Objectives Neoadjuvant chemotherapy may be considered for women with epithelial ovarian cancer who have poor performance status or a disease burden not amenable to primary cytoreductive surgery. Overlap exists between indications for neoadjuvant chemotherapy and known risk factors for venous thromboembolism, including impaired mobility, increasing age, and advanced malignancy. The objective of this study was to determine the rate of venous thromboembolism among women receiving neoadjuvant chemotherapy for epithelial ovarian cancer.
Methods A multi-institutional, observational study of patients receiving neoadjuvant chemotherapy for primary epithelial ovarian, fallopian tube, or peritoneal cancer was conducted. Primary outcome was rate of venous thromboembolism during neoadjuvant chemotherapy. Secondary outcomes included rates of venous thromboembolism at other stages of treatment (diagnosis, following interval debulking surgery, during adjuvant chemotherapy, or during treatment for recurrence) and associations between occurrence of venous thromboembolism during neoadjuvant chemotherapy, subject characteristics, and interval debulking outcomes. Venous thromboembolism was defined as deep vein thrombosis in the upper or lower extremities or in association with peripherally inserted central catheters or ports, pulmonary embolism, or concurrent deep vein thrombosis and pulmonary embolism. Both symptomatic and asymptomatic venous thromboembolism were reported.
Results A total of 230 patients receiving neoadjuvant chemotherapy were included; 63 (27%) patients overall experienced a venous thromboembolism. The primary outcome of venous thromboembolism during neoadjuvant chemotherapy occurred in 16 (7.7%) patients. Of the remaining venous thromboembolism events, 22 were at diagnosis (9.6%), six post-operatively (3%), five during adjuvant chemotherapy (3%), and 14 during treatment for recurrence (12%). Patients experiencing a venous thromboembolism during neoadjuvant chemotherapy had a longer mean time to interval debulking and were less likely to undergo optimal cytoreduction (50% vs 80.2%, p=0.02).
Conclusions Patients with advanced ovarian cancer are at high risk for venous thromboembolism while receiving neoadjuvant chemotherapy. Consideration of thromboprophylaxis may be warranted.
- ovarian cancer
- venous thromboembolism
Statistics from Altmetric.com
Venous thromboembolisms are common among women with ovarian cancer who receive neoadjuvant chemotherapy.
Most thromboembolisms were present at diagnosis, but the incidence during neoadjuvant treatment was 7.7%.
Thromboprophylaxis for women receiving neoadjuvant chemotherapy may be warranted.
Venous thromboembolism is a known potential complication of malignancy and is particularly common among women with epithelial ovarian cancer.1 2 In addition to having active cancer, women with epithelial ovarian cancer are predisposed to other well-established risk factors for venous thromboembolism, including impaired mobility, increasing age, and advanced disease.3 The risk of venous thromboembolism is even greater among patients with clear cell histology, thrombocytosis, high-grade tumors, and co-morbid cardiovascular risk factors (eg, obesity, hypertension, diabetes).4–7 Importantly, the occurrence of venous thromboembolism among patients with epithelial ovarian cancer is associated with increased morbidity, decreased survival, and increased healthcare costs.8–10
Among women with newly diagnosed epithelial ovarian cancer, significant overlap exists between factors promoting the development of venous thromboembolism and indications for consideration of treatment with neoadjuvant chemotherapy. Patients in whom optimal primary debulking is not possible or the risk of surgical complications is high may be offered neoadjuvant chemotherapy followed by interval debulking surgery rather than primary cytoreductive surgery;11 the goal of neoadjuvant chemotherapy is to improve the chance of safely achieving optimal cytoreduction in a high-risk population.12 13 Patients who receive neoadjuvant chemotherapy often have an advanced burden of disease (eg, carcinomatosis, extensive mesenteric involvement, distant metastases) and/or poor performance status at diagnosis, including inadequate nutrition, severe debilitation, or multiple co-morbidities.14
Given the overlap between risk factors associated with venous thromboembolism and indications for neoadjuvant chemotherapy among patients with epithelial ovarian cancer, we hypothesized that women receiving neoadjuvant chemotherapy would be at especially high risk for venous thromboembolism, which could potentially adversely impact patient outcomes. In this study, we sought to determine the rate of venous thromboembolism among patients receiving neoadjuvant chemotherapy for epithelial ovarian cancer and to identify characteristics and outcomes of interval debulking surgery associated with the occurrence of venous thromboembolism during neoadjuvant chemotherapy.
This was a multicenter observational study of patients receiving neoadjuvant chemotherapy for primary ovarian, fallopian tube, or peritoneal cancer at one of three academic institutions (Duke University, University of Oklahoma, West Virginia University). Neoadjuvant chemotherapy was defined as receipt of at least one cycle of chemotherapy prior to surgical debulking. Institutional review board approval was obtained at each site. Data collection at Duke was performed retrospectively from an existing database of patients who received neoadjuvant chemotherapy between January 01, 2000 and September 30, 2013. From October 01, 2013 to the present, data were collected prospectively with informed consent. An exception was made for women who received neoadjuvant chemotherapy after October 01, 2013 but died before informed consent could be obtained. At Oklahoma and West Virginia, patients receiving neoadjuvant chemotherapy from April 16, 2015 to November 27, 2017 and from January 01, 2008 to June 30, 2018, respectively, were identified retrospectively. Patients already on anti-coagulation (warfarin, novel oral anticoagulant, low molecular weight heparin, aspirin, or other) at the time of diagnosis were excluded.
Data were abstracted from electronic medical records at each institution. Study data were collected and managed using REDCap electronic capture data tools hosted at Duke University School of Medicine.15
The primary outcome in this study was rate of venous thromboembolism during neoadjuvant chemotherapy. Venous thromboembolism was defined as deep vein thrombosis in the upper or lower extremities or in association with peripherally inserted central catheters or ports, pulmonary embolism, or concurrent deep vein thrombosis and pulmonary embolism. Both symptomatic and asymptomatic venous thromboembolism were reported.
Secondary outcomes were the incidence of venous thromboembolism during other stages of treatment (including as a presenting symptom at diagnosis, post-operatively following interval debulking surgery (prior to resuming chemotherapy), during adjuvant chemotherapy (from re-initiation to completion of primary chemotherapy treatment), and during treatment for recurrence). The incidence of venous thromboembolism at diagnosis was calculated from the total study cohort, excluding subjects already on anticoagulation at the time of cancer diagnosis. Rates of venous thromboembolism at other time points (including the primary outcome during neoadjuvant chemotherapy) were calculated from the remaining subjects who had not yet had a venous thromboembolism and for whom the defined time interval was applicable with available data.
Patient characteristics and clinical factors reported include age, body mass index (BMI), race, Eastern Cooperative Oncology Group (ECOG) performance status, indication for neoadjuvant chemotherapy (disease volume, co-morbidities, or both), performance of diagnostic laparoscopy, primary tumor origin, tumor histology, tumor grade, and number of neoadjuvant chemotherapy cycles. Continuous variables are reported as mean±SD and compared using Student’s t-test. Categorical variables are described as frequency and percentage and compared using Fisher’s exact test.
Univariate binomial logistic regression analysis was performed to evaluate patient characteristics or clinical factors associated with occurrence of venous thromboembolism during neoadjuvant chemotherapy. Factors analyzed included: age ≥70 years, BMI ≥35 kg/m2, race, ECOG performance status, indication for neoadjuvant chemotherapy, pre-treatment CA125 levels ≥500 U/mL, diagnostic laparoscopy, and number of cycles of neoadjuvant chemotherapy (defined in tertile units of 0–3, 4–6, and 7–10). Primary tumor origin, histology, grade, and races other than Caucasian, non-Hispanic and African American were not included due to limited numbers of subjects with non-ovarian, non-high-grade serous, or other races, respectively.
For subjects who underwent interval debulking surgery and for whom data were available, time to interval debulking surgery and ability to achieve optimal debulking were reported. Optimal debulking was defined as residual tumor ≤1 cm (including no gross residual disease). The association between occurrence of venous thromboembolism during neoadjuvant chemotherapy and subsequent ability to achieve optimal debulking was analyzed using Fisher’s exact test and binomial logistic regression. Mean times from start of neoadjuvant chemotherapy to interval debulking were compared using an unpaired Student’s t-test.
All statistical analyses were performed in R Studio.12 A p value <0.05 was considered significant.
A total of 240 patients receiving neoadjuvant chemotherapy across three institutions were eligible for inclusion. Of these 240 patients, 175 (73%) patients were from Duke University, 44 (18%) patients from the University of Oklahoma, and 21 (8.7%) patients from West Virginia University. Most patients were diagnosed after 2011 (176/240, 73.3%). As shown in Figure 1, 10 patients were excluded due to use of anticoagulation at presentation, leaving 230 patients in the final study population. Patient characteristics are described in Table 1. Median patient age was 64.8 (range 34–84) years; median BMI was 27.2 (range 16–52) kg/m2.
Venous thromboembolism rates and associated stages of treatment are summarized in Figure 2. A total of 63 (27%) patients experienced a venous thromboembolism, with the primary outcome of venous thromboembolism during neoadjuvant chemotherapy occurring in 16 (7.7%) of the remaining 208 patients; eight were deep vein thromboses (including two catheter-associated events), four pulmonary embolisms, and four concurrent deep vein thromboses and pulmonary embolisms. Twenty-two (9.6%) venous thromboembolisms were present at the time of diagnosis. Characteristics of the patients stratified by venous thromboembolism occurrence are reported in Table 1.
Of the remaining patients for whom data were available, 185 patients underwent interval cytoreductive surgery. The incidence of post-operative venous thromboembolism (prior to initiation of adjuvant treatment) was 3% (6/185). Data regarding adjuvant treatment were available for 155 remaining patients, of whom five experienced a venous thromboembolism (5/155, 3%). Fourteen of the 113 remaining patients with documented treatment for recurrence had a venous thromboembolism event (14/113, 12%). One of the aforementioned patients had recurrent venous thromboembolisms, including a pulmonary embolism during neoadjuvant chemotherapy and subsequent pulmonary embolism during treatment for recurrence.
Table 1 further stratifies patient characteristics according to occurrence of venous thromboembolism during neoadjuvant chemotherapy. There was no significant difference in the frequencies of any patient characteristics delineated in Table 1, although there was a trend towards a higher incidence of venous thromboembolic events during neoadjuvant chemotherapy among African American patients. In addition, none of the clinical factors included in the univariate logistic regression analysis in Table 2 (age ≥70 years, BMI ≥35 kg/m2, race, ECOG performance status, pre-treatment CA125 ≥500 U/mL, diagnostic laparoscopy, indication for or number of cycles of neoadjuvant chemotherapy) were found to be predictive of the occurrence of venous thromboembolism during neoadjuvant chemotherapy.
Of patients who underwent interval cytoreductive surgery and for whom data on surgical outcomes were available (n=207), there was a significant association between occurrence of venous thromboembolism during neoadjuvant chemotherapy and failure to achieve optimal debulking at the time of interval cytoreductive surgery (50% vs 80.2%, respectively; OR 0.287, 95% CI 0.098 to 0.86). This result was also supported by descriptive data comparison (p=0.03, 95% CI 1.06 to 11.1) as shown in Table 1. Mean time from start of neoadjuvant chemotherapy to interval cytoreductive surgery (days) was also significantly longer for patients who experienced a venous thromboembolism during neoadjuvant chemotherapy (143.9 vs 105.5, p=0.02).
In this multi-institutional study of women receiving neoadjuvant chemotherapy for epithelial ovarian cancer, we describe a 9.6% prevalence of venous thromboembolism at cancer diagnosis, a 7.7% incidence of the primary outcome of venous thromboembolism during neoadjuvant chemotherapy, and a 6% incidence of venous thromboembolism during the remainder of the primary treatment course. The overall 27% incidence of venous thromboembolism observed in our study cohort was comparable to what has previously been reported in a single-institution study of women receiving neoadjuvant chemotherapy for epithelial ovarian cancer,16 but higher than the incidence among women with epithelial ovarian cancer in general.8 This is likely explained by the fact that patients who are offered neoadjuvant chemotherapy as an alternative to primary cytoreductive surgery typically have a more advanced burden of disease or additional co-morbidities that increase their baseline risk of venous thromboembolism.
Although most venous thromboembolisms were present at the time of diagnosis of the epithelial ovarian cancer, the incidence of venous thromboembolism during neoadjuvant chemotherapy for the remaining patients was high at 7.7%. This rate is comparable to the 7% described by Di Nisio et al in a 2018 systematic review of venous thromboembolism during neoadjuvant chemotherapy across all types of cancer17 but lower than the 11.6% previously reported specifically for patients receiving neoadjuvant chemotherapy for advanced epithelial ovarian cancer.16 18 Several characteristics of the latter study may help explain this difference: it was performed at a single institution with a smaller sample size, and the study period was prior to 2014, which precedes the interval described in the current investigation. As neoadjuvant chemotherapy has continued to gain greater acceptance as an alternative to primary cytoreductive surgery in advanced epithelial ovarian cancer,11 19 20 it may now be considered for a wider spectrum of patients than only those with the most severe disease burden or co-morbidities, thereby reducing the overall risk of venous thromboembolism within a more heterogeneous population.
None of the characteristics evaluated in this study had a significantly higher frequency among patients experiencing a venous thromboembolism during neoadjuvant chemotherapy as compared with those who did not. In addition, no clinical factor evaluated (including age ≥70 years, BMI ≥35 kg/m2, race, ECOG performance status, pre-treatment CA125 ≥500 U/mL, diagnostic laparoscopy, indication for or number of cycles of neoadjuvant chemotherapy) was predictive of venous thromboembolism during neoadjuvant chemotherapy. The power of these subgroup analyses was limited by the size of our overall study cohort and the small number of individuals experiencing a venous thromboembolic event. In addition, the relatively high overall incidence of venous thromboembolism during neoadjuvant chemotherapy observed in this study suggests that these patients could have a baseline risk of venous thromboembolism large enough to mask associations with additional risk factors.
After completion of neoadjuvant chemotherapy, interval cytoreduction surgery remains an important component of treatment for advanced epithelial ovarian cancer. We observed that the average time from start of neoadjuvant chemotherapy to interval cytoreduction was longer for patients who experienced a venous thromboembolism during neoadjuvant chemotherapy as compared with those who had not. Patients who experienced a venous thromboembolism during neoadjuvant chemotherapy were also less likely to have an optimal cytoreductive surgery at the time of interval surgery. Both of the aforementioned associations were evaluated as secondary outcomes; we therefore did not control for the many potential confounding factors and interpretation of these findings is limited. However, further investigation should be considered in future studies.
The high incidence of venous thromboembolism during neoadjuvant chemotherapy reported in this study contributes to a growing body of evidence suggesting that consideration of thromboprophylaxis during neoadjuvant chemotherapy is warranted. According to the current venous thromboembolism management guidelines from the American Society of Clinical Oncology (ASCO), thromboprophylaxis with either apixaban, rivaroxaban, or low molecular weight heparin is recommended for high-risk ambulatory patients with cancer receiving systemic chemotherapy.21 Per the guidelines, high-risk is defined as Khorana score ≥2, which is supported by a study by Carrier et al that demonstrated that chemoprophylaxis with apixaban among ambulatory cancer patients with an intermediate- to high-risk of venous thromboembolism (defined as Khorana score ≥2) significantly reduced the risk of venous thromboembolism (10.2% vs 4.2%).
The incidence of venous thromboembolism reported in the Carrier et al study was consistent with rates described previously for cohorts with Khorana score ≥2 and only slightly higher than the incidence described in our investigation.22 Notably a large portion of our patients would meet inclusion criteria for thromboprophylaxis per ASCO guidelines given that patients with gynecologic malignancies automatically receive one point (additional points are received with BMI >35 kg/m2, platelets >350 000/μL, hemoglobin <10 hgb or white blood cell count >11 000/ng). In addition, the incidence of venous thromboembolism during neoadjuvant chemotherapy reported for our study population slightly exceeds the incidence described for patients meeting ‘high-risk’ Khorana criteria in the original predictive model (Khorana score ≥3, 6.7% to 7.1% over a median of 2.5 months),7 which also supports the practice of thromboprophylaxis among women with epithelial ovarian cancer receiving neoadjuvant chemotherapy.
Discussions regarding the value of thromboprophylaxis must take into account costs associated with venous thromboembolic events as well as with thromboprophylactic medications. Costs attributable to venous thromboembolism among cancer patients have been well-documented: in a 2016 study, Cohoon et al reported that medical expenditures for cancer patients experiencing a venous thromboembolism were nearly two-fold that of matched controls, a finding that has been reproduced in the literature.23 24 These data alone suggest that initiating thromboprophylaxis during neoadjuvant chemotherapy has the potential to decrease costs for patients and the healthcare system overall. Publication of the ASCO guidelines detailed above may also favorably impact reimbursement for oral alternatives to low molecular weight heparin. A dedicated cost-effectiveness analysis is warranted.
Limitations of our study include the relatively small study cohort and, consequently, the small number of venous thromboembolism events identified. In addition, although every effort was made to account for all patients receiving neoadjuvant chemotherapy, this was primarily a retrospective study, which introduces the possibility that some patients were inadvertently omitted. We also did not control for confounding factors such as evolving practices of peri-operative thromboprophylaxis, presence of thrombocytosis, or bevacizumab use as these data were not collected as part of our existing dataset, and interpretation of secondary outcomes is therefore limited.
This is the first multi-institutional investigation of the incidence of venous thromboembolism during neoadjuvant chemotherapy among patients with advanced epithelial ovarian cancer. The relatively high incidence of venous thromboembolism observed in this study supports the importance of further evaluation of the role of thromboprophylaxis during neoadjuvant chemotherapy, especially in the context of the results reported by Carrier et al and current ASCO guidelines.
Contributors JRS: manuscript author, data abstraction and compilation, statistical analysis. KM: aided in development of neoadjuvant chemotherapy database at West Virginia University, assisted in data abstraction. MS: assisted in data abstraction for West Virginia University. RB: aided in development of neoadjuvant chemotherapy database at the University of Oklahoma, data abstraction. JE: assisted in development of neoadjuvant chemotherapy database at Duke University, data abstraction, patient consent. AML: maintenance of neoadjuvant chemotherapy database at Duke University, patient consent, coordination of data sharing among all three institutions. LJH: manuscript review and editing. AAS: leader of neoadjuvant chemotherapy database at Duke University, manuscript review and editing. VGT: leader of neoadjuvant chemotherapy database at West Virginia University. KNM: leader of neoadjuvant chemotherapy database at the University of Oklahoma. BAD: assisted in development of neoadjuvant chemotherapy database at Duke University, coordination of multiple institutions, data abstraction, manuscript review and editing.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient consent for publication Not required.
Provenance and peer review Not commissioned; externally peer reviewed.
Data availability statement Data are available upon reasonable request.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.