ 2) (0.44 (95% CI 0.23–0.85; p = 0.01)) but does not reduce the incidence of acute OIPN (0.41 (95% CI 0.11–1.49; p = 0.18)) [256]. The reduction in acute OIPN incidence with Ca/Mg infusions has not been replicated in a phase I RCT and a large phase III RCT (Table 4) [257‐259].

Recently published ASCO guidelines indicate that duloxetine is the only currently recommended treatment; however, due to a lack of definitive efficacy, no recommendations can be made for exercise therapy, acupuncture, scrambler therapy, gabapentin, pregabalin, topical gel treatment (containing baclofen/amitriptyline plus/minus ketamine), tricyclic antidepressants or oral cannabinoids in the treatment of symptomatic CIPN [232]. Based on current clinical trial data (Table 5), larger, high-quality studies are needed to confirm efficacy and identify risks of treatment [9, 232, 235, 260, 261].

In a systematic review of acupuncture for the treatment of CIPN, two out of three trials found acupuncture to be effective in improving self-reported CIPN measures [262‐264], but one trial found no benefit [265]. A recent systematic review identified 19 RCTs with 1174 patients and showed that acupuncture significantly improved not only pain but also, surprisingly, nerve conduction velocity [266]. Pilot studies of electro-acupuncture and laser acupuncture have shown improvements in self-reported measures and sensory testing in patients with chronic CIPN [267, 268].

A secondary analysis of a large phase III randomised controlled trial of non-pharmaceutical interventions in cancer patients found that exercise reduced sensory symptoms in participants with OIPN, TIPN or VIPN, especially in participants who were older, male or had breast cancer [269]. A recent systematic review and meta-analysis indicated that exercise interventions significantly improve CIPN symptoms, and a sensorimotor-based exercise intervention reduced CIPN-induced loss of postural stability [270].

A large phase III randomised, placebo-controlled trial of participants with CIPN treated with topical 2% ketamine plus 4% amitriptyline showed no benefit on mean pain, numbness or tingling scores when compared to placebo (p = 0.363) [271].However, a pilot study of 44 participants with CIPN treated with topical 10% amitriptyline showed a five-point reduction in mean pain scores after 4 weeks (p < 0.0001) [272].

An in vitro study showed that oxaliplatin modulates the sensitivity of the capsaicin receptor (TRPV1) response through a secondary intracellular messenger [273]. In a single-centre study, the high-dose capsaicin 8% patch reduced pain by 84% in 18 participants with OIPN, 12 weeks after the patch was applied [274]. Similarly, a single-centre, open-label, longitudinal study showed that the capsaicin 8% patch ameliorated neuropathic pain in 16 participants with chronic CIPN, with evidence of regeneration of intraepidermal nerve fibres, suggestive of initial degeneration due to capsaicin [275]. Indeed, the latest ASCO guidelines indicate that the efficacy of the high-dose 8% capsaicin patch should be further explored [232].

In a multicentre, phase IV study, oxycodone and naloxone taken together with gabapentin (≥ 900 mg/day) was found to decrease mean numeric rating scale pain scores from 6.0 ± 1.3 to 4.7 ± 2.1, after 4 weeks (p =  < 0.0001) [276]. Similarly, treatment with controlled-release oxycodone reduced mean pain intensity from 7.6 to 1.3 at day 14 (p < 0.002) [277]. However, close monitoring of long-term opioid therapy, particularly in combination with gabapentinoids, is advised [278].

A double-blind, randomised, placebo-controlled trial found pregabalin to be more effective than both gabapentin and amitriptyline in decreasing pain scores, with a morphine-sparing effect associated with pregabalin monotherapy [279]. A pilot study and a cohort study identified gabapentin as a potential treatment with improved self-reported measures of CIPN [280, 281]. Nevertheless, a randomised, double-blind, placebo-controlled, crossover, phase III trial (n = 115) failed to show a significant change in the pain score with gabapentin in patients with CIPN [282]. Further, the pre-emptive administration of pregabalin did not decrease the risk of painful OIPN [283].

A phase III randomised, double-blind, placebo-controlled, crossover trial of nortriptyline in participants with CisPN showed no benefit on paraesthesia or neuropathic symptoms, although there was an improvement in sleep (p < 0.02) [284]. Amitriptyline has shown no efficacy for the improvement or prevention of CIPN symptoms in two double-blind, randomised, placebo-controlled trials [285, 286].

There is limited evidence for the use of SSRIs in painful CIPN [232].

Studies in experimental models of painful neuropathy have demonstrated a superior antinociceptive effect of norepinephrine compared to serotonin [287], and combined increases in both serotonin and norepinephrine result in a better analgesic effect than an increase in either one alone [288]. A multicentre randomised, double-blind, placebo-controlled crossover trial demonstrated the efficacy of duloxetine in participants undergoing platinum or taxane chemotherapy regimens [289]. At 5 weeks, participants receiving duloxetine reported a greater mean decrease in pain score compared to placebo (1.06 [95% CI 0.72–1.40] vs 0.34 [95% CI 0.01–0.66] p = 0.00)3 [290]. Similar results have been reported in other smaller studies [291, 292]. Further, participants with painful OIPN are more likely to respond to duloxetine that those with TIPN [293]. Duloxetine has a greater effect than venlafaxine on pain scores [294]. Recent ASCO guidelines advise a moderate recommendation for the use of duloxetine in CIPN [232]. A randomised, double-blind, placebo-controlled phase III trial found greater improvements in pain relief by ≥ 50% in participants receiving venlafaxine compared to placebo (p = 0.02) [295]. A retrospective cohort study of participants with painful TIPN or OIPN found that venlafaxine achieved relief of paraesthesia in over half the participants for up to 9 weeks (p < 0.001) [296]. However, a small pilot randomised, placebo-controlled, double-blind study found no significant effect of venlafaxine in the prevention of OIPN [297].

CIPN is a major dose-limiting side effect of chemotherapy, and the burden of CIPN continues to increase with increasing cancer-survivorship. Clinical guidance for the treatment of CIPN highlights the paucity of preventative strategies and symptom management. The diagnosis and assessment of CIPN lacks a reference standard, with studies utilising heterogeneous CIPN assessment tools dependent on self-reported outcome measures. The recent ACTTION recommendations endorse a pathomechanism-driven treatment discovery approach to CIPN. CCM may provide an adjunct to NCS in natural history studies and trials of disease-modifying therapies. Detailed mechanistic research in CIPN and CIPN-related neuropathic pain is needed to address the substantial burden on the patient, families and society.