The Addition of EGFR Inhibitors in Neoadjuvant Therapy for KRAS-Wild Type Locally Advanced Rectal Cancer Patients: A Systematic Review and Meta-Analysis

Background Patients with locally advanced rectal cancer (LARC) are at higher risk of local and distant recurrence and are thus more vulnerable to metastatic diseases. Neoadjuvant chemoradiotherapy (nCRT) and subsequent curative resection with total mesorectal excision (TME) followed by adjuvant chemotherapy have been recommended by the National Comprehensive Cancer Network (NCCN) guidelines as standard of care for LARC patients. However, the efficacy of the addition of epidermal growth factor receptor (EGFR) inhibitors in kirsten rat sarcoma viral oncogene (KRAS)-wild type LARC patients remains uncertain. Materials PubMed, Embase, and Web of Science were searched to retrieve records on the application of EGFR inhibitors in a neoadjuvant setting for LARC patients. pCR was used as surrogate endpoint to perform data synthesis in a single-arm setting. Results Ten cohorts covering 540 subjects were eligible in this systematic review. The pooled pCR rate for EGFR inhibitors was 15% (95% confidence interval (95% CI), 11–20%; I2 = 55.2%); the pooled estimates of Grade 3/4 diarrhea, Grade 3/4 hand–foot syndrome, Grade 3/4 acneiform rash were 17% (95% CI, 4–34%; I2 = 93.3%), 2% (95% CI, 0–5%; I2 = 13.7%), and 15% (95% CI, 9–22%; I2 = 65.4%), respectively. Conclusion The addition of EGFR inhibitors in the nCRT for KRAS-wild type LARC patients provides comparable efficacy and acceptable safety. However, the results should be interpreted cautiously due to the small amount of relevant data and need further confirmation by more future studies.


INTRODUCTION
Rectal cancer has been recognized as one of the most prevalent and lethal cancers for decades (Torre et al., 2015). Patients with locally advanced rectal cancer (LARC) are at higher risk of local and distant recurrence and are thus more likely to suffer metastatic diseases (Schrag et al., 2014). To improve the local control and prognosis of LARC patients through tumor downstaging, the concept of neoadjuvant therapy, or performing preoperative chemoradiotherapy (CRT), has been introduced into clinical practice for years (Sauer et al., 2004;Bosset et al., 2006). According to the National Comprehensive Cancer Network (NCCN) guideline, neoadjuvant CRT (nCRT) and subsequent curative resection with total mesorectal excision (TME) followed by adjuvant chemotherapy have been recommended as standard of care for LARC patients (Rectal cancer V.2., 2020). However, considering the insufficient pathologic complete response (pCR) rates reported in clinical trials investigating various strategies in neoadjuvant setting, LARC patients are badly in need of nCRT regimens with better efficacy (Bosset et al., 2005;Gerard et al., 2006;Craven et al., 2007).
The application of targeted agents to improve the efficacy of nCRT in LARC patients has been a focus of research in the past decade. Our previous work (Zhong et al., 2018) comprehensively reviewed and evaluated the efficacy of the addition of antivascular endothelial growth factor (VEGF) or antiepidermal growth factor receptor (EGFR) targeted agents in meta-analyses of single-arm setting. We reported a promising pooled pCR rate of 27% in bevacizumab-relevant cohorts, but only that of 14% in cetuximabrelevant cohorts. Of note, the kirsten rat sarcoma viral oncogene (KRAS) mutation status of the cetuximab-relevant cohorts is not restricted to be wild-type due to the limitation of included studies, thus the insufficient pooled pCR rate may be accounted for by the potential KRAS-mutated subjects in these cohorts to some extent. In this regard, we performed this systematic review and meta-analysis to assess the addition of EGFR inhibitors in neoadjuvant therapy for KRAS-wild type LARC patients.

Study Selection
This systematic review was conducted according to the Preferred Reporting Items for Systematic Reviews and Meta-analysis (PRISMA) statements checklist (Moher et al., 2009).
The predefined criteria for inclusion and exclusion were: (1) Subjects diagnosed with LARC (clinical T stages 3-4 and/or lymph node metastasis, no distant metastatic diseases observed). (2) KRAS wild-type. (3) Administration of EGFR inhibitors in neoadjuvant setting. (4) Reported pCR and Grades 3-4 treatment-related toxicities such as diarrhea, hand-foot syndrome, and acneiform rash. (5) More recent and larger studies were chosen if research cohorts presented overlapping. (6) Original researches only, excluding reviews, systematic reviews, case reports, case series, letter to the editor.

Data Extraction
The endpoints of interest were pCR and the rates of patients who suffered Grade 3/4 treatment-related toxicities namely diarrhea, hand-foot syndrome, and acneiform rash. The following data concerning baseline characteristics of the eligible studies were extracted: first author and publication year, study design, country/district, subjects, strategy of neoadjuvant therapy, median age, and staging at enrollment. All data were independently extracted by two authors (Yue Zhou and Xi Zhong). Discrepancies were resolved through consensus. The methodological quality of the included studies was evaluated using the Newcastle-Ottawa quality assessment scale (NOS) (Stang, 2010). Studies which scored five or more were deemed of moderate-quality, whereas those with seven or more were considered high-quality (Stang, 2010).

Statistical Analysis
Data of pCR and Grade 3/4 toxicity were within the range of 0-0.3, thus were firstly double arcsine transformed and then pooled using a random-effect model to provide more conservative estimates (Freeman and Tukey, 1950;Liebig et al., 2009). The Cochrane's Q test and inconsistent index (I 2 ) were performed to detect the presence of heterogeneity (Higgins et al., 2003). To detect potential origins of heterogeneity, subgroup analyses were performed based on the type of EGFR inhibitor, the intensity of backbone nCRT, and the region where each included study was conducted, which were selected following the PICOS (Participant, Intervention, Comparison, Outcome, and Study design) principle. A subgroup analysis based on study design was planned; however, considering that only one of the ten included studies was a retrospective study while the other nine were all prospective phase II studies, this subgroup analysis was not performed. Small study effects were evaluated using the Egger linear regression test when data was sufficient (≥10) (Sterne and Egger, 2001). All statistical analyses were performed using STATA version 12.0 (STATA, College Station, TX).

Study Inclusion and the Baseline Characteristics of Eligible Studies
3,494 records were retrieved after database search and manual screening of references; 1,686 duplicates were removed leaving 1,808 records to proceed on title and abstract screening. 20 potential candidates underwent full-text review, of which 10 (Bengala et al., 2009;Pinto et al., 2011;Dewdney et al., 2012;Sun et al., 2012;Helbling et al., 2013;Liang et al., 2017;Hasegawa et al., 2017;Merx et al., 2017;Leichman et al., 2018;Pinto et al., 2018)were adequate for eligibility, after excluding 10 studies which were deemed inadequate because they did not focus on KRAS-wild type LARC patients (shown in Figure 1). Among the eligible studies, six (Bengala et al., 2009;Dewdney et al., 2012;Sun et al., 2012;Liang et al., 2017;Hasegawa et al., 2017;Leichman et al., 2018) focus on the efficacy of cetuximab while the other four (Pinto et al., 2011;Helbling et al., 2013;Merx et al., 2017;Pinto et al., 2018) panitumumab. Concerning the country/ district where these trials were conducted, three (Bengala et al., 2009;Pinto et al., 2011;Pinto et al., 2018) were in Italy, one (Merx et al., 2017) in Germany, one (Helbling et al., 2013) in Switzerland & Hungary, one (Dewdney et al., 2012) in the United Kingdom, Spain & Sweden, one (Leichman et al., 2018) in the United States of America, one (Sun et al., 2012) in the mainland of China, one (Liang et al., 2017) in Taiwan, and one (Hasegawa et al., 2017) in Japan. The detailed baseline characteristics and data of the endpoints of interest of the included studies are shown in Table 1. The results of the methodological quality assessment of the eligible studies are shown in Table 2. Among the ten studies, two with seven points were deemed high-quality, while the remaining eight scored six points and were considered as moderate-quality studies.

Main Findings and Interpretation in Light of the Evidence
KRAS mutation was firstly demonstrated as predictive for lack of response in 2008 (Amado et al., 2008;Karapetis et al., 2008); the studies investigating the roles of EGFR inhibitors in the nCRT for KRAS-wild type LARC patients arose ever since. However, these studies are mostly signal-seeking single-arm phase II trials using pCR, a well-established surrogate endpoint for survival outcomes, as primary endpoint, largely lacking head-to-head survival data comparing neoadjuvant regimens with or without anti-EGFR targeted agents (Bengala et al., 2009;Pinto et al., 2011;Helbling et al., 2013;Zhong et al., 2018). In 2014, an important phase II randomized controlled trial (RCT) (EXPERT-C) by Dewdney et al. (2012). reported a significant improvement in overall survival for KRAS wild-type LARC patients receiving neoadjuvant XELOX and cetuximab (hazard ratio, 0.27; 95% CI, 0.07-0.99; P = 0.034). However, the primary endpoint, pCR, was only 11% in the cetuximab arm compared with 7% in the control arm. In another RCT (SAKK 41/07), a pCR of 10% was reached in KRAS wild-type LARC patients treated with panitumumab and capecitabine compared with 18% in those treated with capecitabine. Of note, the cetuximab/panitumumab arms in these RCTs were included in this systematic review and metaanalysis, while the other eight included studies are either singlearm phase II clinical trials (Bengala et al., 2009;Pinto et al., 2011;Dewdney et al., 2012;Sun et al., 2012;Hasegawa et al., 2017;Merx et al., 2017;Leichman et al., 2018;Pinto et al., 2018) (7) or     single arm retrospective study (Liang et al., 2017) (1). On account of the single arm nature of the ten included cohorts and the widely-acknowledged prognostic role of pCR, we decided to conduct a single arm meta-analysis using pCR as primary endpoint to evaluate the efficacy of neoadjuvant EGFR inhibitors in LARC patients. Besides, to provide a benchmark pCR rate, in our previous work (Zhong et al., 2018), we extracted and pooled relevant data from ten LARC cohorts extracted from the individual patient data-leveled meta-analysis of Maas et al (Maas et al., 2010. In this work, we transformed these data using double arcsine and synthesized a pooled pCR rate of 15% (95% CI, 13-17%), as shown in Figure S1, compared with that of 17% (95% CI, 15-20%) in the previous work where we did not use double arcsine transformation. The baseline characteristics of included cohorts are presented in Supplementary Table S1.
To evaluate the safety of additional EGFR inhibitors, we synthesized the risk of Grade 3/4 treatment-related toxicities namely Grade 3/4 diarrhea (17%, 95% CI: 4-34%), Grade 3/4 hand-foot syndrome (2%, 95% CI: 0-5%), and Grade 3/4 acneiform rash (15%, 95% CI: 9-22%). This safety is acceptable considering that previously published clinical trials reported overall risks of Grade 3/4 toxicity ranging from 13.9% to 27% (Sauer et al., 2004;Bosset et al., 2006;Gerard et al., 2006;Rodel et al., 2012). However, the intensity of backbone nCRT regimens of the included cohorts in safety evaluation is weaker than that of these previous studies, thus more future studies on  REC, representativeness of the exposed cohort; SNEC, selection of the nonexposed cohort; AE, ascertainment of exposure; DO, demonstration that outcome of interest was not present at start of study; SC, study controls for age, sex; AF, study controls for any additional factors; AO, assessment of outcome; FU, follow-up long enough for outcomes to occur; AFU, adequacy of follow-up of cohorts (≥90%). "1" means that the study satisfied the item and "0" means the opposite situation.

Strengths and Limitations
This is the most comprehensive systematic review to evaluate the efficacy and safety of the addition of EGFR inhibitors in the nCRT for KRAS-wild type LARC patients in a single-arm setting, and we used double arcsine transformation to process data to better cope with the feature of all these data lying within the range of 0 to 0.3. Nonetheless, there are several limitations that have to be addressed. First, lack of head-to-head survival data weakens our analyses; pCR is only a surrogate endpoint for prognosis. Second, more future studies evaluating the addition of EGFR inhibitors in neoadjuvant therapy for KRAS-wild type LARC patients are needed to provide more stable and more robust results. Third, due to the lack of currently available studies, this study focused solely on KRAS status; however, increasing evidences demonstrate that the response to anti-EGFR treatment can be also strictly correlated with the mutational statuses of NRAS and BRAF (Sorich et al., 2015;Sanz-Garcia et al., 2017). Future studies stressing more on NRAS and BRAF statuses are warranted. Fourth, although the significant heterogeneity appears to originate from RT-relevant cohorts, only two studies were included in this subgroup. The intensity of backbone nCRT requires to be further accounted for as the amount of eligible trials increases in the future. The preferred backbone nCRT is 5-FU/capecitabine with concurrent RT, considering that the addition of oxaliplatin is now not recommended in neoadjuvant setting for LARC patients, according to the latest version of NCCN guidelines (Rectal cancer V.2., 2020).

CONCLUSION
The addition of EGFR inhibitors in the nCRT for KRAS-wild type LARC patients provides comparable efficacy and acceptable safety. However, the results should be interpreted cautiously due to the small amount of relevant data and need further confirmation by more future studies.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation, to any qualified researcher.