Predictors for the Development of Thromboembolic Events in Cancer Patients Treated with Bevacizumab, Ramucirumab, and Aflibercept: A Single-Institution Retrospective Analysis

Abstract Introduction: The risk of thromboembolic events developing limits the dose of antiangiogenic agents, thereby reducing their efficacy. This retrospective study therefore sought to identify predictors for the development of antiangiogenic agent-induced thromboembolic events and to elucidate whether differences in the likelihood of thromboembolic events exist between different antiangiogenic agents or cancer types, to guide future strategies for optimizing safety, efficacy, and quality of life in patients receiving chemotherapy. Methods: This study retrospectively investigated 468 cancer patients who received chemotherapy with bevacizumab, ramucirumab, or aflibercept at our outpatient chemotherapy center between December 2016 and April 2022. Variables related to the development of thromboembolic events were extracted from the medical records, and multivariate logistic regression analysis was performed to identify predictors for the development of thromboembolic events. The Wilcoxon/Kruskal-Wallis test was used to detect significant differences between groups. Results: Significant factors included serum albumin level (odds ratio [OR] = 0.363, 95% confidence interval [CI] = 0.193–0.685; p = 0.0017) and diabetes mellitus (OR = 5.356, 95% CI = 1.711–16.769; p = 0.0039). Renin-angiotensin system inhibitors (OR = 0.307) had low OR, although it was not significant. No difference in the development of thromboembolic events was evident between cancer types (p = 0.0781), but differences were identified between the three antiangiogenic agents (p = 0.0132). Ramucirumab was associated with a lower likelihood of thromboembolic events. Conclusion: Serum albumin level and diabetes mellitus were identified as significant predictors for the development of antiangiogenic agent-induced thromboembolic events. In addition, the likelihood of thromboembolic events did not differ between cancer types but differed between antiangiogenic agents. Plain Language Summary This retrospective study was designed to identify predictors for the development of antiangiogenic agent-induced thromboembolic events and to elucidate whether differences in the likelihood of thromboembolic events exist between different antiangiogenic agents or cancer types. This study investigated 468 cancer patients who received chemotherapy with bevacizumab, ramucirumab, or aflibercept at our outpatient chemotherapy center between December 2016 and April 2022. Variables related to the development of thromboembolic events were extracted from the medical records, and multivariate logistic regression analysis was performed to identify predictors for the development of thromboembolic events. The Wilcoxon/Kruskal-Wallis test was used to detect significant differences between groups. Significant factors included serum albumin level (odds ratio [OR] = 0.363, 95% confidence interval [CI] = 0.193–0.685) and diabetes mellitus (OR = 5.356, 95% CI = 1.711–16.769). Renin-angiotensin system inhibitors (OR = 0.307) had low OR, although it was not significant. No difference in the development of thromboembolic events was evident between cancer types (p = 0.0781), but differences were identified between the three antiangiogenic agents (p = 0.0132). Ramucirumab was associated with a lower likelihood of thromboembolic events. In conclusion, serum albumin level and diabetes mellitus were identified as significant predictors for the development of antiangiogenic agent-induced thromboembolic events. In addition, the likelihood of thromboembolic events did not differ between cancer types, but differed between antiangiogenic agents.

related to the development of thromboembolic events were extracted from the medical records, and multivariate logistic regression analysis was performed to identify predictors for the development of thromboembolic events.The Wilcoxon/Kruskal-Wallis test was used to detect significant differences between groups.Results: Significant factors included serum albumin level (odds ratio [OR] = 0.363, 95% confidence interval [CI] = 0.193-0.685;p = 0.0017) and diabetes mellitus (OR = 5.356, 95% CI = 1.711-16.769;p = 0.0039).Reninangiotensin system inhibitors (OR = 0.307) had low OR, although it was not significant.No difference in the development of thromboembolic events was evident between cancer types (p = 0.0781), but differences were identified between the three antiangiogenic agents (p = 0.0132).Ramucirumab was associated with a lower likelihood of thromboembolic events.Conclusion: Serum albumin level and diabetes mellitus were identified as significant predictors for the development of antiangiogenic agent-induced thromboembolic events.In addition, the likelihood of thromboembolic events did not differ between cancer types but differed between antiangiogenic agents.

Introduction
Antiangiogenic agents targeting the human vascular endothelial growth factor (VEGF) pathway, such as bevacizumab, ramucirumab, and aflibercept, significantly improve overall survival and progression-free survival for patients with colorectal, lung, gastric, breast, and other cancers [1][2][3][4][5].The toxicity profiles of antiangiogenic agents differ from those of conventional cytotoxic anticancer agents, with the use of antiangiogenic agents commonly associated with increased blood pressure, thromboembolic events, and bleeding [6][7][8][9].The risk of thromboembolic events limits the dose of antiangiogenic agents, thereby reducing their efficacy [5][6][7][8][9][10].However, predictors for thromboembolic events have yet to be elucidated in detail, and few studies have examined differences in the development of thromboembolic events among antiangiogenic agents or cancer types [11].To help guide future strategies to improve the safety, efficacy, and quality of life of patients receiving chemotherapy, this retrospective study attempted to identify predictors for the development of antiangiogenic agent-induced thromboembolic events and to elucidate whether differences exist in the likelihood of thromboembolic events between antiangiogenic agents or cancer types.

Study Period and Participants
Between December 2016 and April 2022, a total of 473 patients with colorectal cancer, gastric cancer, ovarian cancer, breast cancer, lung cancer, or hepatocellular carcinoma received chemotherapy with antiangiogenic agents at our outpatient chemotherapy center.Exclusion criteria were patients with a history of thrombosis and patients with insufficient data.The Medical Ethics Review Committee of Kyoto Prefectural University of Medicine (approval no.ERB-C-867-6) approved this study.All procedures were performed in accordance with the ethical standards of the Medical Ethics Review Committee of Kyoto Prefectural University of Medicine and the 1964 Declaration of Helsinki and its subsequent amendments.No prospective human or animal studies were conducted by any of the authors for this article.Because of the retrospective nature of this work, the need to obtain informed consent from individual participants included in the study was waived in accordance with the ethical standards of the Medical Ethics Review Committee of Kyoto Prefectural University of Medicine.

Extraction of Variables
For the regression analysis of factors associated with antiangiogenic agent-induced thromboembolic events, variables were extracted by manual abstraction from clinical records.The variables evaluated included factors that were considered to potentially affect the development of thromboembolic events, comprising demographic data (sex, age, body mass index [BMI], and body surface area [BSA]), cancer type, adenocarcinoma or not, presence of metastases, presence of comorbidities, laboratory test values, administration of antiangiogenic agents, number of cycles, regimen, and medications that affect thromboembolic events.Creatinine clearance was estimated using the Cockcroft and Gault equation based on serum creatinine, sex, age, and weight.Clinical information was collected before the first dose of bevacizumab, ramucirumab, or aflibercept.

Statistical Analysis
This study used logistic regression analysis, with response Y as a binary categorical variable (onset of thromboembolic event = 1; no onset of thromboembolic event = 0) evaluated simultaneously with several predictors of thromboembolic event = X.Independent variables were analyzed for multicollinearity (correlation coefficient |r| ≥ 0.7) because correlations between variables can lead to unreliable and unstable results from regression analyses.First, univariate logistic regression analyses were performed between outcomes and each potential independent variable.A multivariate logistic regression model was then constructed using those candidate variables with values of p < 0.05 from univariate regression or selected based on previous studies [6][7][8][9].Multivariate logistic regression analysis was used.Receiver operating characteristic (ROC) curve analysis was used to determine optimal cutoff values.Further, we analyzed whether differences existed in the development of thromboembolic events between these three antiangiogenic agents or between cancer types.In addition, glycemic control, as measured by blood glucose levels and glycated hemoglobin, was evaluated both with and without diabetes mellitus (DM).The Wilcoxon/Kruskal-Wallis test was used to determine significant differences between groups.
For all statistical analyses, values of p < 0.05 (two-tailed) were considered significant.All statistical analyses were performed using JMP Pro ® version 16.2 (SAS Institute, Cary, NC, USA).

Results
Of the 473 patients who received chemotherapy with bevacizumab, ramucirumab, or aflibercept, 5 were excluded from the study: 3 patients with a history of thrombosis and 2 patients with insufficient data.Thromboembolic events occurred in 26 patients, comprising 24 patients with deep vein thrombosis (DVT), 1 with pulmonary embolism (PE), and 1 with cerebral infarction.
Table 1 shows the clinical characteristics of the 468 patients enrolled, the potential variables associated with the development of thromboembolic events, and the results of univariate analyses.D-dimer was a significant predictor in univariate analysis but was not used in the multivariate analysis due to the large number of missing data (n = 210).Multicollinearity was observed for BMI and BSA.BSA was used for analysis as the value is considered most clinically relevant for cancer chemotherapy.The selection of factors showing values of p < 0.05 in univariate regression or based on previous studies identified the following candidate variables: reninangiotensin system (RAS) inhibitors, albumin, and DM.
These variables were then used in a multivariate logistic regression analysis.Factors identified as significantly associated with the development of thromboembolic events included albumin (odds ratio = 0.363, 95% confidence interval = 0.193-0.685;p = 0.0017) and DM (odds ratio = 5.356, 95% confidence interval = 1.711-16.769;p = 0.0039) (Table 2).ROC curve analysis of the group most likely to develop thromboembolic events showed that the threshold albumin level ≤3.8 g/dL offered 39.7% sensitivity and 84.6% specificity (area under the curve = 0.65).
A difference in the risk of developing thromboembolic events was noted between antiangiogenic agents (bevacizumab, ramucirumab, and aflibercept; p = 0.0132), but no difference was apparent between cancer types (colorectal, gastric, lung, breast, ovarian, and liver cancers; p = 0.0781).There was a significant difference in glycemic control between the DM and non-DM groups (blood glucose; p = 0.0008, glycated hemoglobin; p = <0.0001).The DM group had significantly worse glycemic control.In addition, we showed a strong correlation between the presence of metastases and elevated d-dimer levels (p = 0.0316).

Discussion
The multivariate logistic regression analysis performed in this study showed that significant predictors for the development of thromboembolic events included serum albumin level and DM.In the ROC curve analysis of potential factors responsible for the development of thromboembolic events, the cutoff value for serum albumin level was ≤3.8 mg/ dL.This study also showed that the likelihood of thromboembolic events differed between antiangiogenic agents but not between cancer types.Univariate analysis showed that ramucirumab was associated with a lower likelihood of thromboembolic events, whereas bevacizumab was associated with a higher likelihood of thromboembolic events.D-dimer was also a predictor in univariate analysis.Although not significant, our study suggests that concomitant use of RAS inhibitors or calcium channel blockers may reduce the risk of developing thromboembolic events due to antiangiogenic agents.
In the current study, low serum albumin levels were extracted as a risk factor for thromboembolic events.Low serum albumin levels may indicate malnutrition.Patients with advanced cancer tend to be malnourished, and low serum albumin levels are a risk factor for the development of thromboembolic events, consistent with findings in routine clinical practice.Several studies have suggested that low serum albumin levels may represent a risk factor for the development of thromboembolic events such as DVT and PE [12][13][14][15].Low serum albumin levels have been suggested to contribute to a prothrombotic state by altering the balance of coagulation and fibrinolysis factors in the blood [13].Clinicians should therefore be aware of the propensity of patients with hypoalbuminemia to develop thromboembolic events, particularly among those with serum albumin levels ≤3.8 mg/dL.Previous studies have identified DM as a potential risk factor for the development of thromboembolic events such as DVT and PE [16].DM is thought to contribute to a prothrombotic state by increasing platelet activation, altering coagulation factors, and impairing fibrinolysis [16][17][18].In addition, DM is often associated with other cardiovascular risk factors, such as obesity, hypertension, and dyslipidemia, which may further increase the risk of thromboembolic events [19].Several studies have shown that individuals with DM are at increased risk of developing thromboembolic events [16][17][18][19].As a result, clinicians need to exercise caution when administering antiangiogenic agents, especially in patients with DM, and maintain good DM control to prevent the development of thromboembolic events.
This study suggests that RAS inhibitors or calcium channel blockers may prevent the development of thromboembolic events, although the result was not significant.Some studies have suggested that RAS inhibitors may have prophylactic effects against thromboembolic events, although the evidence is inconclusive [20,21].One mechanism by which RAS inhibitors may reduce the risk of thromboembolic events is by decreasing the production of angiotensin II, a hormone that constricts blood vessels and promotes inflammation.Angiotensin II has also been shown to increase platelet activation and aggregation, which can contribute to the development of thromboembolic events.By blocking the production of angiotensin II, RAS inhibitors may help reduce platelet activation and aggregation, as well as inflammation and endothelial dysfunction [22].Several observational studies have reported a lower incidence of thromboembolic events among patients treated with RAS inhibitors, including angiotensin-converting enzyme inhibitors and angiotensin II receptor blockers [20,21].RAS inhibitors, such as angiotensin-converting enzyme inhibitors and angiotensin II receptor blockers, may have protective effects against thromboembolic events in patients with DM or hypertension [20,21].However, these studies were observational, and further research is needed to confirm any associations between RAS inhibitors and reduced risk of thromboembolic events.Regarding calcium channel blockers, it was unclear from the literature review whether calcium channel blockers are effective in preventing thromboembolic events.Further validation will be performed in the future.
Previous studies have suggested that thromboembolic events induced by antiangiogenic agents are more likely to occur in lung cancer [23].However, the present study found no difference in the likelihood of thromboembolic events between cancer types.
The present investigation also showed that the likelihood of thromboembolic events differed among antiangiogenic agents.These results are consistent with previous studies [24].Bevacizumab is a monoclonal antibody that targets VEGF, which is involved in the regulation of angiogenesis and vascular permeability.By inhibiting VEGF signaling, bevacizumab may cause damage to the endothelial cells that line blood vessels, leading to an increased risk of clot formation and thromboembolic events.In contrast, ramucirumab is a monoclonal antibody that targets several proteins involved in angiogenesis and does not directly affect VEGF signaling.Therefore, ramucirumab does not carry the same risk of endothelial cell damage and thromboembolic events as bevacizumab.Overall, the mechanisms of action for bevacizumab and the resulting effects on endothelial cells and blood clot formation place bevacizumab-treated patients at higher risk of thromboembolic events than ramucirumab-treated patients.Clinicians should be particularly aware of the potential for thromboembolic events with bevacizumab among antiangiogenic agents [24,25].
D-dimer was also a predictor as determined by univariate analysis, in agreement with previous studies [26].This study showed a significant correlation between elevated d-dimer levels and the presence of metastases.Therefore, clinicians should be aware of elevated d-dimer levels, especially in advanced cancer patients with metastases, in order to prevent thromboembolic events.
The current study had several limitations.First, the retrospective design of the research may have decreased the reliability of the extracted data.Second, because this study was conducted at a single institution, a relatively small number of patients was analyzed.Therefore, a prospective multicenter study is needed to confirm these results.
In conclusion, serum albumin level and DM were identified as significant predictors for the development of thromboembolic events in cancer patients treated with bevacizumab, ramucirumab, or aflibercept.In addition, the likelihood of thromboembolic events did not differ between cancer types but did differ between antiangiogenic agents.However, these preliminary findings need to be confirmed in a larger randomized controlled trial.If cancer patients who may develop adverse effects could be readily identified, selection and/ or treatment with more appropriate doses of antiangiogenic agents could be performed in the future.Nevertheless, these findings may help develop chemotherapeutic strategies that can be used to improve the safety and efficacy of antiangiogenic agents, as well as the quality of life of patients.