Predictive Factors of Deep Vein Thrombosis in Gynecologic Cancer Survivors with Lower Extremity Edema: A Single-Center and Retrospective Study

This study was conducted to examine predictive factors of deep vein thrombosis (DVT) in gynecologic cancer survivors with lower extremity edema (LEE). In the current single-center, retrospective study, there was a total of 315 eligible patients, including 80 patients with DVT and 235 without DVT. They were therefore divided into two groups: the DVT group (n = 80) and the non-DVT group (n = 235). Then, baseline and clinical characteristics of the patients were compared between the two groups. In our study, distant organ metastasis, advanced stage, lymphadectomy, and amount of intraoperative blood loss had a positive predictive value for the occurrence of DVT in gynecologic cancer survivors presenting LEE. In conclusion, our results indicate that it is necessary to consider the possibility of LEE arising from DVT in gynecologic cancer survivors with advanced-stage cancer, distant organ metastasis, lymphadectomy, and intraoperative blood loss over 1500 mL.


Introduction
The relationship between venous thrombosis and malignancy was first described by Trousseau in 1865. Since then, it has been advocated by multiple clinical, pathologic, and laboratory studies [1,2]. According to Virchow, there is a triad of risk factors that contribute to venous thromboembolism; these include venous stasis, endothelial injury, and hypercoagulable states [3]. Patients with cancer are vulnerable to thrombosis arising from hematologic and biochemical abnormalities. For example, ovarian cancer cells are capable of forming and degrading thrombin. In addition, gynecologic malignancies are characterized by increased fibrinolytic activity. Furthermore, patients receiving surgery, chemotherapy, or radiotherapy are at increased risks of developing thrombosis [4,5].
According to epidemiological studies, patients with advanced-stage cancer are at increased risks of developing idiopathic venous thrombosis or thromboembolism. This deserves special attention [6].
In more detail, without prophylaxis, the incidence of VTE is estimated at approximately 10-40% [9]. With prophylaxis, it is estimated at 1.14% in patients diagnosed with gynecological disease, 0.7% in those undergoing laparoscopic gynecological surgery, 0.3% in those undergoing urogynecological surgery, and 4% in those with gynecological malignancies [10][11][12][13]. It remains problematic, however, that most published studies have evaluated symptomatic cases rather than asymptomatic ones as the Exclusion criteria for the current study were as follows: 1. Pregnant women (n = 0) 2. Women receiving any perioperative prophylaxis or anti-coagulation therapies (n = 15) 3. Women who were preoperatively diagnosed with DVT or PE (n = 10) 4. Women who were lost-to-follow-up (n = 25) 5. Women who are deemed to be ineligible for study participation according to our judgment (n = 0). The current study was approved by the Institutional Review Board (IRB) of our medical institution (KUCH 2019-08-027).

Patient Evaluation and Criteria
In our series, we performed a retrospective review of the medical records and thereby analyzed baseline and clinical characteristics of the patients. These include age, the type of malignancy (e.g., cervical cancer, endometrial cancer, and ovarian cancer), TNM stage at initial diagnosis, regional lymph node or distant organ metastasis, duration of disease (the time elapsing from the diagnosis of cancer to the evaluation of edema), BMI, co-morbidities (e.g., diabetes mellitus, hypertension, hyperlipidemia, atrial fibrillation, heart failure, and rheumatic disease), treatment modalities (e.g., surgery, chemotherapy, radiotherapy, and combination of more than two regimens), circumference of the lower extremity measured 10 cm above or below the upper border of the patella, and D-dimer levels.
The anatomical location of the DVT was divided into proximal (the inferior vena cava [IVC], iliac, femoral, and popliteal veins) or distal (the anterior, posterior tibial, peroneal, and muscular veins).
Depending on the presence of DVT on CT venography, the patients were divided into two groups: the DVT group and the non-DVT group ( Figure 1). Then, baseline and clinical characteristics of the patients were compared between the two groups. To consider the possibility of lower extremity edema arising from DVT, a CT venography was performed for (A) the common femoral vein, (B) ① the superficial femoral vein ② the deep femoral vein, (C) the popliteal vein, and (D) the popliteal vein. To consider the possibility of lower extremity edema arising from DVT, a CT venography was performed for (A) the common femoral vein, (B) 1 the superficial femoral vein 2 the deep femoral vein, (C) the popliteal vein, and (D) the popliteal vein.

Statistical Analysis
All data was expressed as mean ± standard deviation. Statistical analysis was done using SPSS 18.0 for Windows (SPSS Inc., Chicago, IL, USA). To compare the baseline and clinical characteristics of the patients between the two groups, we performed a Mann-Whitney U-test. In addition, we also performed the χ 2 -test to identify the correlations between categorical variables and the incidence of DVT. Furthermore, univariate and multiple logistic regression analyses were also performed to identify significant correlations between risk factors of developing DVT and adjusted or unadjusted variables. Their results were expressed as odds ratios (ORs) with 95% confidence intervals (95% CIs). A p-value of <0.05 was considered statistically significant.

Baseline Characteristics of the Patients
A total of 315 patients met inclusion/exclusion criteria. These include 80 patients with DVT and 235 without DVT. They were therefore divided into two groups: the DVT group (n = 80) and the non-DVT group (n = 235). The study flow chart is shown in Figure 2. In addition, baseline characteristics of the patients are represented in Tables 1 and 2. Healthcare 2020, 8, 48 4 of 12 All data was expressed as mean ± standard deviation. Statistical analysis was done using SPSS 18.0 for Windows (SPSS Inc., Chicago, IL, USA). To compare the baseline and clinical characteristics of the patients between the two groups, we performed a Mann-Whitney U-test. In addition, we also performed the χ 2 -test to identify the correlations between categorical variables and the incidence of DVT. Furthermore, univariate and multiple logistic regression analyses were also performed to identify significant correlations between risk factors of developing DVT and adjusted or unadjusted variables. Their results were expressed as odds ratios (ORs) with 95% confidence intervals (95% CIs). A p-value of <0.05 was considered statistically significant.

Baseline Characteristics of the Patients
A total of 315 patients met inclusion/exclusion criteria. These include 80 patients with DVT and 235 without DVT. They were therefore divided into two groups: the DVT group (n = 80) and the non-DVT group (n = 235). The study flow chart is shown in Figure 2. In addition, baseline characteristics of the patients are represented in Table 1 Table 3, the incidence of DVT had no significant correlation with the treatment modalities and co-morbidities. In addition, there were no significant differences in the circumference of the lower extremity, regional lymph node involvement, and D-dimer levels between the two groups. But distal organ metastasis, advanced-stage cancer, lymphadectomy, operation time ≥3 hours, and amount of intraoperative blood loss ≥1500 mL were significantly more prevalent in the DVT group as compared to the non-DVT group (p < 0.05).

Results of Univariate and Multivariate Analyses of Possible Predictive Factors
We performed both univariate and multivariate analyses of predictive factors, such as BMI, distant organ metastasis, advanced stage, lymphadectomy, operation time ≥3 hours, and amount of intraoperative blood loss ≥1500 mL, showing a significant difference between the DVT group and the non-DVT group. This showed that distant organ metastasis, advanced stage, lymphadectomy, and amount of intraoperative blood loss ≥1500 mL were found to be significant predictive factors (Table 4).

Discussion
VTE, presenting as DVT and PE, is a major cause of morbidity and mortality [30]. It is known that patients with malignancy are at a 6-fold greater risk of developing VTE and those with gynecologic cancer are at the greatest risk of developing VTE among all malignancies [31].
VTE is the second cause of mortality in patients with gynecologic cancer, and it has been reported that the risk of DVT and the incidence of PE were estimated at 17-40% and 1-26% in women undergoing gynecologic surgery [9,32].
It has been suggested that various risk factors are involved in cancer-related VTE [33]. Such risk factors are evaluated based on two well-known instruments. The Caprini risk assessment model (RAM), originally developed for surgical patients, aims to promote the derivation of risk factors of developing VTE. To do this, individual risk factors are summed and patients are divided into four categories accordingly: "low risk" (0-1 points), "moderate risk" (2 points), "high risk" (3-4 points), and "highest risk" (≥5 points) [34]. Moreover, Rogers Jr. et al. developed a predictive model of VTE through a logistic regression analysis of data obtained from the Patient Safety in Surgery (PSS) study, thus termed as the Rogers RAM. To do this, individual risk factors are summed and patients are divided into three categories accordingly: "low risk" (1-6 points), "moderate risk" (7-10 points), and "high risk" (>10 points) [35]. Blom et al. conducted a prospective study to estimate the incidence of VTE in 66,329 patients with cancer, thus reporting that it was 12.4/1000 patients within 6 months since the diagnosis of cancer and it was relatively higher when compared to normal healthy individuals [36]. Both thrombin formation arising from the pro-coagulant effects of tumor cells and venous compression leading to stasis are inevitable in patients with cancer [37,38]. Furthermore, cancer-treatment related factors, such as prolonged treatment period, immobilization, radiotherapy, and chemotherapy, may also raise the risk of thromboembolic events in patients with cancer [39].
The deep veins distributed in the lower extremities are classified into two categories: the proximal (the IVC, iliac, femoral, and popliteal veins) and the distal territory (the anterior, posterior tibial, peroneal, and muscular veins). Of note, proximal DVT is more frequently associated with PE and recurrence when compared to the distal one [40,41]. Our results showed that DVT occurred most frequently (31.3%) in the iliac vein. In association with this, Kahn et al. showed that thrombus was present in the common femoral vein and/or iliac vein in 25% of patients with symptomatic DVT of the lower extremities [42].
Our results showed that advanced-stage cancer was significantly more prevalent in the DVT group as compared to the non-DVT one. Presumably, this might be due to an increased immobility of the patients with advanced-stage cancer, which is in agreement with a previous report showing that patients with advanced-stage cancer are at increased risks of developing DVT [43].
In the current study, distant organ metastasis was significantly more prevalent in the DVT group as compared to the non-DVT one, as previously described [44]. But there was no significant difference in the incidence of regional lymph node metastasis between the two groups, which is not in agreement with a previous published study [45]. Involvement of metastases in the risk of DVT has been well described in the literature. Higher risks of DVT in association with distant or regional lymph node metastases may be explained by the process of metastatic dissemination. That is, there is a close association between the presence of metastasis and increased hypercoagulability because the hemostatic system might play an important role in metastatic capacity in malignancies. In more detail, intrusion of tumor cells into the blood or lymphatic fluid is an essential factor for distant metastasis. Distant metastasis is therefore followed by the interaction between tumor cells and the hemostatic system. This leads to the speculation that hypercoagulabulability may already exist in patients with regional spread of cancer [44,45].
We found that the presence of DVT had no significant correlation with the types of treatment modalities. This is in agreement with previous reports describing a lack of statistical significance in it [46,47].
In our series, despite a lack of statistical significance, the duration of cancer was shorter in the DVT group as compared to the non-DVT group. But this is not in agreement with previous published studies showing that risk factors of developing VTE include anatomical sites, biological characteristics, stage, and duration of cancer [48,49].
Lymphadenectomy is commonly used not only to assess lymph node status and the stage of gynecologic malignancies, but also to treat patients with gynecologic cancer [50,51]. But it is often accompanied by complications, such as hemorrhage, hematoma, and lymphocele [52,53]. Of these, lymphocele is one of the most common postoperative complications in that it leads to the occurrence of VTE by venous compression [54,55]. According to a review of the literature, VTE occurred after lymphadectomy at an estimated incidence of 0.8-25% [56][57][58]. This is also seen in our results. We found that lymphadectomy had a significant correlation with the occurrence of VTE.
A substantial amount of blood loss increases the risk of transfusion during perioperative period, and transfusion has been shown to be associated with the postoperative occurrence of VTE in gynecologic surgeries [59,60]. Our results also showed that the amount of intraoperative blood loss was a significant predictive factor of DVT. To summarize, distant organ metastasis, advanced stage, lymphadectomy, and amount of intraoperative blood loss had a positive predictive value for the occurrence of DVT in gynecologic cancer survivors presenting LEE. But our results cannot be generalized, because we retrospectively analyzed a small series of patients at a single, secondary medical institution. The possibility of selection bias could not therefore be completely ruled out. Further large-scale, multi-center studies are therefore warranted to establish our results.

Conclusions
It is such a potentially life-threatening condition that more than 90% of total patients with DVT of the lower extremities develop PE [61]. Clinicians should consider the possibility of LEE arising from DVT in these cases, which should be confirmed with imaging modalities such as CT venography.