Bevacizumab and Surgery: What is the Real Risk?

There are two contrasting theories to explain the mechanism of action of bevacizumab, a monoclonal antibody directed against isoforms of the VEGF receptor (VEGF-R). The best known of these concerns its anti-angiogenic effects, and it is now nearly 40 years since Judah Folkman stated that beyond a certain size, tumors require their own blood supply for growth [1–3]. Since then, almost every anticancer drug has been associated in some way with ‘anti-angiogenesis’, leading to this being called the ‘Holy Grail’ of oncology [4]. A second theory, which is less commonly discussed but of immediate relevance here, suggests that bevacizumab leads to a ‘straightening out’ of the tortuous and kinked blood vessels within a tumor mass, leading to enhanced delivery of chemotherapy. This goes some way to explain why bevacizumab has poor monotherapy response rates but, in combination with chemotherapy, it significantly improves response rates, time to progression and overall survival, compared with chemotherapy alone in patients with metastatic colorectal cancer [5,6]. In addition, bevazicumab is free of routine chemotherapeutic toxicities such as myelosuppression, alopecia and mucositis.

The significant activity of these regimens (and others that contain cetuximab [7]) in patients with colorectal cancer has led to an increased number of patients being considered as candidates for surgical resection of secondary lesions in the liver, along with primary resections of the colonic lesions, which are often, but not always, performed during the same operation. Hepatic resections may occur in stages, along with portal vein embolization and the use of other modalities such as radiofrequency ablation. While liver metastases, present in 20% of patients at presentation, remain the main cause of mortality [8], primary colonic lesions remain liable to obstruct, perforate and bleed, and thus require specific management themselves. It is therefore surprising that the major studies of hepatic resection have not adequately discussed the ‘fate’ of the primary lesions.

The appropriate selection of patients with inoperable disease for ‘neoadjuvant therapy’, with a view to disease ‘down-staging’ and subsequent resection with curative intent, is a laudable aim. As such, neoadjuvant (primary) chemotherapy is being used with increased frequency in patients to decrease tumor bulk, assess the sensitivity of the cancer to systemic treatment and therein to treat potentially micrometastatic disease prior to surgery. Consequently, 5-year overall survival rates of up to 76% have been reported following neoadjuvant chemotherapy and hepatic resections in those with a complete pathologic response to their preoperative treatment [9,10].

However, despite its efficacy, the safety of bevazicumab has been frequently called into question. Phase I and II trials have identified hemorrhage, thromboembolism, proteinuria and hypertension as possible bevacizumab-associated adverse effects. In the original Phase III trials [5,6], the most common side effect was hypertension, which was managed with standard oral antihypertensives. However, gastrointestinal perforation occurred in 1–2% of patients receiving bevacizumab. It is worthwhile mentioning that these patients are part of a preselected, highly monitored trial population treated at major cancer centers, and are not reflective of the everyday clinical situation. This effect of bevacizumab may be directly caused by inhibition of VEGF-related epithelial healing [11,12]; bevacizumab also has a long half-life of approximately 20 days. Given its efficacy, the safety and optimal timing of surgery in patients who receive this agent has been unclear.

To study this further, Kesmodel et al. from MD Anderson Cancer Center (TX, USA) compared surgical complications in two groups of patients, studied retrospectively: those who received neoadjuvant chemotherapy with bevacizumab and those who received chemotherapy alone [13]. They investigated clinical variables known to affect postoperative outcomes in patients undergoing hepatic surgery, including factors previously reported to increase morbidity. Here, no statistically significant associations were observed between bevacizumab use and postoperative complications, despite a numerical imbalance – postoperative complications occurred in 47% overall (43% in those receiving chemotherapy alone versus 49% receiving chemotherapy and bevazicumab). Interestingly, over 70% of these complications were minor (i.e., grade 1 or 2), and in univariate analyses there were no significant differences in the number and types of complications between the two groups. In a multivariate analysis, a lower serum albumin level was associated with increased bleeding, and the authors concluded that neither the use of bevacizumab nor the timing of its administration was associated with complications.

In addition to these impressive data, three other studies that demonstrated a higher numerical rate of complications with bevacizumab, especially wound complications, also notably failed to reach statistical significance [14–16]. While it is tempting to speculate that this may be due to low sample sizes, these studies are retrospective in their nature and this is clearly stated. The type of chemotherapy, the decision to use bevacizumab, and the duration of chemotherapy and bevacizumab administration, along with the timing of surgical consultation, are at the discretion of the treating physician, but as the authors later point out, their study provides a foundation for validation in prospective trials.

It is difficult to decide how many patients and complications (and the size of the corresponding denominator) are required to then state that a drug is safe, or unsafe, to use before surgery, and in what timeframe. Experiences in thousands of patients in the trial and nontrial settings are often required to affirm safety of a drug. A recent meta-analysis of 15 randomized studies, comprising nearly 8000 patients treated with bevacizumab, observed a significantly increased risk of venous thromboembolism with a relative risk of 1.33 (95% CI: 1.13–11.56; p < 0.001) compared with controls [17]. However, this may be a consequence of additional cycles of chemotherapy given with bevacizumab, or a longer time to progression.

Any surgery at the time of receiving cytotoxic treatment is likely to be fraught with complications and difficulties. It is not surprising that we believe that surgery in patients within days of receiving bevacizumab should be contra-indicated. In a pooled analysis of five trials including 1745 patients with metastatic colorectal cancer, breast cancer or non-small-cell lung cancer, serious grade 3 and 4 bleeding occurred in 3.7% of patients treated with bevacizumab versus 1.8% in the control group, with concomitant aspirin intake severely exacerbating the bleeding [18]. The overall incidence of serious bleeding in patients with metastatic colorectal cancer receiving bevacizumab as first-line therapy has been reported at 2.5%, but there are no data, once again, for any group of patients regarding how long we should wait after bevacizumab prior to resection. Similarly, there are no data on when it might be safe to re-commence bevacizumab after surgery.

It is interesting to note, once again, that the focus of these studies is almost exclusively on liver secondaries, hepatic resections and related complications. As aforementioned, removal of the primary colon cancer is often undertaken concomitantly, before or after hepatic resections. We recently managed an elderly patient who had a resected extensive adenocarcinoma at the hepatic flexure. After three cycles of low-dose fluorouracil, oxaliplatin and leucovorin (FOLFOX) and bevacizumab, he developed abdominal pain and computed tomography revealed free air around the primary anastomosis. In this case, we elected to manage him conservatively and he recovered rapidly with antibiotics and reduced oral intake. Other isolated reports of such late anastomotic complications exist [19]; however, the role of bevacizumab in colonic anastomotic failure is unclear, and there is little doubt that future studies should not solely focus on issues related to the liver.

There are several surgical scenarios in these patients that require scrutiny: elective resection of the primary; elective resection of metastases (usually hepatic); emergency surgery for tumor complications such as bleeding, perforation or obstruction; and other elective nontumor-related surgery such as appendicitis, cholecystitis or fractured hips. Sadly, we are lacking data on all of these scenarios, and also lack data on when best to perform hepatic resections. Many of the hepatic issues discussed will be part of the aims of an NCI Phase III clinical trial to evaluate neoadjuvant versus adjuvant FOLFOX plus bevacizumab in patients with potentially resectable hepatic colorectal metastases. The trial will randomly assign patients with resectable colorectal hepatic metastases to FOLFOX plus bevacizumab before resection, followed by the same combination postoperatively, versus hepatic resection first, followed by postoperative FOLFOX plus bevacizumab. Although the primary end point will be progression-free survival, many secondary end points will be evaluated, including perioperative morbidity related to the agents utilized. At present, while bevacizumab appears to be generally safe, the higher numerical rate of complications in these small studies, despite a lack of statistical significance, suggests that increased vigiliance should be exercised when using this drug before, or after, resections. In addition, we should not forget the outcome of primary colonic lesions in these studies either. In terms of toxicity, absence of evidence is not evidence of absence.

Financial & competing interests disclosure

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

No writing assistance was utilized in the production of this manuscript.

Additional information

Funding

The authors have no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.