Platinum-Induced Neurotoxicity and Preventive Strategies: Past, Present, and Future

Implications for Practice:

Neurotoxicity is a burdensome side effect of platinum-based chemotherapy that prevents administration of the full efficacious dosage and often leads to treatment withdrawal. This review summarizes preclinical and clinical evidence of pathogenesis and pathophysiology of platinum-induced peripheral neurotoxicity, as well as available evidence of neuroprotective and therapeutic strategies. These data may help to develop alternative options in the treatment of platinum-induced neuropathy, studies on in vitro models, and appropriate trials planning to find the best patient-oriented solution.

Introduction

Since the discovery of cisplatin in the mid-1960s, many platinum compounds (more than 3,000 compounds) have been developed. Thirty-five of these compounds have exhibited adequate pharmacological advantages (e.g., reaching sufficiently high plasma levels not associated with common toxicities, such as renal toxicity and thrombocytopenia) [1]. Some of them have been registered or are being considered for registration for treatment of different cancers, such as the second (carboplatin, nedaplatin, tetraplatin, and iproplatin) and third (oxaliplatin, lobaplatin, heptaplatin, satraplatin, and LA-12) generation, usually with better safety profiles [2–4].

Despite the efficacy of platinum analogs in cancer treatment, serious side effects, especially peripheral sensory neurotoxicity, often prevent their administration at their full efficacious doses or may considerably affect the quality of life of cancer patients being treated with them [5, 6]. Cisplatin was the first heavy metal used in several kinds of solid tumors, including lung, ovary, testis, bladder, head and neck, and endometrium [7, 8]; most patients develop a symptomatic neuropathy [9]. Second and third generations of platinum compounds have emerged in attempts to reduce the toxicity of cisplatin. Carboplatin, a second generation of platinums used to treat ovarian, non-small cell lung, and refractory testicular cancers, was thought to be associated with a lower risk of developing neurotoxicity [9]. However, the most recent Cochrane review comparing the toxicity of carboplatin versus cisplatin in combination with third-generation drugs for advanced non-small cell lung cancer reported an almost two times higher rate of neurotoxicity in the carboplatin group [10]. Oxaliplatin, as a widely used third-generation platinum analog approved for use in the treatment of metastatic colon cancer, is reported by the Food and Drug Administration to be responsible for more than 70% rate of symptomatic neurotoxicity with any severity [11] and often leads to treatment discontinuation [12–14]. In other studies, approximately 80% of colorectal cancer patients treated with oxaliplatin alone or in combination with other chemotherapeutics experienced neurotoxicity [15–17], and impairment may be permanent. Because the number of patients being treated with a neurotoxic agent is increasing, it is essential to understand the nature of such a problematic side effect. Furthermore, testing and validating available protective strategies in preclinical and clinical settings should be the next steps in overcoming platinum-induced peripheral neurotoxicity.

Clinical Features of Neurotoxicity

Platinum drugs are almost always given in combination with other chemotherapy drugs and/or radiation that may be neurotoxic in their own right. Early presentation of peripheral neurotoxicity can be with numbness, tingling, or paresthesia in fingers and/or toes, a decreased distal vibratory sensitivity, and/or loss of ankle jerks [5]. Moreover, prolonged treatment may also affect proprioception, which may result in ataxic gait.

Oxaliplatin and cisplatin are the two most commonly used neurotoxic platinum agents. Platinum-induced peripheral neurotoxicity can present as two clinically distinct syndromes. The acute transient paresthesia in the distal extremities, which is only commonly seen with oxaliplatin, usually occurs within the early phase of drug administration, whereas the chronic cumulative sensory neuropathy causes more persistent clinical impairments [5]. The latter deteriorates with cumulative doses [18], followed by “coasting,” wherein symptoms worsen even months after treatment withdrawal. Furthermore, patients can develop Lhermitte’s syndrome, which is a shocklike sensation of paresthesia radiating from the neck to the feet triggered by neck flexion. This phenomenon indicates the involvement of the centripetal branch of the sensory pathway within the spinal cord [19]. Neuropathy can also become irreversible. In a prospective multicenter study, Argyriou et al. [20] reported that oxaliplatin can result in an acute and chronic rate of neuropathy in 85% (169 patients of 200) and 73% (145 patients of 200) of patients, respectively.

Hearing loss or ototoxicity is another progressive and irreversible adverse effect of platinum chemotherapy [21] with a high frequency of almost 88% [22], which usually presents bilaterally and can occur during or years after treatment [23]. Nevertheless, the risk of ototoxicity may vary between cisplatin, carboplatin, and oxaliplatin treatments, and cisplatin is believed to be the most ototoxic and oxaliplatin is believed to be the least [24]. In one study, 19%–77% of patients treated with cisplatin developed bilateral sensorineural hearing loss, and 19%–42% developed permanent tinnitus [25]. Cisplatin accumulates in the cochlear tissue, forms DNA adducts, and causes inefficient and dysfunctional protein and enzyme synthesis leading to apoptosis of auditory sensory cells [26].

Diagnosis and Evaluation

The clinical diagnosis is generally not very difficult [27]. Nerve biopsies and neurophysiologic assessments are helpful for the examination of pathological and functional nerve damage (e.g., demyelinating versus axonal pathology; abnormalities in nerve conduction studies, somatosensory evoked potentials, magnetic resonance imaging, threshold tracking techniques, and quantitative sensory testing) [27]. Objective electromyography assessment of motor nerve excitability is a sensitive and specific endpoint of acute oxaliplatin-induced motor nerve hyperexcitability, which has the advantage of being widely available [28, 29]. Additionally, the threshold tracking technique is used to assess axonal excitability [30]. This technique allows the detection of sensory axonal dysfunction before clinical symptoms [18] and can be used as a predictive marker for nerve dysfunction.

Chemotherapy-induced peripheral neurotoxicity is typically a multidisciplinary medical issue, leading to different terminology, measurement, clinical evaluation, and grading, precluding the reliability of neurological assessment. However, standardization is improving. The Functional Assessment of Cancer Therapy/Gynecologic Oncology Group-Neurotoxicity Scale, the FACT-Taxane scales, the Patient Neurotoxicity Questionnaire, European Organization for Research and Treatment of Cancer (EORTC) quality of life questionnaire [QLQ] to assess chemotherapy-induced peripheral neuropathy, and the EORTC QLQ C30 questionnaire are scoring systems that have been used for neurotoxicity assessment to quantify the impact of chemotherapy-induced neurotoxicity on patients’ quality of life [31]. Among the questionnaires, the EORTC questionnaires are widely used nowadays [32]. Among different common toxicity criteria scales that are used for peripheral neurotoxicity assessment, the one developed by the Eastern Cooperative Oncology Group and National Cancer Institute (NCI-CTC) is most widely used [19, 33–35]. Although the reliability of different assessment methods has been tested in different settings, there are vast discrepancies between patient perception and objective tools, particularly in intermediate grades [32, 36], which increase the need for a more effective and standardized method [19].

Nature of Neurotoxicity

The pathophysiology of platinum-induced peripheral neurotoxicity is not completely elucidated. Based on available data, platinum compounds may actively enter the tumor and normal cells through organic cation transporters [37], organic cation/carnitine transporters [38], and some metal transporters, such as the copper transporters [39, 40]. Platinum compounds can be excreted via platinum efflux transporters (e.g., ATP7A, ATP7B, and MRP2) [41–45] (Fig. 1). The platinum adducts are formed intracellularly because of a hydrolysis process [46], resulting in interstrand cross-links, intrastrand cross-links and/or DNA-protein cross-links with platinum, affecting DNA synthesis in cancer cells [47] and mediating apoptosis [48]. Extensive DNA repair is considered as a major mechanism of chemotherapy resistance (Fig. 1), but efficient DNA repair can possibly prevent development of neurotoxicity. In dividing tumor cells, the formation of DNA adducts is supposed to cause growth inhibition and cell kills, hence eliminating the tumor cells.

Platinum products accumulate in the dorsal root ganglia (DRG), which is the main target, and in peripheral neurons (Fig. 2). Because these cells are postmitotic and not dividing, the formation of DNA adducts is not lethal, although the extent of DNA cross-links in DRG neurons at a specific cumulative dose strongly correlates with the degree of neurotoxicity [49]. Cisplatin produces approximately three times more adducts in the DRG compared with oxaliplatin [50], which is consistent with its higher neurotoxicity [12]. Platinum adducts probably cause axonal changes secondary to the neuronal damage [51], whereas brain and spinal cord are to some extent protected by the blood-brain barrier (BBB) [52]. However, there are data showing that cisplatin crosses the BBB and can accumulate when repeated dosages are given [53]. This can cause demyelination and vacuolar changes in the white matter [54]. It is believed that impaired axonal voltage-gated sodium channels kinetics can interfere with channel kinetics by oxalate (a metabolic product of oxaliplatin) [55] and can also reduce sodium ions current [56]. Sittl et al. [57] also showed that cooling in the presence of oxaliplatin induced bursts of action potentials in myelinated A but not unmyelinated C-fibers from human and mouse peripheral axons. Consequently, these alterations led to enhanced resurgent and persistent current amplitudes in large, but not small, diameter DRG neurons. Potassium channel blockade and calcium chelation are also two other etiologic possibilities [56, 58–60]. Besides, oxidative stress and mitochondrial dysfunction are regarded as another probable etiology of the apoptosis [61]. Podratz et al. [62] showed that cisplatin might inhibit mitochondrial DNA replication and cause mitochondrial vacuolization and degradation in DRG neurons in vitro and in vivo. These events can, to some extent, explain the mechanism of the neurotoxic effect of platinum compounds.

Pharmacogenetics

Single-nucleotide polymorphisms (SNPs) may play a key role in determining the induction of neurotoxicity, as well as apoptosis, because they may impair DNA repair pathways, including genes in base excision repair, nucleotide excision repair, mismatch repair, and double-strand break repair pathways [63] (Fig. 2). Moreover, SNPs can alter the drug metabolism, cell cycle control, detoxification, or excretion pathways, which finally may lead to drug toxicities, e.g., neurotoxicity. Several studies evaluated the pharmacogenetic association of SNPs with potential functional changes in the encoded protein that play a role in drug disposition, metabolism, and detoxification, DNA repair, and cancer-cell resistance and that may lead to platinum peripheral neurotoxicity [19]. However, the results are scattered and diverse with several methodological flaws, including small sample size, retrospective study design, and the implementation of a post hoc analysis of oncology-based databases of different, not preplanned sizes as well as lacking a prestudy hypothesis based on the known role of the investigated targets in the peripheral nervous system and the inappropriate outcome measures for neurological impairment [64] (Table 1).

There are controversial reports on the association of polymorphisms in some genes with platinum-induced neurotoxicity. These genes include ATP-binding cassette subfamily B member 1 (ABCB1) [65–68], ATP-binding cassette subfamily C member 1 or 2 (ABCC1, C2, or CG2) [69, 70], alanine-glyoxylate aminotransferase (AGXT) [69, 72, 73, 77, 94], cyclin H (CCNH) [70], catechol O-methyltransferase (COMT) [76], cytochrome P450s (CYPs; e.g., CYP2C8, CYP3A5 exons 3 and 5) [65–68], excision repair cross-complementation group 1 (ERCC1) and ERCC2 (alias XPD, xeroderma pigmentosum group D) [67, 68, 71, 74, 75, 77–87, 88], integrin β3 (ITGB3) [92], glutathione S-transferases (e.g., GSTM1 [69, 77, 78, 85, 86, 88, 89–91], GSTM3 [84, 75], and GSTT1 [88, 91]), voltage-gated sodium channel genes (SCNAs) [20, 71, 98], thiopurine S-methyltransferase (TPMT) [76], and x-ray repair cross-complementing protein 1 (XRCC1) [71, 73]. Although some data support the role of the mentioned genetic variations in the presentation and severity of platinum-induced peripheral neurotoxicity, the results are scattered and diverse (Tables 1, 2), which may form leads for future research.

ABCB1, ABCC1, ABCC2, ABCG2, and probably several other subfamily members mediate the cellular trafficking of drugs, their metabolites, and their endogenous factors, e.g., platinum efflux [99, 100]. CCNH plays an important role in the cell cycle progression, the transcriptional activity of the RNA polymerase II, and the DNA repairing process [101]. Thus, it may deregulate the repair after platinum damage to the dorsal root ganglia neurons [102]. COMT and TPMT, which encode enzymes that metabolize catecholamine-containing chemical and thiopurine drugs via methylation [103], respectively, might be associated with cisplatin-related hearing loss [104, 105].

Glutathione S-transferases (GSTs), a family of enzymes that have an important role in detoxification, have been extensively studied for the relation of SNPs with neurotoxicity induced by platinated compounds. GSTs are involved in detoxification through glutathione conjugation of electrophilic compounds (e.g., GSTM1 and GSTM3). A GSTP1 SNP (rs16953), for example, has been investigated in relation to peripheral neurotoxicity of platinum compounds in 24 studies (Table 1). Among these, 9 studies reported an association of this SNP with the course and severity of peripheral neurotoxicity [68, 74, 77, 80, 85, 88–90, 93], whereas other researchers reported contradicting results in 15 studies with regard to the association of GSTP1 gene variants with neurotoxicity [67, 71, 72, 75, 78, 81, 83, 84, 86, 87, 91, 94–97]. Moreover, recent meta-analysis showed no significant associations between GSTP1 Ile105Val polymorphism and oxaliplatin-induced neuropathy in a dominant model (odds ratio [OR] = 1.08, 95% confidence interval [CI] 0.67–1.74, p = .754), a recessive model (OR = 1.67, 95% CI 0.56–4.93, p = .357), and allelic analysis (OR = 1.22, 95% CI 0.67–2.24, p = .513) [106]. This inconsistency between the findings might be explained by the difference in the cancer type, ethnicity of the population studied, and/or number of the patients enrolled in each study [19].

Other studies evaluated the association of platinum-induced peripheral neurotoxicity with different SNPs in ERCC1 [67, 68, 77, 71, 74, 78–87], ERCC2 [71, 88], and XRCC1 [71, 73], which are parts of the nucleotide excision repair (ERCC1 and ERCC2) and base excision repair (XRCC1) pathways and are required for repair of DNA lesions [107]. Although Lee et al. [73] reported the polymorphism Arg399Gln (rs25487) in XRCC1 associated with less grade 2–4 sensory neuropathy in Korean patients treated with oxaliplatin-based treatment, the recent meta-analysis found it to be generally associated with poor clinical outcomes [108]. AGXT prevents accumulation of glyoxylate in the cytosol by converting it into glycolate, which is subsequently metabolized by lactate dehydrogenase into oxalate, the metabolite of oxaliplatin [109]. Pharmacogenetic analyses evaluated also cytochrome P450s [65–68], which are major enzymes of drug metabolism and bioactivation (e.g., CYP2C8 and CYP3A5), and ITGB3 [92], which belongs to the large family of integrins, known to participate in cell adhesion and cell surface-mediated signaling.

A recent study has provided evidence that SNPs in voltage-gated sodium channel genes (SCNAs; e.g., SCN4A-rs2302237 and SCN10A-rs1263292) can play a causal role in oxaliplatin-based peripheral neurotoxicity [20, 57] (Table 1). A polymorphism in SCN1A (rs3812718) was also reported to be associated with decreased neurotoxicity [85]. However, these results still need to be validated by appropriate larger and prospective studies. Won et al. [71], in a genome-wide pharmacogenomic approach, identified nine novel polymorphisms associated with and predictors of severe oxaliplatin-induced peripheral neurotoxicity, including rs10486003 (tachykinin, precursor 1 [TAC1]), rs2338 (forkhead box C1 [FOXC1]), rs830884 (integrin α1 [ITGA1]), rs843748 (acylphosphatase 2, muscle type [ACYP2]), rs4936453 (B-cell translocation gene 4 [BTG4]), rs17140129 and rs6924717 (phenylalanyl-tRNA synthetase 2 [FARS2]), rs12023000 (calcium/calmodulin-dependent protein kinase II inhibitor 1 [CAMK2N1]), and rs797519 (deleted in lymphocytic leukemia, 7 [DLEU7]) [71]. These genes may account for the mechanism of neurotoxicity prevention by calcium-magnesium infusions or may be associated with the important oxalate and glyoxylate outcome pathway [72]. However, none of the SNPs in the discovery samples (96 patients with colon cancer) surpassed genome-wide significance, and these SNPs were not significant in their validation set (247 patients with colorectal cancer; p = .05–.19) [71]. However, the authors noted that this limitation might be overcome by increasing the sample size in a prospective analysis.

Some evidence demonstrated that mitogen-activated protein kinase pathways, including extracellular signal-regulated kinase 1/2 (ERK1/2), c-Jun N-terminal kinase (JNK)/stress-activated protein kinase (Sapk), and p38, might also have a causal effect on chemotherapy-induced peripheral neuropathies [110]. Normally, there is a balance between ERK1/2 and p38 activation, which regulates neuronal apoptosis, and JNK/Sapk, which preserves neuronal degeneration. This balance is also altered by platinum derivatives [109].

Neuroprotection

Neuro- Versus Chemoprotection

It is essential to demonstrate whether the application of a neuroprotective agent might diminish the efficacy of the therapeutic agent. However, this potential adverse effect has been tested with different agents. As an example, cisplatin and carboplatin are used in induction and myeloablative chemotherapy for high-risk neuroblastoma, but because of significant ototoxicity in children, their administration may be compromised. Harned et al. [169] showed that the exposure of six neuroblastoma cell lines to STS, at 6 hours after cisplatin, did not bind to and eliminate the circulating cytotoxic compound and thus did not affect the antitumor effect of the platinum agent, even under hypoxic conditions. However, a significant undesired protection against cisplatin cytotoxicity was seen when the neuroblastoma cells were simultaneously exposed to both cisplatin and STS combinations. Moreover, Harned et al. [169] demonstrated that in a subcutaneous neuroblastoma xenograft model in nu/nu mice, mice receiving cisplatin alone or cisplatin plus STS after 6 hours had significantly better progression-free survival rates (p < .03) compared with controls or mice treated with concurrent cisplatin and STS administration. Likewise, Muldoon et al. [170] reported that delaying the administration of STS for 6–8 hours after carboplatin did not reduce its antitumor activity in a human small cell lung cancer xenograft model in the rat, but still protected against ototoxicity in guinea pigs. Moreover, Dickey et al. [171] found that adding STS simultaneously or up to 2 hours postcisplatin protected against the antitumor effect of cisplatin in glioblastoma, SKOV3 ovarian carcinoma, medulloblastoma, and small cell lung cancer cell lines, but that delayed STS administration for 6 hours did not show a significant chemoprotection in any of the cell types.

The use of STS to prevent hearing loss in children with a variety of malignancies has been tested in two phase III randomized trials SIOPEL6 (NCT00652132) [172] and COG ACCL0431 (NCT00716976) [173]. In the preliminary report of COG, presented at the 2014 annual American Society of Clinical Oncology (ASCO) meeting [173], a protective effect of STS was found in reducing the proportion of hearing loss compared with observation (29% versus 55%; p = .006). In this trial, 126 cancer patients were randomized to either cisplatin infusions alone or, to prevent cisplatin-induced hearing loss, combined with STS at 16 g/m² IV over 15 minutes beginning 6 hours after the completion of each cisplatin dose. The median postdiagnosis follow-up was 2.1 years. However, the potentially lower survival seen in the patients with disseminated disease receiving STS (event-free survival 60% versus 70%, p = .53; overall survival 75% versus 89%, p = .50) raises some concern of a tumor-protective effect of STS.

Treatment

The efficacy of some antidepressants, anticonvulsants, and a topical gel has been tested in six trials for treatment of platinum-induced peripheral neurotoxicity. Smith et al. [202] studied the effect of duloxetine in a randomized, placebo-controlled, crossover trial of 231 patients with either platinum or taxane neurotoxicity. Patients received 30 mg of duloxetine for the first week and 60 mg of daily duloxetine for 4 more weeks. Duloxetine significantly reduced pain and paresthesia, especially in oxaliplatin group. In contrast, 50 mg of daily amitriptyline or 100 mg of nortriptyline, in two separate trials, failed to demonstrate any significant improvements in patient-reported sensory symptoms, as well as objective scorings [203]. Similarly, trials testing gabapentin at a target dose of 2,700 mg/day [204] or lamotrigine at a target dose of 300 mg/day [205] failed to demonstrate any benefit for treatment of peripheral neurotoxicity. Finally, one trial evaluated a compounded topical gel containing baclofen (10 mg), amitriptyline HCl (40 mg), and ketamine (20 mg) on 208 randomly allocated patients [206], and a significant improvement in motor subscale scores was observed.

Future

There are some promising results favoring the ability of some neuroprotective agents to reduce the rate of subsequent neurotoxicity induced by platinum analogs. However, the most recent update of the Cochrane review on chemo-neuroprotective agents found insufficient data to conclude that any of the available chemoprotective agents is sufficiently effective in preventing or limiting the neurotoxicity of platinum drugs. Albers et al. [9] reviewed 29 randomized controlled trials (RCTs) or quasi-RCTs, in which 2,906 participants received chemotherapy with cisplatin or related compounds. Patients were also evaluated for quantitative sensory testing (primary outcome) or other measures including nerve conduction or neurological impairment rating using validated scales (secondary outcomes) before and 6 months after completing chemotherapy (Table 3). Likewise, the most recent ASCO Clinical Practice Guideline [123] based on a systematic review on 48 RCTs, including 35 RCTs on platinum-induced peripheral sensory neurotoxicity, did not recommend any established agent for the prevention of platinum-induced peripheral neurotoxicity. Only for the treatment of existing oxaliplatin neuropathy, they advised duloxetine, for which intermediate strength of evidence is present, considering the balance between benefit and harm [123].

Altogether, no neuroprotective strategy can yet be recommended for prevention and treatment of platinum-induced neurotoxicity, with a possible exception for duloxetine for oxaliplatin. There is a genetic diversity between patients, leading to differences in drug response including the side effects; hence a neuropathy preventive strategy should be individualized for each patient.

A pharmacogenetic approach might be useful in understanding the cause of peripheral neurotoxicity and tailoring the most suitable chemotherapy for each patient. A genome-wide pharmacogenomic approach may also be useful in identifying novel polymorphism predictors of severe platinum-induced peripheral neurotoxicity that may be used in personalized chemotherapy [71]. However, it is highly recommended that the positive and negative effects of the antineoplastic agents be studied in detail in preclinical settings before implementation in clinical practice. In particular, the central nervous system of animals can be used to quantify the effects of platinated compounds on neurons that corroborate clinical data and suggest them as suitable models for studying possible neurotoxicity of platinum agents [111–121]. Some in vitro models can also be used to investigate morphological parameters affected by platinum compounds [122]. These models enable measuring the effect of the drugs on neurons along with testing the neurogenic potential of neuroprotective compounds.

Conclusion

Our knowledge about the pathophysiology of platinum-induced peripheral neurotoxicity and suggested neuroprotective strategies is diverse and not adequately powered. Therefore, a thorough investigation of available evidence is important to design new, solid studies to tailor appropriate treatment to individual patients. This will minimize the burden of peripheral neurotoxicity, optimizing the potentially positive impact of the chemotherapeutic medication.

Author Contributions

Conception/Design: Abolfazl Avan, Tjeerd J. Postma, Godefridus J. Peters

Provision of study material or patients: Abolfazl Avan, Godefridus J. Peters

Collection and/or assembly of data: Abolfazl Avan

Data analysis and interpretation: Abolfazl Avan, Tjeerd J. Postma, Guido Cavaletti, Elisa Giovannetti

Manuscript writing: Abolfazl Avan, Tjeerd J. Postma, Cecilia Ceresa, Amir Avan, Guido Cavaletti, Elisa Giovannetti, Godefridus J. Peters

Final approval of manuscript: Abolfazl Avan, Tjeerd J. Postma, Guido Cavaletti, Elisa Giovannetti, Godefridus J. Peters

Disclosures

The authors indicated no financial relationships.