Colorectal cancer has now become the fourth common cancer in China and the fifth leading cause of cancer death [1], with many patients diagnosed at advanced stages, bringing a heavy social and economic burden. Until the 1980s, patients with locally advanced rectal cancer (cT3–4 and/or cN+) had high incidences of both local recurrence and distant metastasis. To solve this problem, total mesorectal excision (TME) instead of conventional surgery was developed, which improved outcomes in locoregional control and survival in rectal cancer [2, 3]. Meanwhile, neoadjuvant chemoradiation (neoCRT) was also widely investigated. It was demonstrated that neoCRT significantly reduced pelvic recurrences and increased sphincter-sparing surgery rates, but had no clear improvement on survival [4, 5]. According to the National Comprehensive Cancer Network (NCCN) guidelines [6], neoCRT followed by TME is now recommended as the standard of care for locally advanced rectal cancer.

While locoregional recurrence rates have been reduced to only 4%–8%, 5-year disease-free survival rates in locally advanced rectal cancer remain low at 59%–77% [4, 5, 7], indicating insufficient control over systemic failure. In prevalent treatment schedules, systemic chemotherapy is put in the adjuvant setting after surgery, which means that patients have to receive systemic therapy in about 3–4 months after the diagnosis of cancer; this may account for systemic failure, especially for patients who have already had micrometastases in the initial diagnosis. To address this problem, more effective systemic treatment regimens are needed.

Recently, a series of studies demonstrated that patients with pathological complete response (pCR) after chemoradiotherapy achieved a good prognosis compared with those who had pathological residual disease, in terms of disease-free survival (DFS) and overall survival (OS) [8, 9]. Therefore, some clinical trials were designed to increase pCR by adding new anti-cancer drugs to neoCRT. Oxaliplatin plus capecitabine (the XELOX regimen) has been recommended as an option of induction therapy [10]. The addition of molecular-targeted therapy to increase the rate of responders and to address micrometastases has also been discussed [11].

Bevacizumab, a monoclonal antibody that blocks vascular endothelial growth factor (VEGF), is a critical mediator of tumor angiogenesis [12]. It has been demonstrated that bevacizumab significantly improves OS in patients with metastatic colorectal cancer [13, 14]. Moreover, anti-VEGF antibody has been reported to compensate for the resistance to radiation and augment tumor response in preclinical models [15, 16]. However, the use of bevacizumab as neoadjuvant treatment for locally advanced rectal cancer remains unclear. Several phase II studies have been conducted using bevacizumab in combination with chemoradiotherapy, demonstrating its feasibility with acceptable toxicity [17–19], but none were conducted in Asian populations.

In our institution, we began the use of XELOX as the concurrent chemotherapy regimen in 2007. In our preliminary studies, we demonstrated that induction chemotherapy followed by chemoradiotherapy with the XELOX regimen was well tolerated and brought about a satisfactory pCR in high-risk locally advanced rectal cancer [20–22]. Here we designed a single center pilot study to determine whether adding bevacizumab to neoadjuvant treatment with oxaliplatin and capecitabine plus radiotherapy was a safe and effective strategy for clinical stages II to III rectal cancer.

In this prospective phase II trial, we demonstrated that this multimodal neoadjuvant treatment for locally advanced rectal cancer, consisting of induction chemotherapy and concurrent chemoradiotherapy with capecitabine, oxaliplatin and bevacizumab, produced a satisfactory pCR rate and 3-year survival rate while slightly increasing adverse events and perioperative complications.

The most serious post-operative complication that surgeons were concerned about was anastomotic leak. The incidence of anastomotic leak that needed surgical intervention was 12.4%, slightly higher than studies without bevacizumab [28] and lower than other studies with bevacizumab [18, 29]. As a matter of fact, we did observe an increase in the incidence of anastomotic leak in the early stage of the study, so we suggested prophylactic ostomy as a standard procedure in the later stage. To our delight, the incidence of anastomotic leak was reduced after that and no further surgical intervention was needed. Although the importance of prophylactic ileostomy or colostomy on decreasing anastomotic leak has been widely verified in western countries [30], many patients and some surgeons in China do not seem to have accepted this idea. As a result, the beginning of this trial witnessed the creation of a prophylactic ostomy that was not mandatory but was determined by the surgeon and the patient instead. In fact, one anastomotic leak case that required surgical intervention was preventable, indicating that this patient did not refuse prophylactic ostomy. According to one study [31] in our hospital in 2015 that compared surgery with and without preoperative concurrent chemoradiotherapy for mid and low rectal cancer, the incidence of anastomotic leak in the concurrent chemoradiotherapy plus TME surgery group and the TME group were 2.2% and 8.5%, respectively (P = 0.101). Comparison of the incidence rate of leaks in concurrent chemoradiotherapy with the current study (2/90 vs. 2/13, P = 0.077) revealed a trend towards increased anastomotic leak, although no statistical significance was observed. According to the NCCN guidelines, at least a 6-week interval between the last dose of bevacizumab and surgery was recommended for the risk of healing complications. In our study, patients who developed anastomotic leak all had an interval longer than 10 weeks between the last dose of bevacizumab and surgery. Whether the anastomotic leak is related to bevacizumab remained unclear. Other toxicities were all acceptable. Hand-foot syndrome and oxaliplatin-related neurotoxicity were all mild, and no subsequent treatment was affected. Bevacizumab-related side effects such as hypertension and proteinuria were relatively few. Blood loss and operative time were comparable to patients who did not receive bevacizumab [31]. The incidence of adverse events in the present study was lower than in the literature. This may partly be explained by racial differences, because the acute adverse events that occurred during concurrent chemoradiotherapy in previous studies were also lower in Asian patients than in other ethnic backgrounds [20–22]. Moreover, as a study with small sample size, sample bias might also exist.

Bevacizumab was among the first molecular-targeted agents to be introduced to treat locally advanced rectal cancer when we began the present study in 2013. In preclinical models, bevacizumab has been found to lead to vascular normalization, reduce tumor hypoxia, improve radiosensitivity [32], and augment tumor response [15, 16]. All these effects have made bevacizumab a promising drug whose concurrent application with radiotherapy is worthy of being studied. Studies have also shown that adding bevacizumab to regimens for metastatic colorectal cancer has improved survival [13, 14]. However, in colon cancer stages II and III, the addition of bevacizumab to the systemic treatment has not shown any benefit. Therefore, whether bevacizumab has any advantage in treating micrometastases in locally advanced rectal cancer remains unclear. For better local and systemic control, researchers have attempted to integrate bevacizumab in the multimodal management of locally advanced rectal cancer patients, and several phase II studies have been conducted to test its safety and efficacy. The strategies developed by most studies integrated bevacizumab as a companion to conventional fluoropyrimidine-based [33–35] or intensified oxaliplatin-containing chemoradiotherapy [36, 37], or as an induction treatment before chemoradiotherapy [18].

Oxaliplatin, as the most promising drug in previous single-arm studies, did not improve surgical outcomes such as pCR rate, but increased toxicity in the Italian STAR-01 study [38] as well as the US NSABP R-04 trial [39]. However, a meta-analysis shows that the addition of oxaliplatin to fluoropyrimidine in neoadjuvant chemoradiotherapy could modestly increase the pCR rate and reduce the rate of intra-abdominal or peri-operative metastases, and that the addition of fluoropyrimidine did not result in more surgical complications or postoperative deaths within 60 days despite increased toxicity [40]. Induction of XELOX has also achieved more favorable compliance with less toxicity [10]. In our previous study, we also found that neoadjuvant sandwich treatment [20] with the XELOX regimen added to the conventional radiation is well tolerated. Such discrepancies might be attributed to differences in race as well as drug delivery. This finding was confirmed by our current study, in which we used bevacizumab plus the XELOX regimen as an induction treatment before radiotherapy and concurrent treatment during radiotherapy. Differing from the NSABP R-04 and STAR-01 trials in which oxaliplatin was given weekly during radiation therapy, the current study adopted a strategy of 3 weeks on the XELOX regimen with oxaliplatin given at one loading dose. This difference in the use of oxaliplatin may be beneficial for systemic control.

An intensified preoperative treatment might enhance downsizing and downstaging of the primary tumor, and thus increase the likelihood of achieving pCR. A trend to higher pCR rates was noted when more systemic agents were added to standard chemoradiation [41]. Previous phase III trials of capecitabine or 5-FU-based chemoradiation reported pCR rates being 11%–15% [5, 42]. When bevacizumab was added to conventional fluoropyrimidine-based chemoradiotherapy, pCR rates were increased to 7.5%–32% [33–35], and the addition of bevacizumab to intensified oxaliplatin-containing chemoradiotherapy further increased pCR rates to around 18–40% in several phase II studies [37, 41, 43]. Meanwhile, induction chemotherapy in combination with bevacizumab may also contribute to benefit the pCR rate. In the present study, bevacizumab was employed in both induction and concurrent treatments, and the current cohort showed tumor downstaging in most patients with an ypT0 rate of 42.1%, pCR rate of 39.5%, and near-pCR rate of 21.0%, which were almost the best results of previous phase II studies.

In the present study, a discrepancy between pCR and cCR was also observed. Only four patients (8.9%) achieved CR according to radiology, but the number of pCR cases turned out to be 15 upon pathological examination. This may be related to meticulous clinical practice, in which any sign of residual mass on MRI was considered to be a possible residual tumor. A previous study [44] reported that cCR was associated with pCR in approximately 30% of patients; and this is in accordance with our results.

Survival results were also promising in the present study, with a 3-year PFS rate of 88.6% and a 3-year OS rate of 95.3%, indicating effective control over both locoregional and distant diseases. Three explanations may be applicable. First, this intensified neoCRT treatment strategy further reduced local relapse, thus contributing to survival. Second, the 3-week XELOX regimen plus bevacizumab eradicated micrometastases, resulting in good distant control. Finally, racial differences might exist in patients’ compliance and clinical response. A previous study reported that 3-year disease-free survival could be used as a surrogate parameter for long-term survival in colorectal cancer [45], thus we have reason to believe that patients in this group will have satisfactory long-term survival.

Our study has some potential limitations that require special consideration. Its major weakness is that we did not make prophylactic ostomy a standard of care in the primary design, and the incidence of anastomotic leak was relatively high in the first stage. In the later stage, however, the incidence was reduced when prophylactic stoma was performed. Furthermore, although the pCR rate could be used as a surrogate endpoint for efficacy, longer follow-up is needed to evaluate the impact of bevacizumab on more important endpoints, such as DFS and OS.

Preliminary results suggest that the combination regimen of bevacizumab and XELOX administered in neoadjuvant therapy with radiotherapy may increase the risk of anastomotic leak, and therefore was not considered as a regular recommendation for locally advanced rectal cancer. However, creating a prophylactic ostomy might help to reduce the incident of this adverse event. Current data presented as pCR rate and 3-year survival were satisfactory, indicating promising long-term survival. For patients at high risk with locally advanced rectal cancer, XELOX in combination with bevacizumab may be a potential treatment option.