Oral administration of herbal medicines for radiation pneumonitis in lung cancer patients: A systematic review and meta-analysis

Background Radiation pneumonitis is a common and serious complication of radiotherapy. Many published randomized controlled studies (RCTs) reveal a growing trend of using herbal medicines as adjuvant therapy to prevent radiation pneumonitis; however, their efficacy and safety remain unexplored. Objective The aim of this systematic review is to evaluate the efficacy and safety of herbal medicines as adjunctive therapy for the prevention of radiation pneumonitis in patients with lung cancer who undergo radiotherapy. Methods We searched the following 11 databases: three English medical databases [MEDLINE (PubMed), EMBASE, The Cochrane Central Register of Controlled Trials (CENTRAL)], five Korean medical databases (Korean Studies Information, Research information Service System, KoreaMed, DBPIA, National Digital Science Library), and three Chinese medical databases [the China National Knowledge Database (CNKI), Journal Integration Platform (VIP), and WanFang Database]. The primary outcome was the incidence of radiation pneumonitis. The risk of bias was assessed using the Cochrane risk-of-bias tool. Results Twenty-two RCTs involving 1819 participants were included. The methodological quality was poor for most of the studies. Meta-analysis showed that herbal medicines combined with radiotherapy significantly reduced the incidence of radiation pneumonitis (n = 1819; RR 0.53, 95% CI 0.45–0.63, I2 = 8%) and the incidence of severe radiation pneumonitis (n = 903; RR 0.22, 95% CI 0.11–0.41, I2 = 0%). Combined therapy also improved the Karnofsky performance score (n = 420; WMD 4.62, 95% CI 1.05–8.18, I2 = 82%). Conclusion There is some encouraging evidence that oral administration of herbal medicines combined with radiotherapy may benefit patients with lung cancer by preventing or minimizing radiation pneumonitis. However, due to the poor methodological quality of the identified studies, definitive conclusion could not be drawn. To confirm the merits of this approach, further rigorously designed large scale trials are warranted.


Introduction
Lung cancer is the most common type of cancer in both men and women, and one of the main causes of cancer death worldwide. In 2016, 243,820 new cases of lung cancer were expected, and estimated to comprise 27% of all male cancer deaths and 26% of female cancer deaths [1]. In Korea, lung cancer mortality rates in 2013 were 34.0% for both sexes, accounting for 49.5% of all male cancer deaths and 18.4% of all female cancer deaths [2].
Radiotherapy has been widely used for unresectable and locally advanced non-small cell lung cancer (NSCLC) and small-cell lung cancer (SCLC) [3,4]. However, the lung is more vulnerable to radiotherapy than other organs, and radiation leads to pulmonary toxicity [5,6]. Radiotherapy pneumonitis (RP) caused by radiation-induced lung toxicity is the most serious complication [7,8]. RP typically presents 1-6 months after radiation therapy [6,9,10]. The clinical features usually include mild dry cough, mild fever, and mild dyspnea, but in some cases, severe respiratory failure appears and leads to death [6,8,9]. The incidence of moderate to severe RP with radiotherapy is 10-20%, but varies in clinical studies [8,11]. When RP is left untreated for a long time, it may develop into pulmonary fibrosis, which has a high rate of mortality. The development of RP sometimes makes cessation of radiotherapy or control of the radiation dose necessary. RP not only decreases the treatment success rate for lung cancer, but also reduces the patient's quality of life. Therefore, it is very important to prevent or minimize the incidence of RP.
Despite great efforts to develop agents to reduce the severity and incidence of pulmonary toxicity resulting from radiotherapy, no effective agents currently exist [12]. Amifostine has broad applicability as a protective agent, but the results of clinical trials are controversial. The reported results have not been replicated, and present guidelines and systematic reviews do not support the use of amifostine for the prevention of RP [12,13].
Herbal medicines (HMs) are commonly used complementary and alternative therapies in cancer treatment [14]. Lung cancer patients also use HM while receiving chemotherapy or radiotherapy. A study reported that chemotherapy combined with administration of HMs increased the survival rate among patients with lung cancer [15]. A recent systematic review reported that Astragalus-containing HMs are effective at protecting against RP as adjunctive therapy during conservative radiotherapy [16]. There are also many published trials of HMs other than Astragalus that examined protective effects against RP, but the evidence for HMs protecting against or minimizing RP is not yet compelling. Therefore, we sought to carry out a comprehensive systematic review of the efficacy and safety of HM as adjunctive therapy to prevent RP in lung cancer patients receiving radiotherapy.

Methods
This study was registered with the international Prospective Register of Systematic Reviews (PROSPERO): CRD 42016048066 and the protocol was published [17].

Selection criteria
Types of studies and participants. Only randomized controlled trials (RCTs) were included. Crossover studies, observation studies and case studies were excluded. Patients were included if they were diagnosed with lung cancer, aged 18 or older, and planned to undergo radiotherapy, regardless of tumor stage.
Type of interventions. Studies reporting orally administered HM treatment as adjunctive therapy with radiotherapy were included regardless of the HM type. HMs refers to a treatment involving single herb or a combination of herbs. Studies that included other alternative and complementary therapies, such as acupuncture, moxibustion, massage, etc. were excluded. Trials using control groups receiving radiotherapy or radiotherapy combined with placebo control were included. We only excluded stereotactic radiation therapy among radiation techniques based on Radiation oncologist's opinion. Trials involving other types of therapy, including chemotherapy, were excluded.
Type of outcome measures. Primary outcomes: The rate of incidence f RP after radiotherapy was analyzed as the primary outcome. Trials that reported the diagnostic criteria for RP in terms of commonly used grading systems, such as the National Cancer Institute Common Terminology Criteria (NCICTC) for Adverse Events, the Radiation Therapy Oncology Group (RTOG) score, or the Common Terminology Criteria for Adverse Events (CTCAE) were included [18]. Studies using clinical criteria for RP, in which the diagnosis was made based on the patient's symptoms (including shortness of breath, intermittent low fever, cough, congestion, etc.) combined with radiological manifestations [19] were also included.
Secondary outcomes: The secondary outcome assessments consisted of the Karnofsky performance status (KPS) score, pulmonary function test-especially the diffusing capacity of the lungs for carbon monoxide (DLCO)-and adverse events.

Search methods for the identification of studies
We searched the following databases: three English medical databases (PubMed, EMBASE, The Cochrane Library), five Korean medical databases (Korean Studies Information, Research information Service System, KoreaMed, DBPIA, National Digital Science Library), and three Chinese medical databases (the China National Knowledge Database (CNKI), Journal Integration Platform (VIP), and WanFang Database). Related gray literature and references of included studies were hand searched. The databases were searched from their inceptions up to July 2017. The studies were not limited by language.
The key search terms were "radiation pneumonitis" and "herbal medicine." We used related Medical Subject Heading terms and synonyms in various combinations. The search strategy is presented in online S1 Appendix.

Data collection and analysis
Selection of studies. Two reviewers (KIK and JHJ) selected eligible studies according to the inclusion criteria. The studies were screened based on the study design, patients, intervention, comparator, and outcome, derived from the title and abstract. The full texts of the surviving studies were then reviewed independently for inclusion in the study. Discrepancies were resolved by discussion with a third researcher (HJJ), to determine, by agreement, the final selection of studies.
Data extraction and management. Data were independently extracted by two reviewers (KIK and BJL). Data including publication year, country, inclusion/exclusion criteria, lung cancer type (NSCLC, SCLC), tumor stage, diagnosis criteria for RP, patient's age, sex, interventions (composition, dosage, and type of the intervention, treatment duration), comparator, outcomes, adverse events, radiation dosage, randomization, blinding, and number of withdrawals and dropouts were recorded in an Excel spreadsheet. Any discrepancies in the data were resolved by discussion with another team member (HJJ).

Risk of bias assessment
Three reviewers (KIK, JHJ and BJL) independently assessed the risk of bias for each included study, according to section 5.1 of the Cochrane Handbook [20]. The following criteria were used: (1) random sequence generation, (2) allocation concealment, (3) blinding of participants and personnel, (4) blinding of outcome assessment, (5) incomplete outcome data, (6) selective outcome reporting, and (7) other sources of bias (baseline imbalance). We used "L," "H," and "U" as a code for the judgments, "L" indicating a low risk of bias, "H" indicating a high risk of bias, "U" indicating that the risk of bias was unclear.
Disagreements were resolved by discussion between all the reviewers. When necessary, we contacted the study authors to clarify protocols and obtain any missing information.

Data synthesis
Review Manager Software (RevMan, Version 5.3 for windows; Copenhagen, The Nordic Cochrane Centre, The Cochrane Collaboration, 2014) was used for data analysis. For dichotomous outcomes, we used the risk ratio, with 95% CI and p values, to assess the efficacy and safety of HM. For continuous data collected using the same measurement scale, we calculated the weighted mean difference (WMD) and 95% confidence intervals (CIs). A Chi-square test with a significance level of p < 0.1 was used to assess heterogeneity among studies. To assess inconsistencies among studies, the I 2 test was used. The I 2 statistic indicates the proportion of variability among studies not explained by chance alone, and an I 2 value > 50% or more is considered an indication of substantial heterogeneity. If heterogeneity existed in the pooled studies, a random effects model was applied. Otherwise, the data were analyzed using a fixedeffect model. Funnel plots were used to detect publication bias when 10 or more studies were included in a meta-analysis.

Assessment of reporting bias
A funnel plot and Egger's test will be used to detect publication bias when 10 or more studies are included in the meta-analysis.

Characteristics of the included studies
In this study, a total of 1418 studies were retrieved through electronic and manual searches. Duplicate studies were excluded. During the first screening process, 919 studies were excluded based on their titles and abstracts. After reviewing the full texts of 220 studies, 198 studies were excluded, and the remaining 22 studies  were finally included in this study. The reference of exclude studies in S2 Appendix. The study selection process details are described in  Herbal medicines for radiation pneumonitis in lung cancer patients A total of 1819 subjects were included in the analysis. Twenty-four subjects withdrew or dropped out.

Risk of bias assessment of the included studies
The risk of bias for each study was assessed according to section 5.1.0 of the Cochrane Handbook for Systematic Reviews of Interventions. Seven studies described the method of randomization, and six of them reported using the random number table method, hence were evaluated as "low" [25,32,36,37,39,42]. One study used patient medical record numbers, and was evaluated as "high" [33]. The other studies did not report any randomization procedure, and were evaluated as "unclear." The allocation procedure was not reported for any of the studies, so we evaluated all of them as "unclear." Since placebos were not used in any of the studies, blinding was not performed in any of them. However, when the primary outcome was the RP incidence rate, it was considered that clinical judgements about the occurrence of RP would not always be affected by blinding. Therefore, studies that clearly described the diagnostic criteria for RP, and confirmed RP on X-ray or CT scans, were evaluated as "low" However, studies in which subjective assessment parameters were included in the secondary outcomes, making estimation of the influence of blinding on the study results difficult, were evaluated as "unclear." Studies that did not include objective diagnostic tools such as medical imaging findings in the diagnostic criteria for RP, and performed subjective assessment, were evaluated as "high," since blinding could have affected the study results. Studies with incomplete outcome data were evaluated as "low" if they reported missing data and reasons for the missing data, and it was judged that the missing data did not affect the study results. Studies that did not report missing data were evaluated as "unclear." With regard to selective bias, four studies that did not present evaluation results for the outcome parameters were evaluated as "high." The remaining studies were evaluated as "unclear," since assessment of selective bias in these studies was not possible due to insufficient information. With regard to other bias, all studies were evaluated as "unclear," since full assessment of bias in these studies was not possible due to insufficient information (Fig 2).

Outcome
Incidence rate of RP. We analyzed the incidence rates of RP among patients who were administered HM during radiotherapy and patients who underwent radiotherapy alone, in 22 studies conducted on a total of 1819 clinical trial participants . The relative risk (RR) of RP was significantly less than 1 in a comparison between the HM plus radiotherapy group and the radiotherapy alone group (RR 0.53; 95% CI, 0.45-0.63) (Fig 3). However, an asymmetric funnel plot was observed, suggesting the existence of publication bias (Fig 4).
We also analyzed the incidence rate of severe radiation pneumonitis, scoring 3 or more on the RTOG scale. A total of 11 studies, involving 903 patients, were included in the analysis [21, 22, 28-33, 36, 37, 39]. The analysis result showed that the RR of radiation pneumonitis was significantly less than 1 in the HM plus radiotherapy group compared to the radiotherapy alone group (RR 0.22; 95% CI, 0.11-0.41) (Fig 5).
Effects on quality of life (QoL). As can be seen in Fig 6, six studies (involving 420 patients) that reported the KPS as mean and standard deviation were included [25,27,31,33,36,39]. The HM plus radiotherapy group showed significantly better performance status than the radiotherapy alone group (WMD 4.62, 95% CI 1.05-8.18). Only six trials reported performance status, so a funnel plot was not applicable.
Effects on lung function. One study was identified, including 78 patients that reported lung functions [35]. The forced vital capacities (FVC) was 2.44±0.40 in the HM plus radiotherapy group and 2.28±0.28 in the radiotherapy alone group. The forced expiratory volumes in 1 second (FEV1) was 2.02±0.34 in the HM plus radiotherapy group and 1.83±0.40 in the radiotherapy alone group. The diffusing capacity of the lungs for carbon monoxide (DCLO) was 13.36±3.96 in the HM plus radiotherapy group and 11.22±2.88 in the radiotherapy alone group. The differences in all these lung functions between the two groups were statistically significant (p<0.05). [25][26][27]31]. Of these, three studies reported no side effects in either the HM plus radiotherapy group or the radiotherapy alone group [25,26,31]. One study reported mild diarrhea in two patients from the HM plus radiotherapy group [27]. For assessment of the safety of HMs, liver function tests (LFT) and renal function tests (RFT) were performed in these four studies. No abnormal findings were identified.

Adverse events (AEs). Four trials investigated the side effects of HMs
Herbal composition and frequency. Twenty-two studies reported the herbal formula, in the form of a decoction, capsules, granules, etc. The herbal formulas used in combination with radiation therapy consisted mainly of those that tonify yin, tonify qi, and nourish blood. The detailed herbal compositions are shown in Table 2. Among them, Ophiopogonis Radix was the most commonly used herb for lung cancer as adjunctive therapy with radiotherapy. Ophiopogonis Radix, Adenophorae Radix and Reheanniae Radix Praeparata are also well-known herbs that nourish yin. Astragali Radix is a typical herb with qi-tonifying effects, and Angelicae Sinensis Radix is a classic herb that nourishes blood (Table 3).

Discussion
The present study reviewed 22 studies involving 1819 patients with lung cancer who received radiotherapy. The main finding of this study is that the number of patients who developed RP significantly decreased in the HM plus radiotherapy group compared with the radiotherapy alone group. In addition, the incidence rate of severe RP (scoring more than 3 points on the RTOG scale) decreased significantly in the groups that combined radiotherapy with HM when compared with radiotherapy alone. The KPS values, related to quality of life, also significantly increased in the HM plus radiotherapy group compared with the radiotherapy alone group.
Although there were few studies that reported on the safety of HMs, no severe side effects of HMs were observed in those studies. However, due to problems related to the methodological quality and reporting in most of the trials, we are not able to draw definitive conclusions, and the results must be interpreted with caution.
The methodological quality and reporting of the trials were variable, and often inadequate. Most studies did not explain the randomization process they used, and the allocation method was not mentioned in a single study. Because it is difficult to use a placebo for herbal decoctions, the evaluation of patients after administration of herbal decoctions would be more valid  Herbal medicines for radiation pneumonitis in lung cancer patients if the assessors were also blinded. However, the assessors were not blinded in any of the studies. The quality of evidence for this finding was very low because of the high risk of bias. The quality of reporting was generally poor in the included studies (Table 4). Furthermore, since the number of subjects who were randomized or dropped out from the studies was unclear, it is difficult to tell whether the studies performed per-protocol analyses or intention-to-treat analyses. While it cannot be said, due to the low quality of reporting, that the clinical trials were not conducted properly, improvement in the quality of reporting is necessary, since quality assessments are based on reports.
The chemical components of HMs can change according to the cultivation region, climate, and cultivation season. Therefore, clinical studies must provide detailed information about the HMs used. The Consolidated Standards of Reporting Trials (CONSORT) provides guidelines that list information requirements regarding HMs that must be met by clinical studies to improve the reporting quality of RCTs [43]. Recently CONSORT extensions for Chinese herbal medicine (CHM) formulas (CONSORT-CHM Formulas 2017) was developed to offer more suitable guideline for HM formulas [44]. Traditional Chinese medicine (TCM) and traditional Korean medicine (TKM) is based on its own unique principle and comprehensive theory. The CONSORT-CHM Formulas 2017 reported the key concepts of pattern and the features of CHM formulas to reflect TCM-theory. The selected studies provided poor information on the HMs that they investigated. None of the studies reported herbal medicinal product names, characteristics of the herbal products, dosage regimen and quantitative description, or qualitative testing, which should have been explained. The selected studies only reported the amount of each herb constituting the HM in grams, and the daily dosage. Future clinical studies using HMs must be conducted with more care and recommended to follow the CONSORT-CHM Formulas 2017.
Many of the studies used the RTOG scale, which is an appropriate assessment tool for RP, used internationally. The quality of life was evaluated using the KPS scale; however, strictly speaking, the KPS scale is a performance scale, not an assessment tool. The quality of life needs to be evaluated with internationally validated assessment tools, such as the EORTC-C30 [45] and EORTC-L13 [46] or fact-L questionnaires [47] developed for patients with lung cancer.
The prescriptions mainly comprised herbs that nourish yin, tonify qi and nourish blood. Ophiopogonis Radix, Astragali Radix, and Angelicae Sinensis Radix have been reported to be frequently used in traditional Chinese medicine in combination with chemotherapy for patients with non-small cell carcinoma [48]. Astragali Radix, Adenophorae Radix, Ophiopogonis Radix, Glycyrrhizae Radix et Rhizoma, and Poria are in line with the frequently used herbs as adjuvant therapy to chemotherapy in patients with non-small cell carcinoma [15]. As Herbal medicines for radiation pneumonitis in lung cancer patients shown here, the herbal formulas used with radiotherapy mainly comprised herbs that nourish yin, coupled with qi-tonifying herbs. The proportion of yin-nourishing herbs was higher than in formulas used for chemotherapy. This is because radiation therapy is regarded as a heat toxin pathogen in traditional Chinese and Korean medicine theory, so prescriptions generally focus on removing heat. Despite the limitation that the herbal formulas covered in this review were not consistent, it is promising that co-administration lowered the incidence of radiation pneumonia. Future studies could develop and verify mixed herbal medicines with the most active ingredients. This study had the following strengths. First, 11 databases were searched using a well-structured search formula, and were selected through proper processes. We tried to minimize bias in the assessment of the effects of HMs on the incidence rate of RP, which was the primary  Herbal medicines for radiation pneumonitis in lung cancer patients goal of this study, by using studies that accurately described the diagnostic criteria of RP. Second, studies from all the countries in which RCTs using HMs have been actively conducted were included, including English-speaking countries, China, and Korea. The primary limitation of this study was that the quality of reporting was low in the selected studies, making clear interpretation of the results difficult.
To the best of our knowledge, this is the first systematic review that investigates the efficacy and safety of orally administered HM in conjunction with radiotherapy for preventing radiation induced pneumonitis in patients with lung cancer. The present systemic review showed that administration of HMs during radiotherapy could prevent or minimize the risk of radiotherapy pneumonia. However, due to the poor methodological quality of the identified studies, definitive conclusion could not be drawn. Randomized controlled trials that are larger in scale and have better designs must be conducted to confirm the validity of this finding.