A randomized phase II study of capecitabine-based chemoradiation with or without bevacizumab in resectable locally advanced rectal cancer: clinical and biological features

Surgery is the mainstay of curative therapy for patients with rectal cancer confined to the bowel and regional lymph nodes. Nevertheless, rectal cancers have a high incidence of local failure. Recurrent pelvic disease is associated with significant morbidity and substantially shorter survival [1]. Perioperatory chemoradiotherapy (CRT) improves local control and survival in patients with locally advanced (T3-T4) rectal cancer (LARC) [2]. Moreover, this strategy maximizes downstaging, increases the rate of sphincter-sparing surgery and provides early exposure to systemic therapy.

Capecitabine is being integrated into the treatment of patients with colorectal cancer as an alternative to 5-fluorouracil (5-FU), resulting in improved convenience without compromising efficacy [3]. In addition, radiation induces thymidine phosphorylase and enhances the efficacy of capecitabine, leading to a synergistic effect [4]. A recent non-inferiority phase III study has shown capecitabine can replace 5-FU in adjuvant or neoadjuvant CRT regimens for patients with LARC [5].

In rectal cancer, several trials of bevacizumab with chemoradiation have shown promising results [6-11], but the lack of randomization and the bias associated with single-arm trials raises important concerns when interpreting these data. No randomized study to date has tested the use of bevacizumab in the neoadjuvant setting for localized disease.

Based on our previous experience (back to back submission [12]) and reports of preliminary feasibility of bevacizumab and chemoradiation in rectal cancer, we conducted a randomized trial of neoadjuvant bevacizumab and chemoradiation in patients with resectable LARC. Also, as little data exists on the role of biomarkers as predictors of response to bevacizumab when adding to preoperative CRT in LARC, we explored potential biomarkers that have been previously found to change in response to bevacizumab in other translational trials [11,13-20].

The following possible prognostic factors for tumor angiogenesis were evaluated: vascular endothelial growth factor (VEGF) and circulating soluble VEGF receptor 2 (VEGFR-2) which can be expressed in malignant tumors [21], angiopoietin-2 (Ang-2) expression, a molecule which promotes destabilization of blood vessels, and whose expression decreases along with microvessel density (MVD) after bevacizumab administration supporting the theory on the normalization of the vessels postulated for that drug [18], and the intratumor MVD since an increase in the number of tumor vessels might constitute a higher risk to develop metastasis [22-25].

Between December 2009 and March 2011, 90 patients were randomly assigned through 12 Spanish hospitals, 44 in arm A and 46 in arm B (Figure 1).

Baseline patient characteristics were well balanced between groups (Table 1). The median distance from anal verge was 6.5 cm in arm A and 7.0 cm in arm B.

This study confirms the feasibility of preoperative CRT with bevacizumab and capecitabine in patients with LARC, in a randomized trial. Although the patients who received bevacizumab tended to have a higher pCR rate, the predefined efficacy endpoint was not met. Interesting, Ang-2 plasma levels significantly decreased along the study in patients receiving bevacizumab and, overall, decrease in Ang-2 levels was significantly associated with tumor downstaging: those findings suggest improved tumor shrinkage related to the use of bevacizumab and a potential role of plasma Ang-2 to monitor downstaging.

We conducted this study to further understand the role of bevacizumab in the treatment of LACR. The 5 mg/kg dose was chosen based on the toxicity observed in the Willet et al. study [18], and the Xeberecto Trial (back to back submission [12]), a phase II study of preoperative bevacizumab, capecitabine and radiotherapy for resectable LARC. Our trial was initially designed to include oxaliplatin in both arms; however because the results of two large randomized studies [38,39] did not demonstrate the benefit of oxaliplatin with concurrent irradiation, the study was amended to do not use this drug.

Our patients had a relatively high risk for pelvic recurrence (98% were T3 or T4, and 87% were N+). The reached pCR rate of 16%, albeit not too high, is within the range (13-36%) reported across a number of phase II studies evaluating bevacizumab plus CRT (Table 5) [6-11]. However, caution is needed when comparing pCR rates as this item itself is highly dependent on the quality of the pathological examination [40].

The decrease in the T stage was significantly higher in patients receiving bevacizumab, although the rate of downstaging in both arms was lower than in other single arm studies using preoperative bevacizumab plus CRT [6,10,11]. This should be interpreted with caution because of the preoperative MRI staging technical limitation and lack of central imaging evaluation.

Toxicities were expected and manageable. Early toxicity was mild in both arms. The most serious post-operative complication was anastomotic dehiscence, which occurred more frequently in patients treated with bevacizumab, but in the range reported by others [6,7,9-11].

Suitable biomarkers predicting patients who are likely to benefit from bevacizumab treatment remain elusive [41]. In our exploratory analyses decrease in Ang-2 levels from baseline to d57 was significantly associated with tumor downstaging. Moreover, Ang-2 levels decreased in the bevacizumab arm and increased in the other arm at all time points compared to baseline levels, with significant differences between levels in both groups at all time points. Angiopoietin-2 has been proposed as a gatekeeper of VEGF function and vascular remodeling [42,43], and has been shown to promote metastatic growth [44]. Goede et al. [14] found that serum levels of Ang-2 in patients with metastatic colorectal cancer were significantly higher than in healthy individuals: moreover, in that study, compared with high serum Ang-2 levels, low serum Ang-2 was associated with an outstanding response rate, better disease control and excellent overall survival (OS). Although one can only speculate about this relationship, it seems plausible that adding bevacizumab to standard CRT induces a decrease in serum Ang-2 levels that could facilitate tumor regression.

On the contrary, we found that serum VEGF levels increased in arm A and decreased in arm B at all time points compared to baseline levels, with significant intergroup differences at all time points: this finding had not influence on downstaging. Similarly, several studies have shown acute increases in circulating VEGF after the start of bevacizumab [11]. Nevertheless, changes in VEGF concentrations associated with bevacizumab treatment have not necessarily been predictive of benefit [41]. Detailed analyses are needed of total and free VEGF levels during treatment before circulating VEGF is dismissed as a biomarker.

In accordance with previous reports [11,12,14,45], pretreatment level of VEGF and tumor MVD was not correlated to clinical end points.

This trial has several limitations. Firstly, the selection of patients for the biologic sub-study was opportunistic, by including those patients who gave their consent and in which assessment of selected biomarkers was available at different points, although there is no reason to suspect any differences with other patients where those determinations were not possible. Secondly, the small size of each cohort in that study makes it difficult to find potential associations between changes in different biomarkers, and comparisons between both arms: nevertheless, our findings point in the same direction as that described by other authors. Thirdly, TME was not possible in around 20% of patients, due to the tumor localization: as stated previously [5], TME was mandatory for tumors in the lower two-thirds of the rectum, with PME being permitted for those in the upper third, provided a distal margin of at least 5 cm without coning was observed. Finally, as in other studies realized in patients with LARC, 4 patients were found having distant metastases at the time of surgery, a number not very different of those published by others [39]. As we didn’t make a reevaluation of the baseline studies, we cannot discharge a possible mistake in the inclusion of any patient.

The results of this randomized study support the data described previously in single arm studies about the feasibility of the addition of bevacizumab to a standard neoadjuvant capecitabine-based CRT regimen, as well as its potential role in downstaging. It will be also important to continue observation of these patients to elucidate long-term outcome and morbidity of this strategy. Furthermore, although definitive judgment on the role of Ang-2 as a specific biomarker of outcome to bevacizumab in LARC will require further analysis of larger numbers of patients from phase III trials, the results arising from this study should encourage researchers to further investigate the value of Ang-2 amongst others as a potential biomarker to monitor the added value of bevacizumab in clinically relevant endpoints.