Grazyna Lieberman and William Novotny are employees of Genentech.
Circulating D-dimer levels are better predictors of overall survival and disease progression than carcinoembryonic antigen levels in patients with metastatic colorectal carcinoma
Article first published online: 28 MAY 2004
Copyright © 2004 American Cancer Society
Volume 101, Issue 1, pages 77–82, 1 July 2004
How to Cite
Blackwell, K., Hurẃitz, H., Liebérman, G., Novotny, W., Snyder, S., Dewhirst, M. and Greenberg, C. (2004), Circulating D-dimer levels are better predictors of overall survival and disease progression than carcinoembryonic antigen levels in patients with metastatic colorectal carcinoma. Cancer, 101: 77–82. doi: 10.1002/cncr.20336
- Issue published online: 18 JUN 2004
- Article first published online: 28 MAY 2004
- Manuscript Accepted: 9 APR 2004
- Manuscript Revised: 11 MAR 2004
- Manuscript Received: 11 DEC 2003
- Genentech, Inc., South San Francisco, CA
- colorectal carcinoma;
- carcinoembryonic antigen;
Fibrin formation is required for tumor angiogenesis, metastasis, and invasion. D-dimer, a fibrin degradation product, is produced when crosslinked fibrin is degraded by plasmin. The current study prospectively examined D-dimer levels in patients with metastatic colorectal carcinoma treated in a Phase II randomized trial comparing bevacizumab (Avastin, Genentech, South San Francisco, CA) plus 5-fluorouracil/leucovorin (5-FU/LV) with 5-FU/LV alone.
At least one circulating D-dimer level was evaluable in 98 of the 104 previously untreated patients with metastatic colorectal carcinoma in the current trial. Plasma D-dimer levels were determined using a quantitative immunoassay kit at enrollment, before each treatment, and at the time of trial completion or disease progression.
At trial enrollment, 86 of 104 patients (88%) had elevated D-dimer levels (> 20 ng/mL), and 86 of 102 patients (84%) had elevated carcinoembryonic antigen (CEA) levels (> 3 ng/mL). Baseline D-dimer levels were correlated with the following baseline characteristics: CEA (Pearson coefficient, 0.31; P = 0.002), albumin levels (Pearson coefficient, −0.32; P = 0.002), tumor burden (Pearson coefficient, 0.30; P = 0.003), and number of metastatic sites (Pearson coefficient, 0.21; P = 0.04). At the time of progression, plasma D-dimer levels reached a maximum postbaseline value in 51 of 61 patients (84%), whereas the CEA level was at its maximum postbaseline value in 39 of 55 patients (71%). Baseline D-dimer levels were a strong predictor of overall survival on univariate analysis (P = 0.008) and multivariate analysis (P = 0.03). Overall, treatment with bevacizumab (5 mg/kg) and baseline D-dimer levels were the only predictors of overall survival (P < 0.05).
The current study indicates that fibrin remodeling is an important prognostic feature in metastatic colorectal carcinoma. D-dimer levels should be incorporated into prognostic models, and D-dimer may represent a useful biomarker for patients treated with antiangiogenic agents. Cancer 2004. © 2004 American Cancer Society.
Fibrin turnover in the tumor extracellular matrix (ECM) is essential for tumor angiogenesis and growth.1–5 Crosslinked fibrin in the ECM serves as a stable framework for endothelial cell migration during angiogenesis and tumor cell migration during invasion. D-dimer, a fibrin degradation product, is produced when both intravascular and extravascular crosslinked fibrin is degraded by plasmin. Elevated D-dimer levels have been detected in patients with disseminated intravascular coagulation,6, 7 vasoocclusive crisis in sickle cell disease,8 thromboembolic events,8–10 and myocardial infarction.11
D-dimer levels also are elevated in patients with solid tumors, including lung,12–18 prostate,19–22 cervical,23, 24 and colorectal carcinoma.25–29 Plasma D-dimer levels have been shown to be directly correlated with other tumor markers, including CA 12530–34 and carcinoembryonic antigen (CEA).26, 35 In patients with colorectal carcinoma, D-dimer levels have been shown to be correlated with depth of tumor invasion and lymph node involvement at the time of surgical excision.26, 27
Because elevated D-dimer levels have been found in patients with colorectal carcinoma, and because fibrin degradation is important in tumor angiogenesis, the current study evaluated the correlations between quantitative D-dimer levels and response to chemotherapy, response to antiangiogenic therapy, CEA level, and patient outcome. We postulated that because the D-dimer assay could assess the process of fibrin formation and breakdown, it should serve as a surrogate marker for tumor growth and angiogenic activity. In addition, we evaluated the relation between elevated D-dimer levels and thromboembolic events in patients treated with chemotherapy and antiangiogenic therapy.
MATERIALS AND METHODS
Patients age > 18 years with histologically confirmed colorectal carcinoma and evidence of bidimensionally measurable disease were eligible for the treatment portion of the trial. Further eligibility criteria included Eastern Cooperative Oncology Group (ECOG) performance status 0 or 1 and a life expectancy of > 3 months. Patients with serious wounds, ulcers, fractures, or clinically significant cardiovascular or peripheral vascular disease were excluded, as were those who had undergone a major surgical procedure ≤ 28 days before Day 0. Recent or current use of oral and parenteral anticoagulants (except for the maintenance of central lines) or aspirin was not allowed. Written informed consent was required.
Clinical Study Design
The current study was the prospective correlative science portion of a Phase II randomized study in which eligible patients were assigned randomly to one of three treatment arms: one control arm (5-fluorouracil/leucovorin [5-FU/LV] alone) and two bevacizumab plus 5-FU/LV arms. The full trial design and statistical considerations have been published already.36 The multicenter clinical study comprised 36 sites. All patients received 5-FU/LV weekly for the first 6 weeks of an 8-week cycle according to the Roswell Park regimen: LV, 500 mg/m2, by 2-hour intravenous (i.v.) infusion once weekly for 6 weeks per cycle; and 5-FU, 500 mg/m2, by i.v. bolus (slow-push) 1 hour after initiation of the LV infusion. Patients continued to receive 5-FU/LV in subsequent cycles until disease progression, or for a total of six cycles, whichever occurred first. In addition to 5-FU/LV, patients in the two bevacizumab arms received bevacizumab (5 or 10 mg/kg) as a continuous 90-minute infusion every 2 weeks until disease progression or for up to 48 weeks, whichever occurred first. Patients in the control arm who experienced disease progression were given the option of receiving monotherapy with 10 mg/kg bevacizumab every 2 weeks. Baseline tumor assessments, with prospective identification of index lesions to be followed over the course of the study, included a chest X-ray, abdominal and pelvic computed tomographic scans, and measurement of serum CEA levels (Abbott Laboratories, IL). Serum CEA determination was performed on samples collected at the same time as the D-dimer samples, using published assay procedures. During treatment, tumor status, based on the ECOG tumor response criteria, was assessed at the completion of each 8-week cycle.
Plasma D-Dimer Antigen Measurements
At enrollment, before each cycle of chemotherapy, and at the completion of therapy (disease progression or end of study), 5 mL whole blood was collected from the antecubital vein of patients using a tourniquet, a 20-gauge Vacutainer needle, and a 3.8% sodium citrate Vacutainer collection tube (Becton Dickinson, Rutherford, NJ). All samples were centrifuged within 4 hours of venipuncture, and the plasma components were removed by pipetted and placed in snap-top plastic tubes. Centrifuged plasma samples were stored at −80 °C until assays were performed.
Quantitative D-dimer levels were measured at a central laboratory using the Dimertest immunoassay (American Diagnostica, Greenwich, CT). All samples were run in duplicate according to the manufacturer's recommendations. The commercially available D-dimer monoclonal antibody used recognizes an epitope that is a specific product of crosslinked fibrin that subsequently has been degraded by plasmin. Therefore, the plasma D-dimer assay does not recognize degradation of fibrinogen or noncrosslinked fibrin.
The primary end point for the current study was the evaluation of D-dimer in association with with progression-free and overall survival. Secondary end points included correlation of baseline D-dimer levels with the following variables: time to progression, baseline ECOG status, CEA level, baseline albumin level, presence of liver metastases, presence of lung metastases, and total tumor burden. Finally, the correlation between change in D-dimer levels and disease progression also was evaluated. D-dimer levels were evaluated prospectively as both a continuous variable (Spearman correlation analysis) and a dichotomous variable (≤ median value vs. > median value; Kaplan–Meier analysis).
In the analysis of changes in D-dimer and CEA levels and their associations with disease progression, the following steps were taken: First, D-dimer and CEA data within one cycle of disease progression were required to be available; 2) data from Day 0 in Cycles 2 and 3 were not included in calculating time to maximum postbaseline D-dimer value; Third, maximum D-dimer or CEA value reached within one cycle of disease progression was considered to be concordant with time to disease progression; 4) patients with D-dimer values ≤ 20 ng/mL were excluded (n = 3); and 5) patients with CEA values ≤ 3 ng/mL were excluded (n = 6).
Table 1 describes the patient population enrolled in the treatment portion of the trial. The results of the treatment portion have been reported previously.36 In brief, compared with 5-FU/LV alone, bevacizumab plus 5-FU/LV led to higher response rates, a longer median time to disease progression, and a longer median survival period. Patients had a median tumor burden (the sum of all bidimensional measurements) of 32.4 cm2. Thrombosis was the most common significant adverse event, occurring in 3 of 35 (9%) patients in the control arm, 9 of 35 (26%) patients in the 5 mg/kg bevacizumab arm, and 4 of 32 (13%) patients in the 10-mg/kg bevacizumab arm.
|Characteristics||% of patients|
|Median age in yrs (range)||61 (23–85)|
|No. of metastatic sites|
All 104 patients enrolled in the clinical trial underwent a pretreatment D-dimer evaluation, and 98 of these patients had more than one plasma sample available for determination of D-dimer levels. Two patients did not have enough pretreatment serum available for pretreatment CEA determination and were excluded from the analysis. Four hundred twenty- one D-dimer determinations were performed, and the average number of D-dimer determinations performed per patient was 4.24 (range, 1–8). The mean and median baseline D-dimer values for the population were elevated above normal (i.e., < 20 ng/mL), at 445.1 ng/mL (standard deviation, 751.9 ng/mL) and 133.2 ng/mL (interquartile range, 56–413.6), respectively. Changes in D-dimer levels were not correlated with the development or severity of thromboembolic events in 16 patients who experienced such events.
D-dimer levels were dichotomized at the median value of 133.2 ng/dL. D-dimer levels were not correlated with progression-free survival (Kaplan–Meier P = 0.12; Fig. 1), but they were strongly correlated with overall survival (Kaplan–Meier P = 0.008; Fig. 2).
Univariate analysis (Table 2) demonstrated a significant correlation between baseline D-dimer values and the following variables: CEA level (Pearson coefficient, 0.31; P = 0.002); albumin level (Pearson coefficient, −0.32; P = 0.002); tumor burden (Pearson coefficient, 0.30; P = 0.003); and number of distinct metastatic sites (Pearson coefficient, 0.21; P = 0.04). In addition, patients with liver metastases and patients with lung metastases had higher D-dimer values (P = 0.040 and 0.097, respectively). P = 0.097 was considered indicative of a trend.
|Variable||Mean log (baseline D-dimer level) ± SD||Pearson coefficient||P|
|Baseline ECOG PS||0.737|
|0||4.53 ± 0.26||—|
|1||4.91 ± 0.37||—|
|Absent||3.04 ± 0.53||—|
|Present||5.17 ± 0.21||—|
|Albumin level (g/dL)||0.383|
|≤ 3.5||5.11 ± 0.31||—|
|> 3.5||4.21 ± 0.29||—|
|Tumor burden (cm2)||0.446|
|≤ 32.4||4.36 ± 0.32||—|
|> 32.4||5.04 ± 0.29||—|
|Baseline albumin level||—||−0.315||0.0016|
|Tumor burden at baseline||—||0.296||0.0031|
|No. of metastatic sites at study entry||—||0.208||0.039|
Baseline D-dimer levels and treatment with bevacizumab were the only significant predictors of overall survival in a multivariate analysis that included CEA level, baseline albumin level, and ECOG performance status (Table 3).
|Prognostic variable||P value||Risk ratio (95% CI)|
|Baseline D-dimer level on log scale (≤ 133.2 ng/dL vs. > 133.2 ng/dL)||0.026||1.18 (1.020, 1.371)|
|Treatment with bevacizumab (5 mg/kg)||0.04||0.49 (0.251, 0.967)|
|Treatment with bevacizumab (10 mg/kg)||0.848||0.94 (0.494, 1.771)|
|Baseline ECOG PS (0 vs. 1)||0.342||1.35 (0.728, 2.493)|
|Baseline albumin level (≤ 3.5 g/dL vs. > 3.5 g/dL)||0.066||0.56 (0.304, 1.039)|
|Baseline CEA level on log scale (≤ 3 g/dL vs. > 3 g/dL)||0.761||1.02 (0.904, 1.147)|
Changes in both D-dimer levels and CEA levels and their correlation with disease progression and response were evaluated. Sixty-six patients in the study had documented disease progression during the study period. Of these 66 patients, 5 did not have either a baseline or time-of-progression D-dimer sample and were excluded from the analysis. Eleven patients did not have either a baseline or time-of-progression CEA sample and were also excluded from the analysis. Maximum D-dimer values since baseline occurred at the time of disease progression in 51 of 61 (84%) patients, whereas maximum CEA values since baseline occurred at the time of disease progression in 39 of 55 (71%) patients. Therefore, elevations in D-dimer values exhibited a stronger correlation with disease progression than did elevations in CEA levels (P = 0.01; Fig. 3). Among the 28 patients who demonstrated a documented response to therapy, there was no correlation between response to treatment and change in D-dimer level.
In the current Phase II study of patients with advanced colorectal carcinoma treated with standard 5-FU/LV chemotherapy with or without the anti–vascular endothelial growth factor agent bevacizumab, D-dimer levels were found to be correlated with disease burden, other prognostic variables, and overall survival. Previous studies have demonstrated similar correlations between D-dimer levels and prognosis for other tumor types, but the current prospective study is the first to evaluate the direct correlation between D-dimer levels and overall survival in colorectal carcinoma. More importantly, the current study is the first to examine the correlation of serial D-dimer values with changes in disease response or progression.
The process of metastasis and tumor growth requires a number of steps to occur within a favorable host environment. Malignant cells must first leave the primary tumor, migrate into the lymphovascular and circulatory systems, and establish a new blood supply at the metastatic site to survive. Fibrin remodeling is involved in many steps of metastasis and has been proven to play a crucial role in the formation of new vessels.37–44 Crosslinked fibrin in the ECM serves as a stable framework for endothelial cell migration during angiogenesis and for tumor cell migration during invasion. Knockout mouse models have also revealed the importance of fibrin remodeling in tumor growth and metastasis. Mice that are deficient in plasminogen (which is responsible for the degradation of fibrin) develop larger tumors, have more distant metastases, and have decreased life spans compared with mice with wild-type plasminogen.45 More recently, fibrinogen-deficient mice have been shown to develop fever metastases compared with mice with intact fibrinogen after an i.v. injection of either melanoma or lung carcinoma cells.46 The current study confirms previous studies that demonstrated up-regulated fibrinolytic activity (i.e., the presence of plasma D-dimer) in patients with metastatic disease. In addition, multivariate modeling showed a tight correlation between the presence of elevated D-dimer levels and the presence of visceral metastases. This relation may explain why D-dimer levels were found to be correlated so strongly with overall survival.
The current study demonstrated a discrepancy between the correlation of D-dimer levels with overall survival and the correlation of D-dimer progression levels with free survival. D-dimer levels were strongly correlated with overall survival but not with progression-free survival. This finding would be explainable if D-dimer levels were simply a marker of poor overall patient condition, rather than a marker of disease activity. However, two findings would argue against this hypothesis. First, we failed to show a correlation between D-dimer levels and any other marker of poor patient condition, including ECOG performance status and albumin levels. In addition, increases in D-dimer levels were correlated strongly with disease progression. In fact, changes in D-dimer levels were correlated more strongly with radiologically proven disease progression than with changes in CEA levels. In the future, other coagulation proteins, such as fibrinogen and plasminogen, should be examined to further define the biologic relation between the coagulation system and tumor burden/progression. Larger Phase III studies involving more patients and longer follow-up are underway to evaluate this relation, as well as to determine whether D-dimer levels can be used to predict response to antiangiogenic therapy. Until these studies are completed and further preclinical investigation is performed, it probably is premature to employ D-dimer, either alone or in combination with CEA, as a marker for colorectal carcinoma.
The current study underscores the importance of the tumor microenvironment with respect to growth, metastases, and response to therapy. Further studies evaluating the contribution of fibrin remodeling in cancer therapy are underway and should provide additional insight into potential therapeutic targets. The role of D-dimer as a prognostic marker is currently being studied prospectively in larger studies involving patients with either breast or colorectal carcinoma. It is hoped that D-dimer will serve as another useful tool for monitoring solid tumors, especially in patients undergoing therapy that targets the host environment.
- 23Pretreatment plasma levels of fibrinopeptide-A (FPA), D-dimer (DD), and von Willebrand factor (vWF) in patients with operable cervical cancer: influence of surgical-pathological stage, tumor size, histologic type, and lymph node status. Gynecol Oncol. 1993; 49: 354–358., , , et al.