A Systematic Review of Herbal Medicine for Chemotherapy Induced Peripheral Neuropathy

Background Chemotherapy-induced peripheral neuropathy (CIPN) is a common adverse effect in cancer patients. The aim of this review was to assess the effectiveness of herbal medicine in preventing and treating CIPN. Methods Randomised controlled trials were included in this review. Extracting and assessing the data independently, two authors searched 13 databases. Results Twenty-eight trials involving 2174 patients met the inclusion criteria. Although there were some exceptions, the methodological quality was typically low. Seventeen trials reported the incidence rate of CIPN assessed by various tools and 14 showed a significant difference regarding the decrease of the incidence rate between the two groups. For clinical improvement, 12 trials reported it using various tools and 10 showed a significant difference between two groups. Two cases of adverse events occurred in one trial; the other nine trials reported no adverse events. Conclusions We found that herbal medicines in combination with and/or without other therapies potentially have preventive or therapeutic effects on CIPN. However, conclusions cannot be drawn because of the generally low quality of the methodology, the clinical heterogeneity, and the small sample size for each single herbal medicine. Trials that are more rigorous and report sufficient methodological data are needed.


Introduction
Chemotherapy induced peripheral neuropathy (CIPN) is a common and stubborn adverse effect of chemotherapy agents such as platinum agents, taxanes, vinca alkaloids, epothilones, thalidomide, and bortezomib [1,2]. Approximately 20 to 40% of cancer patients suffer from CIPN [3,4]. Although the mechanisms of CIPN are not fully understood, recent research has shown that the dorsal root ganglion (DGR) of the neural cell bodies is the most common target of chemotherapy agents [1,5]. Microtubule disruption (in the case of taxanes, epothilones, vinca alkaloids, and bortezomib), the nerve terminal damage (in the case of vincristine and taxanes), and dysregulation of neutrophins (in the case of thalidomide and bortezomib) also contribute to inception of CIPN [4,5]. Clinical features of CIPN vary according to the chemotherapy agents used and the site of their action. CIPN is primarily manifested in symptoms of sensory neuropathy, including pain, tingling, numbness, and temperature sensitivity in the hands and feet. Occasionally, when paclitaxel and vincristine are used, the motor neuron can be damaged [4][5][6], and autonomic nerves may be affected by the use of vinca alkaloids [5,7]. Although CIPN is generally dose dependent, it can occur at any time during chemotherapy or after the end of treatment, and it can take different form. Oxaliplatin induced neurotoxicity can occur in a clinically acute and chronic form: the acute form is thought to occur as a result of a dysfunction of nodal axonal voltage, and chronic toxicity is thought to be associated with functional changes in the DGR [5,8]. Late presentations are commonly associated with cisplatin [1,9].
Because of these clinical features, CIPN decreases quality of life (QoL). It also enforces treatment delays and dose reductions of chemotherapeutic agents, thereby limiting their efficacy [1,5]. Numerous pharmacological and nutraceuticals options have been tested for the treatment of CIPN 2 Evidence-Based Complementary and Alternative Medicine including amifostine, glutamine, glutathione, oxcarbazepine, antiepileptics, venlafaxine, vitamin B6, vitamin E, omega-3 fatty acids, acetyl-L-carnitine, and duloxetine [1,5,6,10,11]. Although these agents have been shown to be effective in the prevention and treatment of CIPN, the data are insufficient to demonstrate the efficacy of pharmacological drugs and nutraceuticals [1,5,12].
Herbal medicine has been developed and used widely in East Asia; it is based on the unique theories of Yin and Yang, the five elements, and the visceral manifestation theory [13]. In modern times, herbal medicine is prescribed as therapeutic agents and as an adjunctive method for managing diverse diseases, including cancer [14,15]. Numerous experimental and clinical studies have investigated the effectiveness of herbal medicine and have shown its potential effectiveness in managing CIPN through the action of antioxidants and the enhancement of nerve growth [16,17]. Although some systematic reviews have been published on the effectiveness of herbal medicine for oxaliplatin induced peripheral neuropathy [18,19], to the best of our knowledge, there is no systematic review of randomised controlled trials (RCTs) for the use of herbal medicines in CIPN, regardless of the chemotherapeutic agents or the herbal medicine. To assess the effectiveness of herbal medicines for CIPN, we therefore conducted this systematic review of RCTs without seeking to exclude any particular chemotherapeutic agents or herbal medicines.

Search Strategy.
We searched the following 13 electronic databases, with no limitations on the publication date or the language: MEDLINE, the Cochrane Central Register of Controlled Trials (CENTRAL), EMBASE, AMED, the China National Knowledge Infrastructure (CNKI), the Wanfang Database, the CQVIP Database, and six Korean medical databases (Korean Studies Information, DBPIA, the Korea Institute of Science Technology Information, the Research Information Centre for Health Database, the Korean Traditional Knowledge Portal, and KoreaMed).
The search strategy was developed for MEDLINE (see Appendix A) and CNKI (see Appendix B). We also modified it for each database, based on the search terms. We completed the searches through 17 May 2017.
We registered the protocol of this review in PROSPERO (Available at https://www.crd.york.ac.uk/PROSPERO/display record.asp?ID=CRD42017054369.

Eligibility Criteria.
We included RCTs that investigated herbal medicine for preventing and treating CIPN in cancer patients, regardless of time or language. The participants included in the study met the following criteria: (1) age 18 years or older, (2) patients diagnosed with cancer who had received chemotherapy, regardless of type of cancer, sex, race, or geographic location, and (3) the CIPN diagnosed by clinical assessment or additional investigation, such as nerve conduction velocity. We included data from parallel-group studies for the meta-analysis. In the case of crossover trials, only the first treatment period data was analysed. If we were not able to separate the results of the first and second periods in the crossover trial, we excluded the study. The RCTs in which the data in the methods or the results section were unreliable or unavailable were excluded from the studies.
We included all types of herbal medicines and their major ingredients used for medicinal purposes, regardless of their origin, the composition of prescription, the mode of delivery (e.g., oral, intravenous infusion using fluid, intravenous injection, fumigation, external application, or hand and foot baths), and the dosage or the duration of the treatment. The comparators included no treatment, a placebo, and conventional therapeutic agents. We excluded those trials that used the other type of herbal medicine as a comparison and those that did not use the same baseline therapy.
The primary outcome measures were clinical improvement, incidence rate, and nerve conduction study (NCS). The clinical improvement and the incidence rate were measured by objective methods, including such things as the National Cancer Institute Common Terminology Criteria for Adverse Events (NCI-CTCAE) grade and Levi's grade. We assessed the clinical improvement using three grades with reference to Deng et al. 's 2016 article [19]: complete remission (CR), partial remission (PR), and being not perceptible (NP). CR is completely diminished symptoms and/or the CIPN grade has been reduced to zero. PR is the improvement of symptoms and/or a reduction of more than one grade. NP is either no reduction of the grade or no change of the symptoms, in comparison to the case before treatment. The effective rate is the sum of CR and PR. The NCS included motor and sensory nerve conduction velocity assessed by validated methods. The secondary outcome measures involved the patients' quality of life (QoL), as assessed by objective instruments and adverse effects. We also included information regarding adverse effects induced by chemotherapy.

Data Extraction and Assessment of Methodological
Quality. The two authors (Noh and Park) assessed the titles and abstracts of the studies retrieved from electronic databases and each determined their eligibility for inclusion independently. Hard copies of the relevant studies were retrieved and then the authors used a standard data extraction form to extract the data independently. The form included methodology, participants, intervention, periods of treatment, outcomes, and conclusions. The two authors (Noh and Park) assessed the risk of bias using the Cochrane risk of bias tool [20] and the following items: random sequence generation (selection bias), allocation concealment (selection bias), blinding of participants and personnel (performance bias), blinding of outcome assessment (detection bias), incomplete outcome data (attrition bias), selective outcome reporting (reporting bias), and other biases. We evaluated the domains as "Yes," "No," or "Unclear," according to the Cochrane risk of bias criteria. When there was a discrepancy among the reviewers, a consensus was reached through discussion.

Data Analysis and Synthesis.
We used Review Manager 5.3 software to perform our data analyses. We used the risk ratios (RRs) with 95% confidence intervals (CIs) for dichotomous data and the mean difference (MD) with 95% Evidence-Based Complementary and Alternative Medicine CIs for continuous data. A meta-analysis was planned, which would be conducted using a random effect model, if there was sufficient data to pool. We planned to conduct a standardised mean difference analysis (SMD), with 95% Cis, in the case of the use of different measurement scales for continuous data. Heterogeneity was tested using 2 . In the case of heterogeneity, we planned to analyse the subgroup analyses.
Since the studies did not meet the qualifications, neither publication bias nor sensitivity analyses were performed.

Study Characteristics.
A total of 2174 patients in 28 trials were included; 72 patients were dropped from the study. The mean of the participant sample size was 77.64 people (range: 31 to 186). Eighteen trials studied gastrointestinal cancer, six tested patients with various types of cancer, two studied ovarian cancer or endometrial cancer, and one trial each tested patients with breast cancer and multiple myeloma. The duration of the treatment ranged from 7 days to 40 weeks. All trials described baseline characteristics that were comparable between the intervention groups with the control groups. Although three trials tested herbal medicine during the period of chemotherapy, they did not describe the accurate treatment period [28,34,38]. Twenty-six different formulas were investigated in 28 trials ( Table 3). The oral dosage form of herbal medicine, including decoction or granules, was used in 12 trials. In 11 trials, the herbal formula was used as some form of hand or foot bath. One trial applied herbal medicine through intravenous (IV) infusion and another applied it through fumigation. Three trials investigated oral herbal medicine     Guilong tongluo decoction: Ramulus Cinnamomi (9) (20) (30) combined with hand and foot baths of herbal medicine or the fumigation of herbal medicine. Goshajinkigan granules were used in five trials [24,28,30,35,43] and two used a guilong tongluo decoction [21,32]. The other 24 formulas were investigated in 21 trials. An oxaliplatin based regimen was used to treat cancer patients in 20 trials. A paclitaxel and carboplatin (PC) regimen was used in two trials [27,43] and a docetaxel based regimen was used in one [28]. CIPN patients were treated by various types of chemotherapy regimen in four trials [40][41][42]46]. One trial did not report on the chemotherapeutic agents in detail [48]. Regarding comparators, vitamin B12 was used in four trials [25,28,41,46], glutathione was used in two [23,26], a placebo was used in three [32,35,45], and simple hand and foot baths using warm water were used in three [36,39,40]. The other 16 trials compared interventions with no treatment. In seven trials, the common treatments were used in the treatment and control groups [22, 31, 42-44, 47, 48].

Risk of Bias in the Included Studies.
Nine trials reported the random sequence using a random number table or a computer number generator [25,26,28,30,33,37,40,46,47]. Two trials described an inadequate sequence generated by the date of admission [34,44]. The other 17 trials did not report their random sequence. One trial described allocation concealment done in patients and investigators [30] and two used an opaque sealed envelope [26,32]. Four trials reported double blinding of participants and investigators [30,32,35,45]. Not all of the 28 trials described whether the outcome assessors were blinded or not, although one trial blinded the data collectors. Nine trials reported dropouts [23,30,33,35,39,40,43,45,46] and one applied an intention to treat analysis [30]. For the other six trials, we judged that the proportion of missing outcomes was not sufficient to have a clinically relevant impact on the intervention effect estimate. The remaining 21 trials did not miss outcome data. Two trials reported their prespecified protocols [30,35]. Overall, with some exceptions, the methodological quality was not promising (Figure 2).

Effects of Interventions.
We divided the 28 included trials into two categories: (1) herbal medicine investigated to prevent CIPN in cancer patients receiving chemotherapy (18 RCTs, Table 1) and (2) herbal medicine investigated as therapeutic agents in cancer patients with CIPN (10 RCTs, Table 2).
We analysed the first category as three subgroups, according to the dosage form of the herbal medicine: (1) herbal medicine for hand and foot baths or fumigation, (2) herbal medicine for intravenous infusion, and (3) herbal medicine for oral dosage form. The second category was analysed as two groups: (1) herbal medicine for hand and foot baths or fumigation and (2) herbal medicine for the oral dosage form. We did not pool the data due to clinical heterogeneity for the 21 trials that investigated the differences in herbal medicines. Since different comparators, different outcome measures, or different common treatments were used, we did not perform a meta-analysis for the five and two trials that investigated ?

Preventive Effects: Herbal Medicine for Hand and Foot
Baths or Fumigation. Seven trials compared herbal medicine for hand and foot baths or fumigation with vitamin B12, hand and foot baths of warm water, and no treatment [25,27,29,31,33,34,36]. Six trials reported the incidence rate and all of the trials showed that herbal interventions were significantly effective in decreasing the incidence rate, as compared with the comparators: huoxue tongluo formula, Guo's specified formula, wenjing huoxue formula, siteng yixian formula, yiqi huayu formula for hand and foot baths combined with chaihu jia longgu muli decoction, and Li et al. 's specified formula for hand and foot baths combined with buyang huanwu decoction [25,27,31,33,34]. One trial referred to NCS and the fumigation of Guo et al. 's specified formula showed significant effectiveness in nerve conduction velocity and amplitude, as compared with warm water baths [33].

Preventive Effects: Herbal Medicine for IV Infusion.
One trial investigated the preventive effects of Astragali Radix through IV infusion in comparison to no treatment and to glutathione [26]. Astragali Radix through IV infusion showed a significant reduction of the incidence rate and improvement of KPS (MD 7.68, 95% CI 4.29 to 11.07) in comparison to no treatment. However, in comparing Astragali Radix and glutathione, no significant differences were found in the incidence rate or the KPS.

Preventive Effects: Herbal Medicine for Oral Dosage Form.
Ten trials tested the preventive effects of the six different herbal medicines in comparison to no treatment, vitamin B12, and placebo [21-24, 28, 30, 32, 35, 37, 38]. Ten trials reported the incidence rate of CIPN. For goshajinkigan granules, although Nishioka et al. reported significant effects on the reduction of the incidence rate [24,28], Kono et al. 's results showed no significant deference between the intervention groups and the control groups [30,35]. For the guilong tongluo decoction, Zhu et al. reported a significant reduction in the incidence rate in comparison to no treatment and a placebo [21,32]. The yiqi huoxue decoction combined with neurotropin, yiqi wenjing tongluo decoction, shaoyao gouteng muer decoction, and tanshinone II showed a significant reduction in the incidence rate in comparison to the comparators [22,37,38] [44]. In comparison, Ke's herbal formula [48] and the fumigation of a modified zhixiao xuanbi formula combined with a modified zhixiao xuanbi decoction did not show a significant improvement in the effectiveness rate [42]. However, Ke's herbal formula and the fumigation of a modified zhixiao xuanbi formula combined with a modified zhixiao xuanbi decoction were found to improve the sensory and motor nerve velocity significantly [42,48]. Two trials reported QoL using the Chinese Anti-Cancer Association form, and the results of the intervention groups showed significant improvements in comparison to the control groups [40,46].

Therapeutic Effects: Herbal Medicine for Oral Dosage
Form. Here, two trials were investigated. One compared the bawei yangmai decoction and vitamin B12 [41]. The other compared goshajinkigan granules combined with vitamin B12 and with vitamin B12 alone [43]. The bawei yangmai decoction was found to improve the effectiveness rate in comparison to vitamin B12 (WHO grade; RR 1.77, 95% CI 1.12 to 2.79). However, the goshajinkigan granules combined with vitamin B12 did not show a statistical difference between the two groups. Regarding QoL, the bawei yangmai decoction was found to improve the QoL, as assessed by the CACA form, in comparison with vitamin B12 (MD 6.84, 95% CI 3.72 to 9.96). The goshajinkigan granules combined with vitamin B12 did not show a significant improvement on FACT-Taxane, in comparison to vitamin B12 alone. NCS was not reported. Tables 1 and 2, ten trials reported adverse events of interventions [30,33,36,37,[39][40][41][45][46][47], while the remaining 18 trials did not mention adverse events. One trial reported that two cases of adverse events occurred in the wenjing tongluo formula group [45] Evidence-Based Complementary and Alternative Medicine 13 and the other nine trials reported no adverse events in the treatment and control groups. Eight trials reported adverse events associated with chemotherapy [21,23,24,26,28,32,35,36] and seven reported no significant difference between the intervention groups and the control groups. One trial reported that Astragali Radix for intravenous infusion significantly improved chemotherapy induced anaemia in comparison to no treatment [26].

Discussion
We conducted this systematic review to assess the effectiveness of herbal medicine for CIPN in cancer patients. We searched 819 research studies through 13 English, Chinese, and Korean databases and ultimately identified 28 trials for inclusion with a total of 2,174 patients. Eighteen of the included trials investigated the preventive effects of herbal medicines (used either individually or in combination with conventional therapies), and ten trials tested the therapeutic effects of herbal medicines. Five main outcomes were analysed to assess the effectiveness of herbal medicines combined with or in the absence of conventional therapies in relation to different comparators. Seventeen trials reported an incidence rate of CIPN assessed by various tools and 14 showed a significant difference between the two groups in the decrease of the incidence rate. Regarding clinical improvement, 12 trials reported it using various tools and 10 showed a significant difference between the two groups. Seven herbal formulas were associated with significant improvement in the rate of effectiveness, in comparison with the control groups. The NCS was studied in four trials, all of which indicated that different herbal medicines combined with/without conventional therapies had the advantage in terms of improving nerve conduction velocity. Nine trials reported QoL using KPS, the CACA form, or FACT-Taxane, and eight trials showed treatment groups superior to control groups in improving QoL. Ten trials reported adverse events (with two cases of adverse events occurring in one trial), and nine trials reported no adverse events. Nonetheless, we cannot conclude that the use of herbal medicines in CIPN is safe purely on the grounds that the remaining 18 trials did not report adverse events; in order for us to draw conclusions on safety, the documentation of adverse events must be in full compliance. In addition, eight trials reported adverse events associated with chemotherapy and seven reported no significant difference between the intervention groups and the control groups.
We found some noteworthy features in this review. A relatively large number of studies investigated and tested the external dosage form of herbal medicines, and approximately 15 different herbal formulas were investigated using hand and foot baths or fumigation. We found that due to the characteristics of CIPN-which is a kind of damage to the peripheral nervous system-the direct application of directly administering the herbal formula on lesions would be considered an effective dosage form. A significant variety of herbal formulas was investigated, with a total of 26 different formulas investigated in 28 trials. Noticeably, prescribing diverse herbal formulas for the same disease and modifying the herbs in a particular formula were very common in TEAM. Although these features would be useful in the clinical field, they additionally make it difficult to derive a conclusion regarding the effectiveness of the herbal medicines.
From this review, we found that herbal medicines, in combination with and without other therapies, have potentially preventive and/or therapeutic effects for CIPN. However, some limitations in the review prevent us from drawing clear conclusions. Although we planned to perform a meta-analysis in order to draw global conclusions about herbal medicine, the clinical heterogeneity induced by the different herbal formulas, different outcomes, and different comparators left us unable to pool the data. The participant sizes were also too small to draw conclusions about each herbal formula. A further issue is that, except for a few studies, such as Kono et al. 's research, the methodological quality of the studies was low; most of the included trials did not report sufficient information for random sequences and allocation concealment, and only four of them were performed in a double-blinded manner. As a result, biases might have occurred and influenced the apparent findings on the effectiveness of the herbal medicines.
In conclusion, this review found that herbal medicines showed potentially beneficial effects on CIPN. However, the small participant sizes and low methodological quality make it difficult to draw firm conclusions. More high-quality trials reporting sufficient methodological data should be conducted to corroborate findings of the studies in this review. In addition, more clinical homogeneous trials and further evidence of safety are required before drawing definitive conclusions concerning the effectiveness of herbal medicines.