Non-cardiovascular toxicities of molecularly targeted antiangiogenic agents

INTRODUCTION — Antiangiogenic agents (systemic medications that inhibit angiogenesis) are used in cancer therapy because angiogenesis is required for tumor growth [1].

Several classes of antiangiogenic agents are available that block the vascular endothelial growth factor (VEGF) pathway (figure 1), including:

●Ligand inhibitors – These bind to and inhibit ligand binding to the VEGF receptor, thereby preventing activation of the receptor. Examples include bevacizumab, ramucirumab, and aflibercept.

●Receptor tyrosine kinase inhibitors (TKIs) – These block the enzymatic activity of the intracellular domain of the VEGF receptor (VEGFR). Examples include sunitinib, sorafenib, pazopanib, vandetanib, cabozantinib, axitinib, ponatinib, lenvatinib, regorafenib, tivozanib, and fruquintinib.

Antiangiogenic agents are associated with a wide spectrum of toxicities, which may be fatal in rare cases [2,3]. While some adverse effects are shared with chemotherapy, many are unique to antiangiogenic agents. Toxicities of antiangiogenic agents may be specific to the drug class or to the agent itself.

This topic will discuss the non-cardiovascular toxicities of antiangiogenic agents.

Separate topics discuss:

●Cardiovascular toxicities of antiangiogenic agents – (See "Cardiovascular toxicities of molecularly targeted antiangiogenic agents".)

●Kidney toxicities of antiangiogenic agents – (See "Nephrotoxicity of molecularly targeted agents and immunotherapy", section on 'Antiangiogenic agents'.)

●Infusion reactions to monoclonal antibodies – (See "Infusion-related reactions to therapeutic monoclonal antibodies used for cancer therapy".)

●The basic science of angiogenesis inhibitors – (See "Overview of angiogenesis inhibitors".)

RISK OF FATALITY — Many toxicities of antiangiogenic agents are serious and potentially fatal, particularly the cardiovascular effects and hemorrhage. Careful selection of therapy is based on patient characteristics (reasonable performance status, blood pressure control, and lack of serious cardiovascular events within six months) as well as close monitoring and prompt intervention are necessary to alleviate these risks.

Meta-analyses have demonstrated a small risk of fatal adverse events (approximately 1.5 to 2.5 percent, relative risk [RR] 1.5-2.2) with both antiangiogenic tyrosine kinase inhibitors (TKIs) and bevacizumab [2-4]. In one analysis, bevacizumab was associated with an increased risk of fatal events in combination with taxanes or platinum agents (RR 3.49) but not in combination with other agents (RR 0.85) [2].

In two meta-analyses, hemorrhage was the most common fatal adverse event with both classes of agents; other causes of treatment-related death were cardiac, gastrointestinal tract perforation, hepatic dysfunction, infection, and cerebrovascular events [2,3].

Another meta-analysis examining fatal events with antiangiogenic TKIs noted higher numbers of several complications in the TKI arm [4]:

●Heart failure

●Pulmonary emboli

●Liver failure

●Intestinal perforation

●Pneumonia/respiratory failure

In this analysis, the increased RR for death with the antiangiogenic TKIs was statistically significant for renal cell carcinoma (RCC) but not for lung cancer [4]. However, the increased risk seen in patients treated for RCC may be ascribed partly to increased exposure time to the antiangiogenic TKIs in these patients relative to those with lung cancer.

CLASS SIDE EFFECTS OF VEGF INHIBITORS

Kidney toxicity — Antiangiogenic agents are associated with specific class-effect kidney toxicities such as proteinuria/nephrotic syndrome and thrombotic microangiopathy, among other kidney disorders. Further details on the nephrotoxicity of antiangiogenic agents are discussed separately. (See "Nephrotoxicity of molecularly targeted agents and immunotherapy", section on 'Antiangiogenic agents'.)

Erythrocytosis — In mouse models, very high-grade VEGF inhibition is associated with reversible erythrocytosis, reticulocytosis, and increases in red blood cell mass [5]. Cyclic changes in hemoglobin levels have been noted in patients with metastatic renal cell cancer receiving sunitinib [6,7]. In addition, sunitinib administration is frequently associated with macrocytosis with an elevated mean corpuscular volume (MCV) in the absence of folate deficiency.

Although the mechanism underlying erythrocytosis is not established, it could be due either to temporary loss of intravascular fluid volume (ie, relative polycythemia) caused by inhibition of VEGFR-2 and subsequent reduction in nitric oxide, or secondarily by an increase in erythropoiesis driven by hepatic erythropoietin synthesis [6,8].

Secondary erythrocytosis has also been described in patients treated with sorafenib, axitinib, and bevacizumab [9,10]. Intriguingly, in a study of 10 patients with RCC who developed secondary erythrocytosis with bevacizumab, the peak increase of hemoglobin correlated with longer progression-free survival.

Bleeding — All VEGF-targeted agents have been associated with an increased risk of hemorrhage. Perturbation of endothelial cell function by targeting molecules expressed on the endothelial cell surface, such as VEGFRs, may heighten susceptibility to bleeding. In addition, direct antitumor activity leading to cavitation in an area of tumor that contains poorly developed neovessels that lack a well-formed musculature has been postulated to cause pulmonary hemorrhage, particularly in squamous cell lung cancer [11]. Finally, coexisting thrombocytopenia may aggravate bleeding.

Bevacizumab and aflibercept — Meta-analyses have demonstrated an increase in the risk of bleeding with bevacizumab; most commonly this is grade 1 epistaxis, but serious and, in some cases, fatal hemorrhagic events, including hemoptysis, gastrointestinal bleeding, hematemesis, intracerebral hemorrhage, epistaxis, and vaginal bleeding, have occurred.

●In one trial-level meta-analysis, the risk of major bleeding in patients with advanced solid tumors who were treated with bevacizumab (at any dose) was 2.8 percent (95% CI 2.1-3.6) [12]. The overall relative risk (RR) of high-grade bleeding was 1.60 and was lower at 2.5 mg/kg/week (RR 1.27) compared with 5 mg/kg per week (RR 3.02). Higher risks were observed in patients with non-small cell lung cancer (NSCLC; RR 3.41), RCC (RR 6.37), and colorectal cancer (RR 9.11) receiving bevacizumab 5 mg/kg per week.

●In another trial-level meta-analysis, the incidence of all-grade hemorrhage with bevacizumab was 30 percent, and 3.5 percent were high grade [13]. The overall RR for any grade bleeding was 2.48, with RRs of 3.02 and 2.01 for 5 and 2.5 mg/kg/week, respectively. Most hemorrhages occurred within the first five months of treatment. The most common type of hemorrhage was epistaxis, although hemoptysis, gastrointestinal bleeding, intracerebral hemorrhage, and intratumoral hemorrhage also occurred. The RR of high-grade bleeding was 1.91 and the risk of fatal bleeding was low (0.8 percent) and significantly elevated only in lung cancer (RR 5.02).

●In contrast to these two studies, in an individual patient level meta-analysis evaluating the risk of arterial thromboembolism, the risk of serious bleeding was modestly increased with bevacizumab but the difference was not statistically significant [14]. Grade 3 and 4 bleeding events occurred in 3.7 percent of bevacizumab-treated patients versus 1.8 percent of control patients. The rate of bleeding events per 100 person-years was 5.3 with bevacizumab and 3.3 for controls (ratio = 1.6, 95% CI 0.86 to 2.97). Baseline or on-study aspirin use was associated with a modest 1.3-fold increase in the risk of grade 3 or 4 bleeding events in both treatment groups (from 3.6 to 4.7 percent in bevacizumab-treated patients and from 1.7 to 2.2 percent for control subjects).

In contrast, the risk of severe (grade 3 or 4) bleeding with bevacizumab was not increased in patients receiving primary anticoagulant prophylaxis for venous thromboembolic disease, including low-dose aspirin in an analysis of the BRiTE observational registry [15].

Less information is available for aflibercept. In a phase III trial, epistaxis was noted in 28 percent of aflibercept-treated patients (versus 7 percent with chemotherapy alone), and the rate of grade 3 or 4 hemorrhage was 3 versus 1.7 percent in the control group [16].

Ramucirumab — Although severe and sometimes fatal hemorrhage has occurred in patients treated with ramucirumab, the risk appears to be low. In a 2017 meta-analysis of individual patient safety data from six placebo-controlled randomized trials in a variety of malignancies, the risk of all-grade bleeding was 38 versus 19 percent, but it was not significantly elevated for severe (≥grade 3) bleeding (2.7 versus 2.8 percent) [17]. The product labeling for ramucirumab includes a Boxed Warning to permanently discontinue ramucirumab in patients who experience severe bleeding [18].

Antiangiogenic tyrosine kinase inhibitors — An increased risk of bleeding has also been reported for the antiangiogenic TKIs:

●In a meta-analysis of 27 randomized trials, patients treated with an antiangiogenic TKI (vandetanib, sunitinib, sorafenib, axitinib, pazopanib, or regorafenib) had an overall incidence of all-grade and high-grade bleeding events of 9.1 and 1.3 percent, respectively [19]. The RR for all-grade bleeding was higher for patients receiving a TKI than controls (RR 1.67; 95% CI 1.19-2.33). The risk of all-grade hemorrhage varied significantly by tumor type and the specific TKI. The risk of all-grade hemorrhage was highest in patients with gastrointestinal stromal tumors, while for high-grade hemorrhage, it was highest for melanoma; in both cases, rates were lowest for lung cancer. The TKIs with the highest risk of all-grade hemorrhage were sorafenib, sunitinib, and pazopanib.

●As noted above, in two separate meta-analyses that examined fatal adverse events with antiangiogenic TKIs and bevacizumab in advanced solid tumors, hemorrhage was the most common toxic cause of death with both classes of VEGF inhibitors [2,3]. (See 'Risk of fatality' above.)

Special categories of bleeding

Intracranial bleeding — Concerns have been raised about a potential increase in the risk of intracerebral hemorrhage (ICH) in patients treated with bevacizumab who have brain metastases. In a phase I study, bevacizumab was associated with fatal ICH in a patient with an unsuspected intracerebral metastasis from hepatocellular carcinoma [20]. Based on this single anecdotal report, patients with brain metastases have been excluded from most clinical trials evaluating bevacizumab. However, subsequent accumulation of data has demonstrated that the risk of ICH in patients with treated, nonhemorrhagic brain metastases or with previously undiagnosed brain metastases or with treatment-emergent brain metastases does not appear to be significantly greater than that of patients with brain metastases who are not treated with bevacizumab [21-29]:

●A safety analysis analyzed the incidence of ICH with bevacizumab in approximately 13,000 patients from randomized and nonrandomized trials conducted in patients with breast, non-small cell lung, pancreatic, renal cell, or colorectal cancer [21]. Patients with central nervous system (CNS) metastases were at similar risk of developing ICH, independent of bevacizumab therapy. Among 187 patients who were found to have occult brain metastases, 3 of 91 bevacizumab treated patients developed grade 4 ICH, compared with one fatal ICH among 96 controls.

●In a prospective study, 115 patients with previously treated brain metastases were given bevacizumab in combination with systemic chemotherapy or erlotinib as first or second-line therapy for non-small cell lung cancer (NSCLC) [22]. At a median on-study duration of six months, there were no reported episodes of ICH (95% CI 0.0-3.3 percent). Eighty percent had received prior whole-brain radiation therapy with or without radiosurgery and/or neurosurgery, and 19.1 percent had received radiosurgery alone, and one patient (0.9 percent) underwent neurosurgery alone. A small proportion of patients were on enoxaparin (8.7 percent) and warfarin (7.0 percent), and 9.6 percent of patients were on daily aspirin. Of the five patients who developed a pulmonary or non-CNS, nonpulmonary hemorrhage while receiving bevacizumab, none had a risk factor for hemorrhage.

●An evidence-based review of the incidence of CNS bleeding with anti-VEGF therapy in patients with NSCLC and brain metastases concluded that neither bevacizumab nor sunitinib/sorafenib increased the risk of ICH in patients with treatment-emergent, pretreated, or untreated occult brain metastases [29].

●Even in glioblastomas, which are highly vascular, ICH appears to be uncommon; this adverse event occurred in 0 to 3.8 percent of patients across studies of bevacizumab [25,26]. However, concurrent anticoagulation and bevacizumab therapy may increase the risk of hemorrhage in these patients (11 versus 3 percent overall rate of ICH in anticoagulated versus non-anticoagulated patients in one retrospective review [27]). (See "Management of recurrent high-grade gliomas", section on 'Side effects'.)

●CNS metastases from renal cell cancer have a propensity to bleed. (See "Epidemiology, clinical manifestations, and diagnosis of brain metastases", section on 'Stroke'.)

An early report suggested a high incidence of ICH in patients with metastatic RCC and brain metastases who were treated with sunitinib or sorafenib [30]. In a retrospective review of 67 patients, five died of an ICH within two weeks after initiating therapy with sunitinib or sorafenib, four of whom had known brain metastases, all treated with RT. However, this could not be confirmed in an expanded access series analyzed the safety and efficacy of sunitinib treatment in 4564 patients with advanced RCC, of whom 321 had brain metastases [28]. Only one patient with a brain metastasis had a mild, treatment-related cerebral hemorrhage.

Data regarding ICH risk with other antiangiogenic TKIs used to treat RCC are lacking.

Taken together, these results suggest that patients with a history of treated nonhemorrhagic brain metastases probably should not be excluded from systemic therapy with a VEGF inhibitor as long as they are not on concurrent anticoagulation. In practice, antiangiogenic therapy is generally held during local therapy for brain metastases.

The decision to use bevacizumab in an anticoagulated patient with a recurrent primary brain tumor is more complex and must be based on a careful assessment of the risk to benefit ratio. This subject is discussed in detail elsewhere. (See "Management of recurrent high-grade gliomas", section on 'Side effects'.)

Pulmonary hemorrhage and cavitation — Pulmonary hemorrhage is a known complication of bevacizumab, especially in patients with squamous cell NSCLC [11,31,32]. Hemoptysis has not been seen in patients receiving bevacizumab for advanced colorectal or breast cancer.

Central tumor cavitation is common with the use of antiangiogenic agents, and is reported in 14 to 25 percent of cases of NSCLC during bevacizumab treatment [32,33]. In a retrospective study of phase II trials, the presence of baseline cavitation but not central tumor location was associated with a higher risk of hemorrhage [32,34].

Management — The VEGF inhibitor needs to be discontinued in the setting of a severe bleed and supportive transfusions should be instituted. Given the short half-life of antiangiogenic TKIs (unlike bevacizumab), cessation of the drug may rapidly reverse bleeding in this context. Aspirin and anticoagulants should be discontinued, with careful consideration of the risk/benefit ratio before resumption.

Minor bleeding (eg, epistaxis) may be managed symptomatically with no discontinuation or only temporary cessation of the agent. (See "Approach to the adult with epistaxis".)

Bronchoscopic laser coagulation, electrocautery, argon plasma coagulation, hemostatic tamponade, and bronchial artery embolization may be useful in severe pulmonary hemorrhage [35,36]. Endobronchial irrigation with cold saline or epinephrine solution appears to have limited efficacy [37]. Radiation therapy and surgery may be necessary to salvage some cases. (See "Evaluation and management of life-threatening hemoptysis".)

The management of intracranial hemorrhage should follow good clinical practice guidelines. (See "Spontaneous intracerebral hemorrhage: Acute treatment and prognosis".)

Delayed wound healing

Bevacizumab — Bevacizumab has been associated with impaired wound healing in a variety of settings. This adverse effect is likely due to the critical role played by VEGF and angiogenesis during the early stages of wound healing. (See "Basic principles of wound healing".)

The incidence of wound healing problems with bevacizumab was addressed in a meta-analysis comparing fluorouracil (FU)-based chemotherapy with or without bevacizumab in patients who underwent surgery after beginning chemotherapy for metastatic colorectal cancer (mCRC) [38]. Wound healing complications were observed in a higher fraction of those receiving bevacizumab (3.4 [1 of 29] versus 13 [10 of 75] percent), although this difference was not statistically significant. Of the 10 patients who experienced wound healing complications after surgery with bevacizumab plus chemotherapy, the time interval between bevacizumab and surgery was 0 to 29 days in five patients, 30 to 59 days in five; no patient with more than 60 days between administration of bevacizumab and surgery had a wound complication.

Based on these data, and the long half-life of bevacizumab (20 days), at least 28 days (preferably six to eight weeks) should elapse between surgery and the first dose of bevacizumab (or the last dose of bevacizumab prior to surgery), when possible [38]. (See "Potentially resectable colorectal cancer liver metastases: Integration of surgery and chemotherapy", section on 'Issues related to bevacizumab'.)

The safety of surgery performed ≥6 weeks after the last dose of bevacizumab was confirmed in a retrospective study of three prospective trials of bevacizumab in patients with metastatic breast cancer [39]. There was a low risk of severe bleeding (0.1 to 0.9 percent) and a low risk of severe wound-healing complications (1.3 to 2.2 percent).

Hepatic metastasectomy — Impaired wound healing is especially germane to patients receiving bevacizumab prior to hepatic metastasectomy. Concerns about impaired wound healing and possibly impaired hepatic regeneration may affect the safety of metastasectomy, particularly if performed too soon after bevacizumab administration. It is generally recommended that six to eight weeks should elapse between the last dose of bevacizumab and elective hepatic resection. This subject is discussed in detail elsewhere. (See "Potentially resectable colorectal cancer liver metastases: Integration of surgery and chemotherapy".)

Recurrent glioma — Among patients treated for glioma, use of preoperative bevacizumab also appears to be associated with delayed wound healing or wound dehiscence [40,41]. In a series of 209 patients undergoing a second or third reoperation for recurrent glioblastoma, significantly more patients receiving preoperative bevacizumab developed wound healing complications than did non-bevacizumab-treated patients (35 versus 10 percent) [41]. Wound healing complications developed in only 6 percent of those who only received bevacizumab postoperatively. The incidence of wound complications was most striking for the third craniotomy (44 percent) and for a shorter delay between bevacizumab and surgery. Based on these results, the authors recommend that craniotomy not be performed until at least 28 days after the last dose of bevacizumab.

Chest wall port placement — Timing of administration also affects wound healing after chest wall port placement. In two studies, administration of bevacizumab within 10 to 14 days of chest wall placement of an implantable venous access device (port) was associated with a higher incidence of wound healing complications and wound dehiscence [42,43]. When possible, bevacizumab should be held for at least two weeks after placement of a port.

Tracheoesophageal fistula — The development of tracheoesophageal fistula has been reported in patients who were given bevacizumab with concomitant chemoradiotherapy or in the setting of prior thoracic irradiation:

●Two independent phase II trials were discontinued for safety reasons after enrolling 29 patients with limited stage small cell lung cancer and five patients with advanced NSCLC, respectively [44]. Among the 34 patients, there were four confirmed cases of tracheoesophageal fistula and one suspected clinically. In all cases, patients were receiving bevacizumab with concomitant chemoradiotherapy.

●There are at least two reported cases of tracheoesophageal fistula developing in patients treated with bevacizumab who have a prior history of mediastinal irradiation [45,46].

Bevacizumab should be used cautiously in patients who have received prior mediastinal irradiation and avoided altogether in patients receiving concomitant chemoradiotherapy.

Ramucirumab — There are few data on ramucirumab use in patients with serious or nonhealing wounds. In a 2017 meta-analysis of individual patient safety data from six randomized placebo-controlled trials of ramucirumab in a variety of malignancies (totaling 2748 ramucirumab-treated patients and 2248 placebo-treated patients), wound healing complications developed in only 14 patients treated with ramucirumab, compared with four patients in the control group (0.5 versus 0.2 percent). Nevertheless, because of the potential risk of delayed wound healing, ramucirumab should be withheld prior to surgery [47].

Antiangiogenic tyrosine kinase inhibitors — Impaired wound healing (and reopening of previously healed wounds) has also been observed following treatment with antiangiogenic TKIs [28,48-51]:

●In a retrospective review of all patients undergoing cytoreductive nephrectomy at MD Anderson over a seven-year period, use of presurgical systemic targeted therapy with a broad range of VEGF inhibitors including TKIs was predictive of having a complication >90 days postoperatively, having multiple complications, and having wound complications [49].

●Another study comparing perioperative complications and surgical outcomes among 14 patients undergoing surgery after neoadjuvant sunitinib or sorafenib versus a control group of 73 consecutively treated patients who underwent surgery in the absence of prior systemic therapy observed less hemorrhagic and wound healing issues but a significant increase in incidence and severity of intraoperative adhesions during debulking nephrectomy in patients who received preoperative therapy [28]. The median time from TKI discontinuation to surgery was two weeks.

Based on these findings, in our view, all antiangiogenic TKIs should be ideally held for one week prior to elective major surgery, unless otherwise specified in the United States Prescribing Information for these agents, and all should be held for two weeks postoperatively and/or until wounds are reasonably healed.

At many institutions, therapy with these agents is held for four weeks after major surgery and for at least two weeks after minor surgery, although there are no prospective data validating this approach. The decision to resume therapy following a major surgical intervention should be based on clinical judgment of recovery from surgery.

Intestinal perforation/fistula formation — All VEGF-targeted therapies can cause gastrointestinal perforation (GIP) and fistula formation, although this complication is best described in patients receiving bevacizumab. The mechanism by which these drugs contribute to GIP has not been proven, but proposed mechanisms include intestinal wall disruption (ulceration) in areas of tumor necrosis, disturbed platelet-endothelial cell homeostasis causing submucosal inflammation and subsequent ulcer formation, impaired healing of pathologic or surgical bowel injury, and mesenteric ischemia from thrombosis and/or vasoconstriction [52-54].

Bevacizumab — GIP is an infrequent but potentially fatal toxicity of bevacizumab. GIP may lead to peritonitis requiring emergency operative intervention, fistula formation [55], or intraabdominal abscess. GIP has been reported in patients treated with bevacizumab for a variety of malignancies, but are most often described in the setting of mCRC and epithelial ovarian cancer (EOC). Clinicians should maintain a high index of suspicion for GIP in patients who develop acute abdominal pain while receiving bevacizumab, even if they have no apparent risk factors.

Diseases other than ovarian cancer — The risk of GIP in patients treated with bevacizumab for conditions other than EOC has been addressed in the following studies:

●In randomized trials of chemotherapy with and without bevacizumab in patients with mCRC and community-based cohort studies, the incidence of GIP in patients treated with bevacizumab has ranged from 1 to 4 percent. In a large community-based BRiTE observational cohort of patients treated with bevacizumab for mCRC, 37 of 1953 evaluable patients (1.9 percent) experienced GIP [56]. The majority of GIPs (26 of 37) occurred ≤6 months after starting bevacizumab (median, 3.35 months). In multivariate analysis, age ≥65 years was significantly associated with lower GIP risk (1.1 versus 2.6 percent), while intact primary tumor (3 versus 1.7 percent) and prior adjuvant radiation therapy (3.4 versus 1.7 percent) were associated with increased risk. Neither a prior history of peptic ulcer disease nor use of aspirin or other NSAIDs was a risk factor. No cumulative risk was observed with duration of bevacizumab exposure.

●All patients receiving bevacizumab are at risk, regardless of the type of malignancy; however, risk is variable. In a meta-analysis of over 12,000 patients from 17 randomized trials conducted in a variety of tumor types (six in mCRC, three NSCLC, two RCC, two pancreatic, four breast cancer, no trials in ovarian cancer), the incidence of GIP was approximately 1 percent overall regardless of tumor type, and the mortality rate was 22 percent [57]. The RR compared with controls not receiving bevacizumab was 2.14 (95% CI 1.19-3.85), and the risk was greater with higher bevacizumab doses (RRs for 5 and 2.5 mg/kg per week were 2.67 [95% CI 1.14-6.26] and 1.61 [95% CI 0.76-3.38], respectively). Risks were highest in patients treated for CRC (RR 3.10, 95% CI 1.26-7.63) and RCC (RR 5.67, 0.66-48.42). Among patients with NSCLC, risk was not significantly elevated compared with controls (RR 1.55, 95% CI 0.37-6.59).

Although several risk factors have been described for GIP during bevacizumab treatment, bowel perforation may occur even in the absence of predisposing risk factors [58], and it remains difficult to predict which patients will develop this complication. Many cases involve perforation of an in situ bowel primary [59]. However, surgical site complications can also occur at previously resected primary sites, often in the setting of previous irradiation or a prior anastomotic leak [55,60,61]. Perforation of an ileal neobladder has also been reported [62].

GI tract ulcers, possibly caused by bevacizumab, may predispose to perforation. One report described 18 patients with advanced colorectal cancer who participated in a phase III study and who were receiving oxaliplatin, capecitabine, and bevacizumab with or without cetuximab for mCRC who developed a GI ulcer (n = 6), a GIP (n = 8), or a perforated ulcer (n = 4) [63]. The occurrence of the perforations early in treatment, the established role of VEGF in ulcer healing [64], and the inhibitory effect of bevacizumab on wound healing support the causative role of bevacizumab in ulcerogenesis.

Accumulating data suggest a significantly increased risk of perforation in patients treated with bevacizumab for mCRC who subsequently undergo placement of an enteral stent. Thus, colonic stenting should not be performed in patients who are receiving bevacizumab. These data are addressed in detail elsewhere. (See "Enteral stents for the management of malignant colorectal obstruction", section on 'Stenting in the setting of adjunctive therapy'.)

Another potential risk factor for GIP in patients receiving bevacizumab for treatment of glioma is use of concurrent high-dose glucocorticoids [65]. (See "Management of recurrent high-grade gliomas", section on 'Bevacizumab'.)

Ovarian cancer — Although GIP has been reported in patients treated with bevacizumab for a variety of cancer primaries, it has attracted substantial attention in epithelial ovarian cancer (EOC) due to early reports noting high rates of GIP in these patients, and the common involvement of both the mesentery and the bowels in women with this disease.

Early studies suggested rates as high as 10 to 11 percent:

●An early phase II study was terminated early when five GIPs were observed among 44 women (11 percent) treated with bevacizumab monotherapy for chemorefractory disease [66].

●In another report of 112 patients with EOC who were treated with bevacizumab-containing regimens, 12 experienced a serious gastrointestinal event (10 [9 percent] had a GIP, and two a fistula) [67]. In this series the 30-day mortality rate was 50 percent, and 30 percent had died within one week of diagnosis. The only risk factor associated with GIP/fistula formation was rectovaginal nodularity (OR 3.64, 95% CI 1.1-12.1).

These studies lacked control groups for comparison, and they were conducted in populations of women with advanced, chemotherapy-refractory disease. It has been suggested that GIP may be more common in patients who are heavily pretreated with chemotherapy or who have diffuse peritoneal disease, significant small bowel disease, or bowel obstruction [52,68-70], and that the exclusion of such patients from bevacizumab therapy results in an underestimation of the true rate of GIP [71]. Even lower rates of GIP and fistula formation have in fact been reported in subsequent phase III trials comparing bevacizumab-containing versus non-bevacizumab chemotherapy for women with EOC, possibly because women were excluded who have evidence of GI tract obstruction, and many of the trials were undertaken in the first-line setting, or for women with recurrent, platinum-sensitive disease rather than in heavily pretreated women with advanced disease. As examples (see "First-line chemotherapy for advanced (stage III or IV) epithelial ovarian, fallopian tube, and peritoneal cancer", section on 'Incorporation of angiogenesis inhibitors' and "Medical treatment for relapsed epithelial ovarian, fallopian tube, or peritoneal cancer: Platinum-sensitive disease", section on 'Angiogenesis inhibitors'):

●In Gynecologic Oncology Group trial 218, in which 1873 previously untreated women with ovarian cancer were randomly assigned to paclitaxel plus carboplatin with or without bevacizumab, the risk of GIP/fistula formation with and without bevacizumab was 3 versus 1 percent [72]. In a later analysis, GI tract fistulas of any grade were reported in 4 of the 625 patients treated with chemotherapy alone compared with 8 of the 1248 patients receiving bevacizumab; the corresponding numbers for perforation were 2 of 625 versus 18 of 1248 [73]. Specific risk factors for adverse GI events with bevacizumab included a history of treatment for inflammatory bowel disease and bowel resection at primary surgery.

●Similarly, low rates of GIP were reported in the ICON7 trial of first line carboplatin plus paclitaxel with or without bevacizumab [74]. Among 1528 women enrolled to the trial, GIP occurred in 10 bevacizumab treated patients (versus 3 in the control group, 1.3 versus <1 percent), and fistula formation/abscess developed in 13 (versus 10 patients in the control group, 1.7 versus 1.3 percent).

●The role of bevacizumab in recurrent platinum-sensitive EOC was studied in the OCEANS trial, in which 484 women were randomly assigned to carboplatin plus gemcitabine with or without bevacizumab [75]. There were no reported cases of GIP, and rates of fistula/abscess formation were also low in the bevacizumab group (4 versus 1 in the control group, 1.6 versus 0.4 percent).

Aflibercept — Fewer data are available for aflibercept. In a phase III trial conducted in 1226 patients with mCRC, GIP occurred in the same percentage of patients treated with chemotherapy alone or with aflibercept (0.5 percent) [16]. Rates of fistula formation of the GI tract with and without aflibercept were 1.1 versus 0.3 percent, and for fistulas from other than GI origin, rates were 0.3 versus 0.2 percent.

The elimination half-life of aflibercept is shorter than for bevacizumab (6 versus 20 days). There are no data addressing the safety of surgery in patients receiving treatment with this agent.

Ramucirumab — Rates of GIP are also low with ramucirumab. In a 2017 meta-analysis of individual patient safety data from six placebo-controlled trials of ramucirumab in a variety of malignancies, the incidence of GIP was 1.1, versus 0.3 percent with placebo [17].

Antiangiogenic tyrosine kinase inhibitors — Whether antiangiogenic TKIs increase the risk of GI perforation is uncertain. There are few cases reported in trials and case reports [76-81]. A systematic review of 5352 patients with a variety of solid tumors from 20 clinical trials found an overall 1.3 percent incidence of GI perforation among patients receiving antiangiogenic TKIs (with no statistically significant increase compared with controls and with a mortality of 29 percent) [82]. When compared with patients treated with a control medication in these trials, there was no significant increase in the risk of GI perforation with the use of antiangiogenic TKIs, although the confidence intervals were wide (OR 2.99, 95% CI 0.85-10.53). Subgroup analysis suggested no variability in the incidence of GI perforation according to tumor type, specific TKI, or treatment regimen.

Management — In order to minimize the risk of GIP and fistula formation, at least 28 days (preferably six to eight weeks) should elapse between surgery and last dose of bevacizumab, except in emergency situations [38]. Given the shorter half-life of the antiangiogenic TKIs, at least two weeks may suffice in this context.

Any patient treated with an anti-VEGF agent, particularly bevacizumab, should be considered at risk for GIP. Mortality rates may be as high as 50 percent [83], and early detection might help reduce the morbidity and mortality of this complication.

Perforation may be asymptomatic [84], or it can present with abdominal pain due to peritoneal contamination, free air, hemoperitoneum, or intraabdominal abscess. Postoperative patients and those being treated for rectal cancer can present with a fistula or anastomotic leak.

For all patients with new-onset abdominal pain while receiving bevacizumab, the potential diagnosis of GIP should be evaluated urgently. The evaluation should include a complete history, physical examination (to rule out signs of peritonitis), and abdominal imaging (ie, radiograph to evaluate for free air, or non-contrasted computed tomography). Evaluation and diagnosis of suspected GIP are discussed in detail separately. (See "Evaluation of the adult with nontraumatic abdominal or flank pain in the emergency department", section on 'Ancillary studies'.)

Any case of GIP should result in the immediate and permanent discontinuation of VEGF-targeted therapy. Beyond that, there are no specific recommendations for management of documented GIP in patients receiving VEGF-targeted therapies. Patients with confirmed or highly suspected GIP whose overall condition is unstable secondary to the GIP should be considered for immediate surgical repair or diversion. Those who are more stable can be considered for less invasive management strategies such as bowel rest and broad-spectrum antibiotics with or without percutaneous drainage of concurrent abscesses [84,85]. The timing of the presentation, the patient's overall condition, their goals and wishes, and overall prognosis are important factors in the decision to explore these patients surgically.

Fatigue — Fatigue is common with all antiangiogenic TKIs, and bevacizumab may enhance fatigue caused by other agents given in combination, eg, interferon or chemotherapy. While mild fatigue is common, severe fatigue is seldom seen. In a meta-analysis of randomized trials, the RR of all-grade and high-grade fatigue with single agent VEGFR TKIs was 1.35 (95% CI 1.22-1.49) and 1.33 (95% CI 0.97-1.82), respectively [86]. The mechanism of fatigue due to angiogenesis inhibitors is unclear. Contributing factors may be other drugs, hypothyroidism (especially with sunitinib), anemia, dehydration (secondary to diarrhea, nausea, or vomiting), and cardiac dysfunction, which should be addressed. (See "Cancer-related fatigue: Prevalence, screening, and clinical assessment" and 'Thyroid dysfunction' below.)

Supportive care and psychostimulants may be employed. Severe fatigue may warrant dose modification or, rarely, discontinuation. (See "Cancer-related fatigue: Treatment".)

Dysphonia — Dysphonia has been observed with antiangiogenic TKIs, especially with newer more potent agents, such as axitinib, regorafenib, lenvatinib, and tivozanib [87-89]. Dysphonia was also reported in 25 percent of patients treated with aflibercept in a phase III trial conducted in mCRC (compared with 3 percent of the chemotherapy alone group) [16]. In contrast, dysphonia has not been reported in patients treated with bevacizumab.

Osteonecrosis of the jaw — Medication-related osteonecrosis of the jaw (MRONJ) is defined by areas of tissue breakdown and exposure of bone in the maxillofacial region that fail to heal within eight weeks in a patient who has not received jaw irradiation. (See "Medication-related osteonecrosis of the jaw in patients with cancer", section on 'Nomenclature and definition'.)

Isolated case reports have described MRONJ with VEGF inhibitors, mostly bevacizumab, sunitinib, and lenvatinib [90-100]. However, the incidence appears low, at least in the absence of other risk factors. As examples:

●In a retrospective study of 3560 patients receiving bevacizumab-containing therapy for advanced breast cancer in two double-blind, randomized trials and a large nonrandomized safety study, MRONJ occurred in 0.3 percent of patients receiving bevacizumab in the blinded phase of the two randomized trials and in 0.4 percent of patients in the single-arm study [101].

●In a randomized trial in patients with thyroid cancer, there was only one case of MRONJ among 261 lenvatinib-treated patients (0.4 percent), versus none with placebo [88]. Furthermore, at least some data support the view that risk for MRONJ is increased with lenvatinib only in patients with concomitant exposure to other risk factors such as high potency bisphosphonates, everolimus, glucocorticoids, or invasive dental procedures [102].

As a result of these data, the US Food and Drug Administration (FDA) and the European Medicines Agency has issued an MRONJ advisory only for bevacizumab and sunitinib. Sunitinib should be withheld for at least three weeks prior to scheduled dental surgery or invasive dental procedures, if possible. Lenvatinib should be withheld for at least one week prior to scheduled dental surgery or invasive dental procedures.

However, there is at least a potential risk of MRONJ associated with several other TKIs that target VEGF, including sorafenib, axitinib, vandetanib, regorafenib, and others. The incidence of MRONJ with these other agents is insufficiently characterized.

Concurrent use of antiangiogenic agents does represent a risk factor for MRONJ among patients with bone metastases who are also receiving therapy with an antiresorptive agent for prevention of skeletal related events. These data are discussed elsewhere. (See "Osteoclast inhibitors for patients with bone metastases from breast, prostate, and other solid tumors" and "Medication-related osteonecrosis of the jaw in patients with cancer", section on 'Concurrent antineoplastic therapy'.)

In view of the difficulty in treating established MRONJ, prevention is emphasized. Patients at risk could have a comprehensive dental examination and preventive dentistry (preemptive extraction of unsalvageable teeth and optimized periodontal health) before beginning therapy with an antiangiogenic agent, particularly if they are also receiving concomitant therapy with antiresorptive agent such as a bisphosphonate or denosumab for skeletal metastases. Oral examinations and hygiene status should be monitored during treatment. Invasive dental procedures (eg, placement of dental implants) should be avoided during therapy if at all possible. (See "Medication-related osteonecrosis of the jaw in patients with cancer", section on 'Prevention'.)

Treatment objectives for patients with an established diagnosis of MRONJ are to eliminate pain, control infection of the soft tissue and bone, and minimize the progression or occurrence of bone necrosis. Treatment has generally shifted away from aggressive surgical interventions and towards conservative therapy with limited debridement, antibiotics, and oral rinses with chlorhexidine or hydrogen peroxide. (See "Medication-related osteonecrosis of the jaw in patients with cancer", section on 'Treatment of established MRONJ'.)

Reversible posterior leukoencephalopathy and brain capillary leak syndrome — Reversible posterior leukoencephalopathy syndrome (RPLS) is a clinical radiographic syndrome of heterogeneous etiologies that are grouped together because of similar findings on neuroimaging studies. It is also often referred to as:

●Posterior reversible encephalopathy syndrome (PRES)

●Reversible posterior cerebral edema syndrome

●Posterior leukoencephalopathy syndrome

●Hyperperfusion encephalopathy

●Brain capillary leak syndrome

None of these names is completely satisfactory; the syndrome is not always reversible, and it is often not confined to either the white matter or the posterior regions of the brain. (See "Reversible posterior leukoencephalopathy syndrome", section on 'Introduction and terminology'.)

The pathogenesis of RPLS remains unclear, but it appears to be related to disordered cerebral autoregulation and endothelial dysfunction. (See "Reversible posterior leukoencephalopathy syndrome", section on 'Pathogenesis'.)

The clinical syndrome of reversible posterior leukoencephalopathy syndrome (RPLS) is characterized by headaches, altered consciousness, visual disturbances, and seizures; hypertension is frequent but not invariable. (See "Reversible posterior leukoencephalopathy syndrome", section on 'Clinical manifestations'.)

A wide variety of medical conditions have been implicated as causes of RPLS (table 1), including rapidly developing, fluctuating, or intermittent hypertension, and therapy with a VEGF-targeted agent (bevacizumab, sunitinib, sorafenib, lenvatinib, and pazopanib [103-113]). (See "Reversible posterior leukoencephalopathy syndrome", section on 'Related conditions'.)

Prevention and management — Stringent control of blood pressure is critical to avert this rare complication. In suspected cases of RPLS, discontinuation of the offending agent is essential. (See "Reversible posterior leukoencephalopathy syndrome", section on 'Management'.)

CLASS SIDE EFFECTS OF VEGF RECEPTOR TYROSINE KINASE INHIBITORS

Thyroid dysfunction — Thyroid dysfunction is frequently observed in patients treated with sunitinib. Typically this has been manifested by hypothyroidism [114-116]. In one series, 15 of 42 euthyroid patients with intact thyroid glands (36 percent) became hypothyroid (as defined as a persistently elevated level of thyroid stimulating hormone [TSH]) while receiving sunitinib for advanced gastrointestinal stromal tumors (GIST) [115]. The risk increased with longer duration of therapy (18, 29, and 90 percent in patients treated for 36, 52, and 96 weeks, respectively). The mean time to development of hypothyroidism was 50 weeks.

The mechanism by which sunitinib causes hypothyroidism has yet to be determined [117]. Inhibition of iodine uptake may be responsible [118].

Several reports have also described transient thyrotoxicosis with sunitinib treatment in patients with metastatic renal cell carcinoma (RCC) [119-121]. In one report, six patients developed thyrotoxicosis after starting treatment with sunitinib, and four later became hypothyroid [119]. Some of these patients appeared to have a destructive thyroiditis, characterized by transient thyrotoxicosis (with low radioiodine uptake, which distinguishes it from Graves' disease) followed by hypothyroidism associated with atrophy of thyroid follicular cells [119,120]. However, a case of lymphocytic thyroiditis accompanied by transient thyrotoxicosis has also been reported [122]. (See "Overview of thyroiditis" and "Disorders that cause hyperthyroidism", section on 'Thyroiditis'.)

Thyroid dysfunction also occurs in patients treated with sorafenib for metastatic RCC, but this appears to be less frequent [116,123-128]. In an analysis of thyroid function in 39 patients, hypothyroidism was seen in seven (18 percent) and hyperthyroidism in one [123]. Two of the hypothyroid patients (5 percent of the total) were symptomatic and required thyroid replacement therapy. Others report a frequency of thyroid hormone replacement therapy among patients taking sorafenib of 2.6 and 6 percent, respectively [116,128]. As with sunitinib, patients treated with sorafenib have also been reported to develop thyrotoxicosis with associated thyroiditis [123,124,127,129,130].

Hypothyroidism is also reported among patients treated with pazopanib, axitinib, and tivozanib and hence appears to be a class-effect of these agents [131-133]. Pazopanib appears to have the lowest reported incidence, with <10 percent of patients developing hypothyroidism in a phase III trial [132]. Hypothyroidism rates are approximately 11 percent with tivozanib [134].

Early reports with axitinib suggest a very high incidence. In a phase I trial, 89 percent of patients had elevations in TSH [131].

Finally, another potential mechanism for hypothyroidism that may be seen in patients being treated with antiangiogenic tyrosine kinase inhibitors (TKIs) for GISTs is consumptive hypothyroidism due to excessive degradation of thyroid hormone; however, this appears to be caused by overexpression of the thyroid hormone inactivating enzyme type 3 iodothyronine deiodinase (D3) within large GISTs and not a drug-related adverse effect. This subject is discussed in detail elsewhere. (See "Disorders that cause hypothyroidism", section on 'Consumptive hypothyroidism' and "Clinical presentation, diagnosis, and prognosis of gastrointestinal stromal tumors", section on 'Adults'.)

Lenvatinib is a multitargeted TKI approved for advanced, radioiodine-refractory, differentiated thyroid cancer and in combination with everolimus for RCC after one prior vascular endothelial growth factor (VEGF) inhibitor [1]. It has been associated with impaired suppression of TSH in patients receiving exogenous thyroid hormone supplementation [135]. TSH levels should be monitored monthly and thyroid replacement medication adjusted as needed. (See "Differentiated thyroid cancer refractory to standard treatment: Systemic therapy", section on 'Targetable mutation not identified'.)

Management — Because of the high prevalence of hypothyroidism, regular surveillance of TSH levels is warranted during therapy with antiangiogenic TKIs, more frequently in patients receiving sunitinib and lenvatinib. We suggest that thyroid function be evaluated at baseline and monitored every 4 to 12 weeks thereafter and earlier if dictated by symptoms. (See "Diagnosis of and screening for hypothyroidism in nonpregnant adults" and "Treatment of primary hypothyroidism in adults".)

Thyroid hormone supplementation should be given to symptomatic patients with hypothyroidism. Discontinuations and dose modifications of the VEGF receptor TKI is usually not necessary.

Transient thyrotoxicosis may be seen preceding hypothyroidism and there are no reports suggesting that therapeutic intervention is warranted.

Myelosuppression — Myelosuppression has been observed more commonly when utilizing antiangiogenic TKIs that more potently target Flt-3 (FMS-related tyrosine kinase 3 receptor) and KIT, which are required for hematopoiesis [136]. When comparing across different trials, sunitinib and sorafenib appear more myelosuppressive than more selective TKIs such as pazopanib, axitinib, and cabozantinib. A meta-analysis of trials of single agent sunitinib revealed high-grade neutropenia in 12.8 percent, thrombocytopenia in 10.7 percent, and anemia in 6.2 percent, respectively [137]. Indeed, the combination of sunitinib with relatively non-myelosuppressive chemotherapy regimens, eg, carboplatin plus paclitaxel, has appeared prohibitively myelosuppressive [138]. Moreover, thrombocytopenia and leukopenia induced by sunitinib appeared to be Flt-3 genotype and CYP1A polymorphism dependent, respectively, in retrospective studies [139,140]. The incidences of sorafenib-associated high-grade anemia, neutropenia, and thrombocytopenia in a meta-analysis were 2.0, 5.1, and 4.0 percent, respectively [141].

Bevacizumab is not known to cause significant myelosuppression.

Recurring grade 3 or 4 neutropenia or thrombocytopenia (table 2) persisting for at least five days and/or febrile neutropenia/bleeding require dose alterations.

Oral toxicity — Stomatitis and other manifestations of oral toxicity may be seen with all the antiangiogenic TKIs. Rates are variable, but it appears to be more common with sunitinib (29 to 48 percent), sorafenib (25 to 28 percent), cabozantinib (51 percent), and lenvatinib (40 percent) than with pazopanib (4 to 12 percent) and axitinib (15 percent) [87,142,143]. Symptoms of oral pain, dysesthesia/mucosal sensitivity, dysgeusia, dysphagia, dry mouth, and aphthous ulcers may occur rather than conventional oral mucositis [144]. The mechanism is unclear but may be related to poor healing of microtrauma (similar to the proposed mechanism for hand-foot skin reaction). (See 'Cutaneous toxicity' below and "Oral toxicity associated with systemic anticancer therapy", section on 'Etiology and risk factors'.)

In a phase III trial of FOLFIRI with or without aflibercept in metastatic colorectal cancer, rates of mucositis were higher in the combined therapy arm (55 versus 35 percent for all-grade, and 14 versus 5 percent for severe) [16], suggesting that anti-VEGF agents may exacerbate the risk of mucositis from chemotherapy; however, this has not been seen with bevacizumab.

Gastrointestinal toxicities — Diarrhea, nausea, and emesis have been observed with all antiangiogenic TKIs and are generally mild.

In clinical trials, diarrhea of any grade has been reported in 30 to 79 percent of patients (highest rates with vandetanib), with severe diarrhea (grade 3 or 4) in 3 to 17 percent [87,145-151]. Vandetanib is a moderately potent epidermal growth factor receptor (EGFR) inhibitor in addition to being a vascular endothelial growth factor receptor (VEGFR) inhibitor, which explains the somewhat higher incidence of diarrhea associated with this agent [147,148].

In clinical trials, nausea of any grade has been reported in 23 to 58 percent of treated patients, and rates are highest in patients treated with sunitinib, lenvatinib, cabozantinib [88,142,152,153]. Vomiting of any grade has been reported in 10 to 48 percent of treated patients; rates are lowest with pazopanib and sorafenib and highest with sunitinib, lenvatinib, and cabozantinib [88,142,152,153].

Pancreatic atrophy been reported in patients receiving long-term sorafenib. The possibility of pancreatic insufficiency should be considered if a patient treated with sorafenib has refractory diarrhea with clinical features suggesting of pancreatic exocrine insufficiency. (See 'Pancreatic atrophy' below and "Exocrine pancreatic insufficiency", section on 'Clinical manifestations'.)

Management — Symptomatic management of grade 1 or 2 diarrhea with an oral antidiarrheal agent, such as loperamide, and avoidance of foods and supplements (eg fiber) that increase gastrointestinal motility generally suffices [154]. Treatment interruption is necessary for grade 3 or 4 diarrhea, and dose modifications may be necessary to control diarrhea if the drug is continued.

Nausea and vomiting is rarely severe and can usually be controlled with oral antiemetics. However, caution should be exercised in combining vandetanib, lenvatinib, sunitinib, and sorafenib with serotonin antagonists such as granisetron and ondansetron, due to the potential for QTc interval prolongation and torsade de pointes. (See "Cardiovascular toxicities of molecularly targeted antiangiogenic agents", section on 'Prolongation of the QTc interval and cardiac arrhythmias' and "Prevention of chemotherapy-induced nausea and vomiting in adults", section on 'Cardiac issues'.)

Cutaneous toxicity — A range of cutaneous toxicities has been reported in patients receiving antiangiogenic TKIs (table 3). (See "Cutaneous adverse events of molecularly targeted therapy and other biologic agents used for cancer therapy", section on 'BCR-ABL tyrosine kinase inhibitors'.)

Hand-foot skin reaction (HFSR) is a common cutaneous manifestation of toxicity with oral antiangiogenic TKIs. HFSRs appear to be more common with sorafenib than sunitinib, and more common with axitinib, lenvatinib, regorafenib, and vandetanib, as compared with pazopanib [87,88,145,155]. In a meta-analysis that examined the types of cutaneous toxicities, vandetanib exhibited the highest incidence of rash (41 percent), while sorafenib was most commonly associated with HFSR (37 percent) and pruritus (14 percent) [156]. A meta-analysis of 831 patients who received cabozantinib demonstrated an overall incidence of HFSR of 35 percent and grade 3 HFSR in 10 percent [157]. Interestingly, as has been seen with hypertension, an association between skin toxicity and improved outcomes has been described [158,159]. The mechanisms of HFSR include inhibition of pericyte-mediated endothelial survival, resulting in damage to the capillary endothelium in the hands and feet. Notably, KIT is strongly expressed in the ductal epithelium of eccrine glands, where the drug can be excreted [160]. (See "Cutaneous adverse events of molecularly targeted therapy and other biologic agents used for cancer therapy", section on 'BCR-ABL tyrosine kinase inhibitors' and "Cardiovascular toxicities of molecularly targeted antiangiogenic agents", section on 'Association with antitumor efficacy'.)

Multiple case reports have also identified an increased risk of keratoacanthomas and squamous cell carcinoma in patients treated with sunitinib or sorafenib. (See "Cutaneous adverse events of molecularly targeted therapy and other biologic agents used for cancer therapy", section on 'Squamoproliferative lesions'.)

Vandetanib also inhibits the EGFR and has been associated with a generalized maculopapular erythematous acneiform rash that is typical of other EGFR inhibitors (eg, gefitinib, erlotinib, cetuximab). (See "Cutaneous adverse events of molecularly targeted therapy and other biologic agents used for cancer therapy", section on 'VEGFR/PDGFR inhibitors'.)

Prevention and management — The United States Prescribing Information provides suggested dose modifications for sorafenib based on the grade of HFSR (table 4) during therapy [161]. Guidelines are not available for the other agents, although the institution of supportive measures, dose reductions, or permanent discontinuations may be necessary depending on the severity and persistence of dermatologic toxicity.

Prevention and management strategies for HFSR in patients receiving sunitinib or sorafenib are also available from an international consensus group [154]. This subject is presented in detail elsewhere. (See "Cutaneous adverse events of molecularly targeted therapy and other biologic agents used for cancer therapy", section on 'Hand-foot skin reaction'.)

Prevention and management strategies for the acneiform eruption in patients receiving vandetanib and other EGFR inhibitors are provided in detail elsewhere. (See "Acneiform eruption secondary to epidermal growth factor receptor (EGFR) and MEK inhibitors", section on 'Prevention' and "Acneiform eruption secondary to epidermal growth factor receptor (EGFR) and MEK inhibitors", section on 'Management'.)

Given the early onset of skin toxicities and the association with the dose of regorafenib, a randomized phase II trial demonstrated that a dose-escalation strategy (starting dose 80 mg/day orally with weekly escalation by 40 mg increments to 160 mg/day if no significant drug-related adverse events occurred) may be an acceptable alternative to a standard-dose strategy (160 mg/day orally) for 21 days of a 28-day cycle [162]. Interestingly, the trough concentration of M5, a major metabolite of regorafenib, was associated with the incidence of skin toxicities [163]. However, pharmacokinetic monitoring is not used clinically. (See "Second- and later-line systemic therapy for metastatic colorectal cancer", section on 'Regorafenib'.)

Definitive guidelines for continuing versus discontinuing sorafenib in patients who develop squamoproliferative lesions while on sorafenib or sunitinib therapy have not been established. (See "Cutaneous adverse events of molecularly targeted therapy and other biologic agents used for cancer therapy", section on 'Squamoproliferative lesions'.)

Hepatotoxicity — Severe and occasionally fatal hepatotoxicity has been observed in clinical studies with all VEGFR TKIs. The risk for elevations in serum alanine transaminase (ALT), aspartate transaminase (AST), alkaline phosphatase (ALP), and bilirubin during therapy with VEGFR TKIs was addressed in a meta-analysis of 18,282 patients from 52 randomized controlled trials of a variety of VEGFR TKIs [164]. The incidence of hepatic failure with VEGFR TKIs was 0.8 percent overall. Compared with patients not treated with a VEGFR TKI, the relative risks (RRs) for all-grade elevations in ALT, AST, ALP and bilirubin in patients receiving a VEGFR TKI were 1.57 (95% CI 1.38-1.79), 1.57 (95% CI 1.36-1.81), 1.20 (95% CI 1.09-1.83), and 1.55 (95% CI 1.21-1.97), respectively. The RRs for high-grade elevations were 1.66 (95% CI 1.25-2.20), 1.61 (95% CI 1.21-2.14), 1.02 (95% CI 0.70-1.47), and 1.34 (95% CI 1.0-1.81), respectively. The RR of all-grade hepatotoxicity did not significantly differ between relatively more specific VEGFR TKIs, such as axitinib, as compared with less specific VEGFR TKIs (sunitinib, sorafenib, pazopanib, vandetanib, cabozantinib, ponatinib, regorafenib).

Monitoring and management — Patients treated with any of these agents should have a baseline evaluation of liver function tests (LFTs) and periodic re-evaluation during therapy. For patients treated with pazopanib, the United States Prescribing Information recommends assessing LFTs at baseline and at weeks 3, 5, 7, and 9, 12 and 16, and thereafter periodically as clinically indicated. Treatment should be interrupted or discontinued if evidence of hepatotoxicity is observed.

Specific recommendations for individual drugs include the following:

●Pazopanib – Polymorphisms in the uridine diphospho-glucuronosyltransferase 1A1 (UGT 1A1) enzyme that cause Gilbert's syndrome (the UGT1A1*28 allele) may be associated with pazopanib-induced hyperbilirubinemia. Isolated hyperbilirubinemia in these patients may represent a benign manifestation of Gilbert's syndrome, and continuation of pazopanib monotherapy is reasonable in this setting. This subject is discussed in more detail elsewhere. (See "Hepatotoxicity of molecularly targeted agents for cancer therapy", section on 'Pazopanib'.)

For patients with pre-existing moderate hepatic impairment, the United States Prescribing Information recommends that starting doses of pazopanib be reduced and that the drug be avoided in patients with pre-existing severe hepatic impairment, defined as total bilirubin >3 times the upper limit of normal (ULN) with any level of ALT/AST elevation.

●Ponatinib – As hepatic elimination is a major route of excretion, hepatic impairment may result in increased ponatinib exposure. The United States Prescribing Information recommends avoiding use of the drug in patients with moderate to severe (Child-Pugh B or C (table 5)) hepatic impairment unless the benefit outweighs the possible risk of ponatinib overexposure. (See "Hepatotoxicity of molecularly targeted agents for cancer therapy", section on 'Ponatinib'.)

●Sorafenib – Sorafenib-induced hepatitis is characterized by a hepatocellular pattern of liver damage with significant increases of transaminases, which may result in hepatic failure and death. Increases in bilirubin and International Normalized Ratio (INR) may also occur. The United States prescribing information recommends that the drug be discontinued in case of significantly increased transaminases without alternative explanation such as viral hepatitis or progressing underlying malignancy. There are conflicting data on the need for dose reduction in patients with preexisting liver dysfunction. (See "Hepatotoxicity of molecularly targeted agents for cancer therapy", section on 'Sorafenib'.)

●Sunitinib – The United States Prescribing Information recommends treatment interruption for bilirubin levels >3X ULN or AST/ALT >5X ULN, and discontinuation of sunitinib if there is no resolution or if patients subsequently experience severe changes in LFTs or have other signs and symptoms of liver failure. (See "Hepatotoxicity of molecularly targeted agents for cancer therapy", section on 'Sunitinib'.)

●Lenvatinib – The United States Prescribing Information recommends discontinuing the drug for grade 3 or worse liver impairment during therapy, and that the drug be permanently discontinued for hepatic failure. A reduced starting dose is recommended for patients with preexisting severe hepatic impairment (Child-Pugh C cirrhosis (table 5)). (See "Hepatotoxicity of molecularly targeted agents for cancer therapy", section on 'Lenvatinib'.)

●Axitinib – When used in combination with avelumab or pembrolizumab, the United States Prescribing Information recommends treatment interruption for ALT or AST ≥3X ULN but <10X ULN without concurrent total bilirubin ≥2X ULN, and discontinuing the drug for ALT/AST increase to >3X ULN with concurrent total bilirubin ≥2X ULN or any ALT/AST increase >10X ULN. A reduced starting dose is recommended for all patients with preexisting moderate hepatic impairment (Child-Pugh class B cirrhosis). (See "Hepatotoxicity of molecularly targeted agents for cancer therapy", section on 'Axitinib'.)

Pancreatitis — Elevations of the pancreatic enzymes lipase and amylase are frequently reported in patients treated with antiangiogenic TKIs, although overt pancreatitis appears to relatively rare, with the possible exception of ponatinib [149,165-167]:

●In a phase II trial of ponatinib in 81 patients with refractory chronic myelogenous leukemia, pancreatitis developed in 11 (14 percent) and was severe (grade 3 or worse) in four (5 percent) [166].

●The risk of pancreatitis in patients treated with VEGFR TKIs other than ponatinib, cabozantinib, or regorafenib was addressed in a meta-analysis that included 10,578 patients from 22 randomized trials [167]. Pancreatitis of any grade occurred in 25 of 5569 patients receiving a VEGFR TKI (0.4 percent), and high grade pancreatitis occurred in 22 (0.4 percent). The RR for all grade and high-grade pancreatitis for the TKI versus no TKI arms was 1.95 (95% CI 1.02 to 3.70) and 1.89 (95% CI 0.95 to 3.73), respectively. There was no differential impact of malignancy type or specific TKI agent on RR of either all grade or high-grade pancreatitis.

We closely monitor patients with pancreatic enzyme elevations and continue therapy with the TKI unless patients develop clinical pancreatitis.

Hypoglycemia — Blood glucose levels may be reduced in patients who are treated with antiangiogenic TKIs for a variety of malignancies. As an example, hypoglycemia has occurred in clinical trials in approximately 2 percent of patients receiving sunitinib for renal cell cancer or GIST, and in approximately 10 percent of patients treated with sunitinib for a pancreatic neuroendocrine tumor [168].

There are conflicting data as to whether hypoglycemia is more common in diabetic as compared with nondiabetic patients:

●In a retrospective series of 19 type II diabetic patients whose blood glucose was monitored at baseline and during treatment with sunitinib, blood glucose levels fell from 149 mg/dL at baseline to 117 mg/dL at four weeks (8.3 to 6.5 mmol/L) [169]. In contrast, blood glucose levels fell only minimally in nine nondiabetic patients (106 to 95 mg/dL [5.89 to 5.26 mmol/L]) who were similarly monitored. No serious episodes of hypoglycemia were reported.

●However, other data suggest that declines in blood glucose occur both in diabetic and nondiabetic patients [170]. In this retrospective review of diabetic (n = 17) and nondiabetic (n = 61) patients treated with sorafenib, sunitinib, imatinib, or dasatinib, all four drugs were associated with significant mean declines in blood glucose, which were greatest (52 mg/dL) with dasatinib. The magnitude of decline in blood glucose was similar in diabetic and nondiabetic patients. On the other hand, revised product labeling for sunitinib suggests that reductions in blood glucose levels may be worse among diabetic patients treated with sunitinib as compared with nondiabetic patients [168].

Blood glucose levels should be routinely checked during and after treatment for all patients who are treated with antiangiogenic TKIs. Antidiabetic medications should be adjusted if necessary to minimize the risk of hypoglycemia.

AGENT-SPECIFIC EFFECTS

Bevacizumab

Nasal septal perforation — Nasal septal perforation has been observed with bevacizumab administered in combination with chemotherapy [171,172]. The US Food and Drug Administration (FDA) recognized this complication in 2006 after seven postmarketing reports became available [173]. However, this is a rare complication. To date, only 18 cases have been published in the literature [174]. Although the data are sparse, some groups of patients may be at a higher risk of this complication. The perforations occurred in combination with taxanes and higher doses of bevacizumab (15 mg/kg once every three weeks) in many cases [171], but they have also been reported in patients treated with bevacizumab and fluorouracil (FU)-based chemotherapy [172]. The overall risk in this series of 100 patients treated for metastatic colorectal cancer was only 1 percent.

The mechanism may be related to underlying mucositis, delayed tissue repair, and additive or synergistic antiangiogenic activity of taxanes.

Management — When patients develop this complication, other potential causes of perforation should be excluded, even if the patient is receiving bevacizumab. The evaluation should include inspection of the nasal septum for evidence of infection and obtaining local cultures to exclude other causes of perforation [175]. Septal perforation is often seen in those who use intranasal cocaine. It may also result from septal surgery, atrophic rhinitis, granulomatosis with polyangiitis, and several other disorders (table 6). (See "Etiologies of nasal obstruction: An overview", section on 'Disorders affecting the septum'.)

Patients with a nasal septal perforation should be referred to an otolaryngologist for evaluation and management. Most perforations, particularly those that are posterior and asymptomatic, can be medically managed with frequent irrigation of the area with saline, and application of lubricant gels and other supportive deal with epistaxis or pain [174]. In most instances, surgical repair is not undertaken, but closing the perforation with a bridge flap, or more commonly, a button, is an option [176].

The decision as to whether to continue bevacizumab in a patient who has developed a nasal septal perforation must be individualized. In the published literature, a few patients continued with bevacizumab and seemed to do fine [171,177,178]. However, given the propensity of bevacizumab to delay wound healing, it seems prudent to wait for some evidence of perforation stability and healing prior to continuing the drug [174].

Sorafenib — Long-term treatment with sorafenib has been associated with muscle wasting and pancreatic atrophy.

Muscle wasting/sarcopenia — Cancer cachexia is characterized by diminished nutrient intake and progressive tissue depletion, both of which lead to weight loss. A disproportionate and excessive loss of lean body mass is the hallmark of cancer cachexia. (See "Pathogenesis, clinical features, and assessment of cancer cachexia", section on 'Changes in body composition'.)

Sarcopenia (skeletal muscle wasting) may also be an adverse effect of treatment with the antiangiogenic TKIs, particularly sorafenib:

●Sarcopenia was linked to dose-limiting toxicities in a phase I study of sorafenib in renal cell carcinoma (RCC) [179].

●In a subset analysis of 80 patients with RCC who were treated in the TARGET trial, body weight was measured and skeletal muscle mass was serially assessed by computed tomography (CT) [180]. At six months, sorafenib therapy was associated with a statistically significant decrease in total body weight and muscle mass compared with placebo (-2.1 versus + 0.8 kg and -7.4 versus -3.1 cm², respectively). The loss in weight and muscle mass in patients treated with sorafenib was progressive during treatment from 6 to 12 months.

●An associated between sarcopenia, sorafenib exposure, and dose limiting toxicity within the first month of treatment has also been seen in patients receiving sorafenib for hepatocellular carcinoma (HCC) [181].

The loss of muscle mass appears to be additive to that caused by advanced cancer and may contribute to asthenia and fatigue.

Whether or not sarcopenia represents a class effect of VEGF-targeted therapy is unclear. At least some data report muscle loss in patients treated with bevacizumab-containing chemotherapy for metastatic cancer, but there was no control group in either study, and the effects could have been attributable to the chemotherapy that was given in conjunction with bevacizumab or progression of the cancer itself [182,183].

Pancreatic atrophy — A single report describes two patients who developed irreversible pancreatic atrophy while on long-term treatment with sorafenib [184]. One patient developed intermittent diarrhea within three months of starting sorafenib, with remissions when treatment was interrupted and recurrence with rechallenge; pancreatic exocrine insufficiency was diagnosed 18 months after treatment initiation. The second developed diarrhea two months after treatment initiation that was responsive to pancreatic enzyme replacement and was shown to have a 35 percent decrease in the volume of the pancreas by CT by 37 months after treatment initiation. This complication has also been described with sunitinib [185].

Pazopanib and muscle pain — Myalgias and muscle spasms can occur in patients treated with pazopanib. In one study of 369 patients treated with pazopanib or placebo for advanced soft tissue sarcoma (STS), musculoskeletal pain of any grade occurred in 23 percent of pazopanib-treated patients versus 9 percent of the control group [186].

SUMMARY AND RECOMMENDATIONS

●General principles – Antiangiogenic agents (systemic medications that inhibit angiogenesis) have unique toxicities that differ from those of cytotoxic chemotherapy. Toxicities may be specific to the drug class or to the agent itself.

●Classification – The classes of antiangiogenic agents include:

•Ligand inhibitors – These agents bind to and inhibit ligand binding to the vascular endothelial growth factor (VEGF) receptor, thereby preventing activation of the receptor. Examples include bevacizumab, ramucirumab, and aflibercept.

•Receptor tyrosine kinase inhibitors (TKIs) – These agents block the intracellular domain of the VEGF receptor (VEGFR). Examples include sunitinib, sorafenib, pazopanib, vandetanib, cabozantinib, axitinib, ponatinib, lenvatinib, regorafenib, tivozanib, and fruquintinib.

●Selection of therapy – Many toxicities of antiangiogenic agents are serious and potentially fatal, particularly the cardiovascular effects and hemorrhage. Careful selection of therapy is based on patient characteristics (reasonable performance status, blood pressure control, and lack of serious cardiovascular events within six months), as well as close monitoring and prompt intervention are necessary to alleviate these risks. (See 'Risk of fatality' above.)

●Kidney toxicity – Nephrotoxicity of VEGF inhibitors is discussed separately. (See "Nephrotoxicity of molecularly targeted agents and immunotherapy", section on 'Antiangiogenic agents'.)

●Erythrocytosis – Erythrocytosis can occur with VEGF inhibitors. (See 'Erythrocytosis' above.)

●Hemorrhage – All VEGF-targeted agents have been associated with an increased risk of hemorrhage. This is most commonly grade 1 epistaxis, but serious and sometimes fatal hemorrhagic events, including hemoptysis (particularly in patients with squamous cell lung cancer), gastrointestinal bleeding, hematemesis, intracerebral hemorrhage, epistaxis, and vaginal bleeding, have occurred (risk of major bleeding approximately 2 to 3 percent). (See 'Bleeding' above.)

Because of the risk of massive hemoptysis, bevacizumab is contraindicated in patients with squamous cell lung carcinoma and in any patient with hemoptysis (>2.5 mL of blood) within three months. (See 'Pulmonary hemorrhage and cavitation' above.)

The risk of intracranial bleeding is low, even in patients with nonhemorrhagic brain metastases and recurrent glioma. Patients with a history of treated nonhemorrhagic brain metastases or a recurrent primary brain tumor need not be excluded from systemic therapy with a VEGF inhibitor. (See 'Intracranial bleeding' above.)

●Impaired wound healing – Bevacizumab and antiangiogenic TKIs have been associated with impaired wound healing in a variety of settings. At least 28 days (preferably six to eight weeks) should elapse between surgery and the first dose of bevacizumab (or the last dose of bevacizumab prior to surgery), when possible. (See 'Bevacizumab' above.)

If the clinical situation permits, all antiangiogenic TKIs should be interrupted for at least one week before surgery and not reinitiated until adequate wound healing has occurred, usually at least two to four weeks after major surgery.

●Gastrointestinal perforation – Although best described with bevacizumab, all VEGF-targeted therapies can cause gastrointestinal perforation (GIP) leading to peritonitis, fistula formation, or intraabdominal abscess. Clinicians should maintain a high index of suspicion for GIP in patients who develop acute abdominal pain while receiving bevacizumab or ramucirumab, even if they have no apparent risk factors. (See 'Intestinal perforation/fistula formation' above.)

●Osteonecrosis of the jaw – Isolated case reports have described medication-related osteonecrosis of the jaw (ONJ) with VEGF inhibitors, but the overall incidence appears low. (See 'Osteonecrosis of the jaw' above.)

●Reversible posterior leukoencephalopathy syndrome – Antiangiogenic therapy can cause reversible posterior leukoencephalopathy syndrome (RPLS), thought to be due to brain capillary leak and characterized by headaches, altered consciousness, visual disturbances, and seizures with or without hypertension. (See 'Reversible posterior leukoencephalopathy and brain capillary leak syndrome' above.)

●Fatigue – Fatigue is common with all VEGF-targeted agents; it is usually mild. (See 'Fatigue' above.)

●Dysphonia – Dysphonia has been reported with antiangiogenic TKIs and aflibercept but not bevacizumab. (See 'Dysphonia' above.)

●Thyroid dysfunction – Thyroid dysfunction is a class effect of antiangiogenic TKIs; it is most frequent with sunitinib and lenvatinib. (See 'Thyroid dysfunction' above.)

●Myelosuppression – Mild myelosuppression is a class effect of antiangiogenic TKIs and is seen most often with sunitinib and sorafenib because they target Flt-3 (FMS-related tyrosine kinase 3 receptor) and KIT, which are required for hematopoiesis. (See 'Myelosuppression' above.)

●Oral toxicity – Stomatitis may be seen with all the antiangiogenic TKIs, but it appears to be most common with sunitinib, sorafenib, and lenvatinib. (See 'Oral toxicity' above.)

●Gastrointestinal toxicity – Diarrhea, nausea, and emesis have been observed with all the antiangiogenic TKIs; rates of diarrhea are higher with vandetanib, probably because it also inhibits epidermal growth factor receptor (EGFR). (See 'Gastrointestinal toxicities' above.)

●Cutaneous toxicity – Cutaneous toxicity with antiangiogenic TKIs includes hand-foot skin reaction (especially with sorafenib), increased risk of keratoacanthomas and squamous cell carcinoma (with sunitinib or sorafenib), and acneiform rash (with vandetanib). (See 'Cutaneous toxicity' above.)

●Hepatoxicity – Sunitinib, ponatinib, sorafenib, lenvatinib, and especially pazopanib have been associated with severe and occasionally fatal hepatotoxicity. Patients treated with these agents should have a baseline evaluation of liver function tests (LFTs) and periodic re-evaluation (at least monthly) during therapy. (See 'Hepatotoxicity' above.)