Preoperative laboratory parameters associated with deep vein thrombosis in patients with ovarian cancer: retrospective analysis of 3,147 patients in a single institute

Abstract

Objective

Patients with ovarian cancer have a high risk of developing thrombosis. We aimed to investigate laboratory parameters associated with deep vein thrombosis (DVT) in patients treated for ovarian cancer.

Methods

We retrospectively analyzed pre-operation laboratory data of patients with ovarian cancer for DVT at the National Cancer Center, Korea, between January 2000 and February 2021. The test items were white blood cell count, absolute neutrophil count (ANC), hemoglobin, platelets, monocytes, serum glucose, CA125, D-dimer, fibrinogen, prothrombin time (PT), activated partial thromboplastin time (aPTT), and body mass index (BMI). Differences between patients with and without DVT were compared with Wilcoxon rank-sum test. We analyzed the variables using logistic regression. Items with significant odds ratios were included in multivariate logistic regression. Significant variables were selected using backward elimination. Items were further categorized based on reference ranges. Univariate and multivariate analyses were performed to identify items with abnormal values associated with DVT.

Results

From 3,147 patient samples analyzed, 286 (9.1%) patients with DVT were selected. Differences between patients with vs without DVT were statistically significant for hemoglobin, monocyte, serum glucose, CA125, PT, aPTT, fibrinogen, D-dimer, and BMI. After univariate and multivariate analysis, monocyte, glucose, and PT remained significant. Among the categorical variables, low hemoglobin, high monocyte, high CA125, prolonged PT, and high BMI remained significant after univariate and multivariate analysis.

Conclusion

Pre-operation laboratory data of low hemoglobin, high monocyte percentage, high serum glucose, high CA125, prolonged PT, and high BMI were associated with DVT.

Synopsis

We retrospectively analyzed pre-operative laboratory findings of patients with ovarian cancer who underwent surgery in a single institute and evaluated the difference between patients with thrombosis vs. patients without. Patients with thrombosis had higher percentage of monocytes, higher level of serum glucose, prolonged prothrombin time (PT), anemia, high level of CA125 and high level of body mass index.

INTRODUCTION

Venous thromboembolism (VTE), including deep vein thrombosis (DVT), is a leading secondary cause of death in patients with cancer [1]. Ovarian cancer is associated with one of the highest incidence of VTE among solid tumors [2]. Large tumors arising within the pelvis with accumulated ascites can compress the pelvic veins, leading to hemostasis and increased thrombosis risk [3]. Moreover, an advanced stage at diagnosis may contribute to VTE through cellular mechanisms including vessel wall irritation, inflammation, and thrombocytosis [3,4,5].

Numerous studies have investigated the risk factors for cancer-associated thrombosis and algorithms have been developed to calculate the risk factors [6] however, much of the data on thrombotic events in solid tumors are based on heterogeneous patient populations with multiple cancer types [7]. Additionally, many studies were mainly conducted in the United States and Europe. Several studies were conducted in Japan, but very limited elsewhere in Asia. This study, aimed to evaluate the laboratory findings of patients with ovarian cancer patients and DVT to compare them with those of patients without DVT in the same cancer group of the Korean population.

MATERIALS AND METHODS

Data were retrieved from the clinical research search portal of the National Cancer Center (NCC), Korea. Patients diagnosed with both ovarian cancer and DVT were identified through the search portal. We collected laboratory data before patients underwent surgery between January 2000 and February 2021. The evaluated test items were pre operation level of white blood cell count, absolute neutrophil count, hemoglobin, platelet count, monocyte percentage, serum glucose, CA125, prothrombin time (PT), activated partial thromboplastin time (aPTT), fibrinogen, D-dimer, and body mass index (BMI). All patients undergoing cytoreductive surgery were treated with prophylactic anticoagulants, preoperative single dose of low molecular weight heparin (Dalteparin) 2,500 IU and postoperative 2,500 IU dalteparin were injected every day for 5 days (patients treated from year 2000–2007) and 7 days (patients treated from year 2007 until now) and intermittent pneumatic compression devices (elastic stockings) were applied for 5 days after the operation to prevent thrombosis following the operation.

We compared the collected pre-operation laboratory data between patients with and without DVT using the Wilcoxon rank-sum test. We then analyzed the variables with univariate logistic regression. Items with significant odds ratios (ORs) were included in multivariate logistic regression, and significant variables were selected using backward elimination. The items were further categorized based on reference ranges. Univariate and multivariate analyses were performed to identify items with abnormal values associated with DVT. Statistical analyses were performed using R software version 3.6.3 (R Foundation for Statistical Computing, Vienna, Austria). This study was approved by the Institutional Review Board of the National Cancer Center in Korea (IRB No. NCC2021-0059)

RESULTS

Samples from 3,147 patients were analyzed, and 286 (9.1%) patients with DVT were selected based on past electronic medical records. Clinical characteristics are present in Table 1. The mean duration between the operation and onset of thrombosis was 5.3 months with median value of 2 months ranging from -8 months to 64 months (Fig. 1). 58.2% of thrombosis occurred within 3 months after the surgery. Differences in mean values between those with vs those without DVT were statistically significant for hemoglobin (10.3 g/dL vs. 10.8 g/dL, p<0.001), monocyte (6.7% vs. 6.3%, p=0.015), serum glucose (140 mg/dL vs. 125 mg/dL, p<0.001), CA125 (1,011 U/mL vs 671 U/mL, p<0.001), PT (15.7 seconds vs 14.9 seconds, p<0.001), aPTT (38.8 seconds vs. 37.4 seconds, p=0.001), fibrinogen (290 mg/dL vs. 308 mg/dL, p=0.003), D-dimer (3.36 ug/mL vs. 3.03 ug/mL, p=0.001) and BMI (23.6 vs. 23.3, p=0.018) (Table 2, Fig. 2).

Table 1. Clinical characteristics of patients.

Variables		Values
Pathology
	Serous carcinoma	2,037 (64.72)
	Mucinous carcinoma	210 (6.67)
	Endometrioid carcinoma	200 (6.36)
	Clear cell carcinoma	250 (7.94)
	Sex cord-stromal tumors	52 (1.65)
	Germ cell tumors	78 (2.48)
	Other specified malignant neoplasm	112 (3.56)
	Unknown	208 (6.61)
Age at diagnosis (yr)
	Mean±SD	52.50±12.90
	Median (min–max)	53 (5–93)

Values are presented as number (%) not otherwise specified.

Fig. 1. Duration between the cytoreductive surgery and onset of thrombosis.

Table 2. Comparison of mean values of pre operation test items between patients with DVT vs. without DVT.

Test items	No DVT (n=2,861)			DVT (n=286)			p-value*
	No.	Mean±SD	Median (min–max)	No.	Mean±SD	Median (min–max)
WBC (×103/uL)	2,760	6.46±2.74	6.07 (1.3–70.86)	270	6.33±2.39	5.80 (1.58–18.13)	0.276
ANC (/uL)	2,414	7,270.83±4,773.33	6,032 (400–50,898)	237	7,635.81±4,369.94	6,988 (1,020–26,241)	0.060
Hemoglobin (g/dL)	2,859	10.84±1.98	10.9 (0.1–16.1)	285	10.33±1.87	10.30 (5.10–15.60)	<0.001
PLT (×103/uL)	2,760	284.98±108.39	269 (42–906)	270	286.51±114.88	265.50 (53.0–714.0)	0.976
Monocyte (%)	2,848	6.26±2.69	5.9 (0.4–33.7)	282	6.71±3.15	6.25 (1.50–27.60)	0.015
Glucose (mg/dL)	2,836	125.39±43.89	108 (58–391)	283	140.29±51.74	124.00 (67.00–358.00)	<0.001
CA125 (U/mL)	2,658	670.81±2,384.25	83.9 (5.0–74,000)	267	1,011.38±2,499.20	183.00 (4.6–26,760)	<0.001
PT (sec)	2,846	14.90±2.89	14.2 (10.6–54.6)	284	15.76±3.1	15.20 (10.80–27.10)	<0.001
aPTT (sec)	2,844	37.36±9.29	35.8 (20.4–180)	284	38.81±8.83	37.40 (22.00–87.00)	0.001
Fibrinogen (mg/dL)	2,698	307.97±135.03	282 (34–957)	269	290.40±143.86	251.00 (85.0–835.0)	0.003
D-dimer (ug/mL)	1,521	3.03±2.91	2.38 (0.14–20.0)	189	3.36±2.37	2.82 (0.22–13.22)	0.001
BMI (kg/m2)	2,861	23.25±3.50	22.86 (14.23–43.12)	286	23.59±3.11	23.47 (15.09–34.45)	0.018

p<0.05 in bold.

ANC, absolute neutrophil count; aPTT, activated partial thromboplastin time; BMI, body mass index; DVT, deep vein thrombosis; No., number; WBC, white blood cell; PLT, platelet; PT, prothrombin time; SD, standard deviation.

^*Wilcoxon rank sum test.

Fig. 2. Comparison of mean values of pre operation test items between patients with DVT vs without DVT (A) glucose, (B) hemoglobin, (C) prothrombin time, (D) activated partial thromboplastin time € fibrinogen (F) monocyte percentage.

aPTT, activated partial thromboplastin time; DVT, deep vein thrombosis; Hb, hemoglobin; PT, prothrombin time; THROMBOSIS_YN N, no thrombosis; Y, thrombosis.

After univariate logistic regression analysis of continuous variables, significant ORs included hemoglobin (OR=0.879; 95% confidence interval [CI]=0.828–0.934; p<0.001), monocyte (OR=1.054; 95% CI=1.013–1.096; p=0.009), serum glucose (OR=1.006; 95% CI=1.004–1.009; p<0.001), CA125 (OR=1.000; 95% CI=1.000–1.000; p=0.050), PT (OR=1.085; 95% CI=1.047–1.124; p<0.001), aPTT (OR=1.013; 95% CI=1.003–1.024; p=0.014), and fibrinogen (OR=0.999; 95% CI=0.998–1.000; p=0.434) (Table 3). Multivariate analysis using significant variables selected after backward elimination found that monocyte (OR=1.065; 95% CI=1.023–1.108; p=0.002), serum glucose (OR=1.005; 95% CI=1.003–1.008; p=0.001), and PT (OR=1.065; 95% CI=1.025–1.107; p=0.001) were significantly associated with DVT. A receiver operating curve (ROC) was drawn using the remaining three factors, monocyte, serum glucose, and PT, to obtain an area under the curve (AUC) of 0.6253 (data not shown).

Table 3. Univariate and multivariate logistic regression analysis using continuous variables as factors affecting thrombosis.

Test items	No.	Event	Univariate		Multivariate (n=3,101/Event=280)
			OR (95% CI)	p-value	OR (95% CI)	p-value
WBC (×103/uL)	3,030	270	0.980 (0.931–1.031)	0.434
ANC (/uL)	2,651	237	1.000 (1.000–1.000)	0.258
Hemoglobin (g/dL)	3,144	285	0.879 (0.828–0.934)	<0.001
PLT (×103/uL)	3,030	270	1.000 (0.999–1.001)	0.825
Monocyte (%)	3,130	282	1.054 (1.013–1.096)	0.009	1.065 (1.023–1.108)	0.002
Glucose (mg/dL)	3,119	283	1.006 (1.004–1.009)	<0.001	1.005 (1.003–1.008)	0.001
CA125 (U/mL)	2,925	267	1.000 (1.000–1.000)	0.050
PT (sec)	3,130	284	1.085 (1.047–1.124)	<0.001	1.065 (1.025–1.107)	0.001
aPTT (sec)	3,128	284	1.013 (1.003–1.024)	0.014
Fibrinogen (mg/dL)	2,967	269	0.999 (0.998–1.000)	0.043
D-dimer (ug/mL)	1,710	189	1.037 (0.989–1.087)	0.131
BMI (kg/m2)	3,147	286	1.028 (0.993–1.063)	0.118

p<0.05 in bold.

ANC, absolute neutrophil count; aPTT, activated partial thromboplastin time; BMI, body mass index; CI, confidence interval; DVT, deep vein thrombosis; No., number; OR, odds ratio; PLT, platelet; PT, prothrombin time; WBC, white blood cell.

When the variables were categorized based on reference range, univariate logistic regression analysis showed that decreased hemoglobin (p<0.001), increased monocyte count (p=0.028), increased serum glucose (p<0.001), increased CA125 (p=0.001), prolonged PT (p<0.001) and aPTT (p=0.001), decreased fibrinogen (p=0.001), and increased BMI (p=0.040) were associated with DVT. Multivariate analysis from significant variables selected based on backward elimination method found that decreased hemoglobin (p=0.004), increased monocyte count (p=0.014), increased CA125 level (p=0.003), prolonged PT (p=0.001), and increased BMI (p=0.026) were significant factors associated with DVT (Table 4). ROC was drawn using the remaining five factors, hemoglobin, monocyte, CA125, PT, and BMI, to obtain an AUC of 0.6376 (data not shown). Since the two groups were largely unequal in size, propensity matching was done with pathology and age at diagnosis as correction variables (Tables S1, S2, S3).

Table 4. Univariate and multivariate logistic regression analysis using categorial variables based on reference range as factors affecting thrombosis.

Variables		No.	Event	Univariate		Multivariate (n=2,900/Event=263)
				OR (95% CI)	p-value	OR (95% CI)	p-value
WBC (×103/uL)
	<10	2,573	248	1 (ref)
	≥10	187	22	1.221 (0.770–1.935)	0.396
ANC (/uL)
	<7,500	1,451	129	1 (ref)
	≥7,500	963	108	1.261 (0.965–1.650)	0.090
Hb (g/dL)
	<12	1,975	235	1 (ref)		1 (ref)
	≥12	884	50	0.475 (0.347–0.652)	<0.001	0.592 (0.413–0.848)	0.004
PLT (×103/uL)
	<450	2,503	235	1 (ref)
	≥450	345	47	1.427 (0.941–2.163)	0.094
Monocyte (%)
	<9	2,503	235	1 (ref)		1 (ref)
	≥9	345	47	1.451 (1.040–2.024)	0.028	1.558 (1.093–2.222)	0.014
Glucose (mg/dL)
	<110	1,466	111	1 (ref)
	≥110	1,370	172	1.658 (1.292–2.129)	<0.001
CA125 (U/mL)
	<35	830	53	1 (ref)		1 (ref)
	≥35	1,828	214	1.833 (1.342–2.504)	0.001	1.621 (1.174–2.239)	0.003
PT (sec)
	<14.9	1,689	123	1 (ref)		1 (ref)
	≥14.9	1,157	161	1.911 (1.494–2.445)	<0.001	1.704 (1.290–2.251)	0.001
aPTT (sec)
	<45	2,499	229	1 (ref)
	≥45	345	55	1.740 (1.269–2.384)	0.001
Fibrinogen (mg/dL)
	200–400	1,575	132	1 (ref)	(0.001)
	<200	591	85	1.716 (1.286–2.290)	0.001
	≥400	532	52	1.166 (0.834–1.631)	0.369
BMI (kg/m2)
	<22.9	1,443	126	1 (ref)		1 (ref)
	≥22.9	1,418	160	1.292 (1.012–1.650)	0.040	1.340 (1.035–1.735)	0.026

p<0.05 in bold.

ANC, absolute neutrophil count; aPTT, activated partial thromboplastin time; BMI, body mass index; CI, confidence interval; No., number; OR, odds ratio; PLT, platelet; PT, prothrombin time; WBC, white blood cell.

DISCUSSION

In our study, mean hemoglobin level was significantly lower in patients with DVT. Several studies have reported the relationship between iron deficiency anemia (IDA) and thrombosis. Here, concurrent thrombocytosis is frequently referred to as the major associated factor [8,9,10] however, platelet number did not differ between the two groups. When we evaluated the data aside from pre operation data on several occasions (data not shown), platelets were significantly increased in the DVT group. The mechanism of action between IDA and thrombosis is unknown. One study investigated the levels of cytokines, thrombopoietin, erythropoietin, interleukin (IL)-6, IL-11, and leukemia inhibitory factors in patients with IDA. Here, only elevated levels of erythropoietin were correlated with thrombocytosis [11].

In contrast, anemia is associated with the functional aspect of platelets. In one study, anemic blood samples showed increased platelet aggregation velocity, maximum clot firmness, and shortened clot formation time [12]. In another study, anemia was associated with a delay in initiating the coagulation cascade of a final clot with superior strength and viscoelastic properties. The authors explained that in anemia, the initial rate of fibrin formation is delayed until a certain amount of thrombus was generated. From that point on, the coagulation profile was superior in terms of a more rapid fibrin build-up with a stronger clot and better viscoelastic properties [13]. This may explain the relationship between anemia and thrombus formation.

In another study, the authors suggested that anemia may reflect underlying conditions related to thromboembolism, such as nutritional deficiency [14,15], chronic inflammation [16,17,18], cancer, and chemotherapy [6,19]. Moreover, they suggested that anemia may predispose to endothelial dysfunction [20], blood stasis [21] and/or a hypercoagulable state [22] which leads to a greater risk of thrombus formation.

In our study, increased monocyte levels remained significant after multivariate logistic regression analysis using both continuous and categorical variables. A previous study reported that monocytosis alone can act as an independent variable for predicting DVT [23] because tissue factor is strongly present on monocytes. This activates the coagulation cascade, where monocytes and neutrophils are the first circulating cells seen in DVT formation [24]. Although the aforementioned study evaluated the percentage of monocytes in patients who were admitted for DVT with a high OR (9.35; 95% CI=3.20–27.30) [23], our data suggest that patients prone to DVT had continuously high percentage of monocytes from the pre operation time point.

In this study, patients with ovarian cancer and DVT had higher serum glucose levels. Diabetes mellitus (DM) is commonly associated with both micro- and macrovascular complications [25]. Its pathogenesis can be explained by endothelial dysfunction, coagulation activation, microparticles, and platelet activation [26]. The hyperglycemic environment generates advanced glycation end products, which accumulate in the vessel walls, directly affecting cell structure and function [27,28]. Hyperglycemia induces platelets to generate thrombin. This is impaired by increased levels of plasminogen activator inhibitor-1 from insulin resistance. Low-grade inflammation that increases the circulating levels of IL-6, fibrinogen, and tissue factors all lead to a hypercoagulable state [29,30,31]. Platelets from patients with type 2 DM produce more tissue factors than platelets from matched control subjects [32]. Platelets of patients with DM are hyperreactive with enhanced adhesion, aggregation, and activation. These result from dysregulation of several signaling pathways [33] thereby leading to thrombus formation.

Ovarian cancer is a bulky malignant tumor of the peritoneal cavity and we can assume that there is a high risk of thromboembolism in such patients. In patients with ovarian cancer, a massive tumor forms in the pelvis. Subsequently, venous flow in the lower extremities stagnates and leads to a large volume of ascites, causing intravascular hypovolemia and thrombi formation [34,35]. Our data showed that high CA125 level, a tumor marker representing tumor volume, was related to DVT. This shows that larger tumor volumes are more likely to form tumor thrombi.

In a risk model developed by Khorana et al., patients with a BMI of 35 kg/m² or more had an OR of 2.5 (95% CI=1.3–4.7) and were included in the risk scoring system [6]. In our population, higher BMI was associated with DVT. Obesity is a known risk factor for VTE in the general population [36] but it has not been studied specifically in cancer patients. None of the patients in our study had a BMI >35 kg/m². This questions the applicability of the BMI criteria in the risk scoring system (6) in the Korean population.

In our study, the mean preoperative D-dimer level was significantly increased in patients with DVT. D-dimer is a specific degradation product of cross-linked fibrin breakdown by plasmin. High D-dimer levels reflect increased fibrin formation and an efficient fibrinolytic system. Several studies have emphasized the clinical utility of D-dimer cutoff level as a predictive marker of DVT in patients with ovarian cancer, presuming that a large number of patients have already developed VTE prior to cancer treatment [37]. However, in our study, D-dimer was not significant after logistic regression analysis.

In our study, prolonged PT was associated with DVT in multivariate logistic regression analysis using both continuous and categorical variables. In univariate analysis, aPTT was also prolonged in the DVT group, which allowed us to evaluate situations where both PT and aPTT were prolonged. These include common pathway factor deficiencies or inhibitors, warfarin, vitamin K deficiency, disseminated intravascular coagulation, liver disease, direct thrombin inhibitors, direct Xa inhibitors, lupus anticoagulant, and heparin use [38]. Patients with DVT had low fibrinogen levels in univariate analysis using categorical variables. In thrombotic settings, fibrinogen has many roles. These include forming fibrin, providing binding sites for plasma proteins involved in clot formation (thrombin), and binding to monocytes to enhance monocyte activation; thus resulting in decreased levels of fibrinogen [39]. Although the data were collected before the operation, the time point of DVT occurrence was not considered in the analysis; hence, prior anticoagulant treatment and thrombosis may be confounders, which is also a limitation of this study.

In conclusion, pre-operation laboratory data of low hemoglobin, and high monocyte percentage, serum glucose, CA125, prolonged PT, and BMI were associated with DVT. Although the patients were treated with prophylactic anticoagulants and compression stockings to prevent DVT, patients still developed thrombosis, which suggests that careful observation and follow-up are needed in patients showing a significant combination of related pre operation data which reflects patients with high tumor volume, diabetes mellitus, increased BMI, increased percentage of monocytes and prolonged PT.

SUPPLEMENTARY MATERIALS

Table S1

Age at diagnosis and pathology were used as correction variables for propensity score matching

Table S2

Method 1: Results using propensity score matching data (1:2) with age and pathological variables as matching variables

Table S3

Method 2: Adjusted pathology and age at diagnosis to multivariable model