A Meta-Analysis of the Efficacy and Toxicity of Twice-Daily vs. Once-Daily Concurrent Chemoradiotherapy for Limited-Stage Small Cell Lung Cancer Based on Randomized Controlled Trials

Background: Currently, the accepted standard management of limited-stage small cell lung cancer (SCLC) is concurrent chemoradiotherapy (CCRT), but the frequency of radiotherapy is controversial. Therefore, this meta-analysis, which compared the efficacy and toxicity between twice-daily (BID) and once-daily (OD) CCRT, was performed to help clinicians make better decisions. Methods: Relevant randomized controlled trials (RCTs) were collected by searching the PubMed, Ovid MEDLINE, Embase, ScienceDirect, Web of Science, the Cochrane Library, Scopus and Google Scholar databases to assess antitumor effects (overall survival, OS; progression-free survival, PFS; overall response rate, ORR) and toxicity (adverse effects, AEs). Results: We screened 1499 articles and included 5 RCTs including 1421 patients. We found that BID CCRT improved OS (hazard ratio, HR = 0.88, 95% confidence interval, CI 0.78–0.99, p = 0.03), the 1-year OS rate (OSR-1y, risk ratio, RR = 1.07, 95%CI 1.01–1.13, p = 0.03), and OSR-4y (RR = 1.22, 95%CI 1.03–1.43, p = 0.02), with better trends in OSR-2y, OSR-3y, and OSR-5y, compared to OD CCRT. In addition, BID CCRT had a higher complete response (CR, RR = 1.31, 95%CI 1.01–1.70, p = 0.04) than OD CCRT. PFS (HR = 0.92, 95%CI 0.79–1.07, p = 0.29), annual PFS rate, ORR (RR = 0.99, 95%CI 0.93–1.05, p = 0.72), and AEs for all grades (RR = 1.00, 95%CI 0.98–1.01, p = 0.57), and grades 3–5 (RR = 1.02, 95%CI 0.95–1.09, p = 0.60) were similar between the two arms. Conclusions: BID CCRT appears to be better than OD CCRT for limited-stage SCLC, with better antitumor effects (OS, OSR, and CR) and similar AEs. However, the high levels of AEs in both arms should be taken as a sign of caution. More large sample and high-quality RCTs need to be conducted to confirm our conclusions.


INTRODUCTION
It is estimated that 228,150 new cases of lung cancer will be diagnosed in 2019, ∼15-30% of which will be small cell lung cancer (SCLC) (1,2). As reported by the Surveillance, Epidemiology, and End Results (SEER) Cancer Statistics Review, the annual incidence rates of SCLC are decreasing, but the 5-years period survival is only 6.5% (3). This deadly neuroendocrine tumor, which has the characteristics of a short doubling time and early metastasis, is difficult for oncologists to treat (4). For limited-stage SCLC, the current standard treatment is concurrent chemoradiotherapy (CCRT) (5). However, the optimal frequency of CCRT is still controversial.
In light of the National Comprehensive Cancer Network (NCCN) guidelines, twice-daily (BID), and once-daily (OD) CCRT are recommended, but it is not clear which treatment is better (6). Based on the Intergroup 0096 study, 45 Gy in 30 fractions BID CCRT was considered better than 45 Gy in 25 fractions OD CCRT (7). However, this result was not widely accepted because BID CCRT resulted in severe esophagitis and logistical problems (8). The latest large phase III randomized controlled trial (RCT) has reported that the survival outcomes and toxicity are similar between the two regimens (9). Meanwhile, some studies have even shown that OD CCRT improves survival or reduces toxicity compared with BID CCRT (10,11). To date, the two regimens are both in clinical use with different doses.
As a result, this meta-analysis was performed to compare the efficacy and toxicity of BID CCRT with OD CCRT for limitedstage small cell lung cancer to provide information to clinicians so that they can make better decisions.

MATERIALS AND METHODS
We performed this meta-analysis in accordance with the Preferred Reporting Items for Systematic review and Meta-Analysis (Table S1) (12).
were browsed and manually screened. The retrieval strategies used to search the electronic databases are presented in Table S2.

Selection Criteria
We set the inclusion criteria based on the principles of PICOS as follows: (1) patient: patients with limited-stage SCLC; according to a 2-stage (limited-stage and extensive-stage) classification scheme, limited-stage SCLC should include patients with a primary tumor and nodal involvement limited to the ipsilateral hemithorax or without distant metastasis (13); (2) intervention and comparison: BID CCRT vs. OD CCRT; (3) outcome: antitumor effect (overall survival, OS; progression-free survival, PFS; overall response rate, ORR), and toxicity (adverse effects, AEs); and (4) study design: RCTs published in English.
We excluded articles without original data, with abstracts only, or with duplicated data as well as reviews, meta-analyses, animal experiments, and unpublished conference papers and clinical trials.

Data Extraction
The data collected from each included study were general information; characteristics of the study participants; specific modalities of radiation and chemotherapy; survival outcomes (OS, PFS, ORR), including follow-up periods; and AEs, which included the total AEs, all grades and grade 3-5 AEs. The whole information extraction process was completed by two researchers independently, and any disagreements were solved through discussion.

Quality Assessment
We analyzed the methodological quality of the RCTs according to the Cochrane Handbook, which considers seven dimensions of risk of bias (14). Each dimension contains at least one item, which was used to determine "low risk, " "high risk, " and "risk unknown." One point was given for "low risk, " and high-quality studies had a score ≥3 points.
We rated the evidence quality for each outcome according to the GRADE (Grading of Recommendations Assessment, Development, and Evaluation) system, as follows: (1) High; (2) Moderate; (3) Low; and (4) Very low (15). The included RCTs may have been degraded by the following five factors: risk of bias, inconsistency, indirectness, imprecision, and publication bias.

Statistical Analysis
We used Review Manager 5.3 and STATA 12.0 for data analysis. For the time-to-event data, such as OS and PFS, the hazard ratio (HR) with the 95% confidence interval (CI) was used and pooled as an effective indicator (HR > 1 favors the OD arm). If the HR did not appear in the literature, we obtained it from Kaplan-Meier curves through the method described by Tierney et al. (16). In addition, for dichotomous outcomes, such as ORR and AEs, we calculated and pooled risk ratios (RR) with the 95% CI (ORR: RR > 1 favors the BID arm; AEs: RR > 1 favors the OD arm). Subgroup analyses of OS and PFS were carried out for the continent, models of CCRT regarding the radiotherapy dose (fractional and full) and timing, and chemotherapy cycle. The fixed effect model was initially used. χ 2 tests and the I 2 index were used to investigate heterogeneity among trials. When substantial heterogeneity appeared (I 2 > 50% or p < 0.1 for the χ 2 test), the random effect model was used and sensitivity analysis was conducted (14). Publication bias detection was performed for OS and PFS. P < 0.05 indicated a significant difference.

Search Results and Study Quality Assessment
A total of 1499 studies were identified from the search of the electronic databases, and six related articles based on five RCTs including 1,421 patients were eligible for inclusion (7,9,(17)(18)(19)(20). Bonner et al. (17) and Schild et al. (18) described the same study and different results. The former described AEs and the response rate, and the latter described survival outcomes. The literature selection is shown in Figure 1. All of the studies were of high quality. The detailed quality assessment of each included study is shown in Figure S1. The evidence quality of 24/46 outcomes was "High, " and the rest was "Moderate" (Table S3).

Basic Characteristics of the Studies
Of the five studies, two were conducted in America (7,17,18), one in Norway (20), one in England (19), and one with patients from multiple centers from eight countries (9). Eligible patients included in these trials were required to have a good performance status (the performance score was recorded as 0-2) and a limited-stage disease that met the above definition. The mode of CCRT was not identical in each experiment. Physical dose was converted into equivalent dose in 2 Gy/f (EQD2) linear quadratic model, with alpha/beta =10 for early reaction tissues (tumor tissue) and alpha/beta = 3 for late reaction tissues (normal tissue), according to the method provided by Fowler (21). Patients who achieved complete response (CR), which referred to the complete disappearance of the tumor in this meta-analysis, received prophylactic cranial irradiation (PCI). The specific basic characteristics of the studies are presented in Table 1.
Three studies analyzed PFS. We found that PFS was similar (HR = 1.08, 95%CI 0.93-1.25, p = 0.30, I 2 = 0%; Figure 2B) between the two arms. The annual progression-free survival rate (PFSR) within 5 years tended to favor the BID arm but without a significant difference ( Figure 3B and Figure S3).
Four studies compared the treatment-related mortality (TRM). We found that the TRM (RR = 0.96, 95%CI 0.50-1.87, p = 0.92, I 2 = 33%; Figure S4A) was similar between the two arms. The top three causes of treatment-related deaths were pulmonary effects, infection and neutropenic sepsis, and no significant difference was found between the two arms (Table S4).

Subgroup Analysis
To explore whether the survival outcomes changed, we established subgroups for OS and PFS according to continent, radiotherapy dose (fractional and full) and timing, and chemotherapy cycle. We found that BID CCRT led to better OS vs. OD CCRT in standard fractionation (1.8 or 2 Gy/f, HR = 0.87, 95%CI 0.76-0.98, p = 0.03, I 2 = 0%). The results of other subgroups tended to support BID CCRT compared with OD CCRT although no significant differences were found ( Table 4).

Sensitivity Analysis
To evaluate sensitivity and stability, we conducted a sensitivity analysis for CR with significant heterogeneity. We found that the outcome of CR was stable after omitting one study ( Figure S5).

DISCUSSION
The median survival of limited-stage SCLC has been reported to be 15-20 months, indicating that this cancer extremely malignant (22). For limited-stage SCLC that is not suitable for surgery, the accepted standard treatment is CCRT (23). However, there is much debate about whether radiotherapy should be BID or OD (24). This is the first meta-analysis to determine the optimal scheme between the two regimens; this meta-analysis includes five RCTs including 1,421 patients. The results of this metaanalysis indicated that BID CCRT improved OS, OSR-1y, and OSR-4y, with better trends in OSR-2y, OSR-3y, and OSR-5y, compared to OD CCRT. Subgroup analyses of OS in standard fractionation also supported this result. In addition, BID CCRT had a higher complete response than OD CCRT. No significant differences were found for PFS, annual PFSR, ORR, or AEs of grade 3-5 or all grades between the two arms. Other subgroup analyses of OS and PFS regarding continent, radiotherapy dose (fractional and full) and timing, and chemotherapy cycle tended to favor BID CCRT but without significant differences between the two arms. In fact, Turrisi et al. reported that the OS of BID CCRT was better than that of OD CCRT (p = 0.04), which agreed with the results of this meta-analysis (median OS: 25.6 months vs. 21.6 months) (7). The annual OSR suggested that there was a potential survival advantage in BID CCRT compared to OD CCRT. Schreiber (25). However, a cohort study in Asia found that OD CCRT had a significantly longer median OS than BID CCRT (47.2 months vs. 32.8 months) (10). The studies we included in this meta-analysis mainly focused on Europe and the Americas, and the results of this meta-analysis may only apply in these areas. Subgroup analyses of OS in standard fractionation indicated BID CCRT was superior to OD CCRT but no significant difference was found in hypofractionation (>2 Gy/f) and other subgroup analysis results, which might be caused by an insufficient sample size. Xia et al. study also shown that 2.5 Gy/f OD CCRT had favorable survival (26). The ongoing CALGB 30610/RTOG 0538 study (NCT00632853) is further exploring the effects of high dose fractions (27). According to the NCCN guidelines, which recommend the first or second cycle of chemotherapy be combined with radiotherapy, we performed a set of subgroup analyses and found that OS was improved in the BID arm when radiotherapy was initiated from the first or second cycle of chemotherapy vs. the OD arm (HR = 0.85, 95%CI 0.75-0.97, p = 0.02). One possible reason for this improvement is that repopulation of tumor cells is accelerated during chemotherapy cycles and proliferating cells may be more sensitive to BID irradiation (28). Another possible reason for the improvement is that lower overall doses lead to improved treatment delivery of BID CCRT compared with OD CCRT; thus, more patients receive full-dose irradiation. The PFS and PFSR results were consistent with the included studies, which tended to favor BID CCRT but without statistically significant differences compared with OD CCRT. The subgroup analysis reported similar results. However, a previous cohort study reported that the median PFS was longer in the OD arm vs. the BID arm (20.1 months vs. 18.8 months), although there was no significant difference. We recommend that high-quality and well-designed RCTs should take the Asian region into account.
In all of the included studies, the ORRs to CCRT of the two arms were both nearly 90 percent, and no significant difference was reported, which formed the basis of our results. Kubota et al. also found that 95 percent of limited-stage SCLC patients receiving BID CCRT combined with etoposide and cisplatin had an objective response in a large sample RCT (29). In general, for treatment of a tumor with poor prognosis, the CR and PR results should support the same arm (30). However, in this metaanalysis, the higher CR and lower PR in the BID regime only demonstrated an advantage due to the high ORR. These results are in line with the RHA study and NCCTG study but not the ECOG study, in which the difference between the two arms was not significant. Salama et al. found low CR (41%) and high PR (47%) in connected Cancer and Leukemia Group B limited-stage SCLC trials (39808, 30002, and 30206) using high dose OD CCRT, which was similar to our results (OD: 44% and 51%) (31). It was possible that BID CCRT provided a higher biological effective dose for a shorter time of radiation compared with the OD arm, thus preventing further tumor replication.
Regarding AEs, although no significant difference was reported between the two arms in most of the included studies, the NCCTG study found that BID CCRT had a higher incidence of grade 3-5 AEs (54 vs. 39%) and thrombocytopenia (45 vs. 59%) and the ECOG study found that BID CCRT had a higher incidence of esophagitis (63 vs. 44%) vs. OD CCRT. Thus, we conducted a subgroup analysis of grade 3-5 esophagitis and found similar results in the subgroup from North America, where the included studies were published in 1999 (RR = 2.06, 95%CI 1.46-2.89, p < 0.0001; Figure S4B). However, a recent study reported that the incidence of severe esophagitis after BID CCRT was only 2% (29). A possible reason for this result is that previous studies used large and 2-dimensional fields with elective nodal irradiation, which increased radiation exposure to normal tissues (9). At present, 3D-or intensity-modulated radiotherapy provides fewer, higher dose fractions and achieves similar results as the hyper-fractionated regimen (32). We suggest that future studies should use advanced radiotherapy techniques to determine the optimal dose and fractionation for limited-stage SCLC therapy.
There were several limitations of this meta-analysis: (1) Only five RCTs were included, which impacted the representativeness of the results; (2) substantial heterogeneity appeared in the analysis which might make the results unstable: most of the participants were from Europe and North America, which affected the representativeness of the results; only the CONVERT study reported similar improvements in OS in elderly patients for both regimes, which might not allow the elimination of the interference of age (33); PCI might be unbalanced between the two arms (RR = 1.07, 95%CI 1.00-1.14, p = 0.06, I 2 = 0%; Figure S4C), which had been reported to improve OS (34); at baseline, all patients received CT scans, and only some patients from CONVERT received PET/CT imaging. In fact, pretreatment with PET scans had been reported to improve OS (35); only the RHA study had a definite staging on CT scans before admission, and the rest were performed later, which might lead to stage migration; (3) some of the time-to-event data were obtained from Kaplan-Meier curves, which were unstable; factors such as necrosis and hypoxia could affect the alpha/beta value, which made the inaccuracy of the formula to estimate EQD2 increase; and (4) there is an economic argument for BID CCRT as compared to OD CCRT because of the shortened treatment time, but no formal economic analysis had been done on both regimens. These findings suggest that further well-designed prospective studies need to be carried out.
CONCLUSION BID CCRT appears to be better than OD CCRT, with better antitumor effects (OS, OSR, and CR) and similar AEs for limitedstage SCLC patients. However, the high levels of AEs in both regimens should be a sign of caution. Subgroup analysis of OS and PFS tended to favor BID CCRT, especially comparing OD CCRT in standard fractionation. Moreover, more large sample and well-designed RCTs need to be conducted to confirm our conclusions.

DATA AVAILABILITY STATEMENT
All datasets generated for this study are included in the article/Supplementary Material.

AUTHOR CONTRIBUTIONS
QW: full access to all of the data in the manuscript and takes responsibility for the integrity of the data, the accuracy of the data analysis, and statistical analysis. QW and WZ: drafting of the manuscript. QW, WZ, YX, SZ, XC, and FY: critical revision of the manuscript for important intellectual content. YW and WZ: supervision. All authors: concept and design and acquisition, analysis, or interpretation of data.

FUNDING
This study was supported by National Natural Science Foundation of China (NSFC), number of grants (81560345), Natural Science Foundation of Jiangxi Province (Grant number: 20181BAB215027). The funding had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

ACKNOWLEDGMENTS
The authors thank professor Bentong Yu, MD (The first affiliated hospital of Nanchang University) for her advice and professor Xiaoshu Cheng, MD, Ph.D. (The second affiliated hospital of Nanchang University) for his data collection.