This prospective cohort study was performed at Dr. Kariadi Hospital, the main teaching hospital for the Medical Faculty of Diponegoro University, Semarang, Indonesia. This hospital is a tertiary referral hospital for all patients with cancer in Central Java province. The endpoint of this study was objectively confirmed DVT within 3 months after first-line chemotherapy was given, without verification of either venous or arterial thrombosis at the time of enrollment. All patients were evaluated by the Wells’ score probability test and D-dimer level. Patients with a positive D-dimer level (≥500 ng/mL) and/or a high probability Wells’ score (≥2) were referred to undergo a vascular duplex ultrasound. Short-cycle chemotherapy was administered as outpatient therapy, while a protocol consisting of more than 2 days of chemotherapy was administered as inpatient therapy. None of the patients received primary prophylaxis with anticoagulation because the patients who were included in this study had an Eastern Cooperative Oncology Group (ECOG) performance status ≥2 to maintain their active mobilization. This study was approved by the Internal Review Board of Dr. Kariadi Hospital. Written informed consent was obtained from all subjects. This study was conducted in accordance with the Declaration of Helsinki.
Patient selection and data collection
From November 2016 to February 2017, a total of 246 consecutive and unselected newly diagnosed cancer patients were screened. Forty consecutive patients with active cancers undergoing chemotherapy were enrolled (see Figure 1 for patient enrollment). The inclusion criteria were as follows: newly diagnosed cancer with a histological confirmation, age over 18 years, ECOG performance status ≤ 2, consent to participate, and signed written informed consent. The exclusion criteria were as follows: overt bacterial or viral infection within the last two weeks, hepatic and renal dysfunction, venous or arterial thromboembolism within the last three months and the use of an anticoagulant, aspirin or statin and surgery or radiotherapy within the last two weeks. Inherited VTE risk factors were not taken into consideration. The probability test for Wells’ score (which will be described later) indicated “DVT unlikely” for all participants.
Before this study was conducted, all patients had been informed about the study’s details in an individual interview. Then, anamneses on patients’ cancer history, tumor site, tumor histology and tumor stage were documented. Patients who met the inclusion criteria were selected as study subjects. Samples were examined twice (before and after chemotherapy) to measure the plasma levels of sP-selectin, vWF:Ag and ADAMTS-13. Age, sex, smoking history, ABO blood group, body mass index, DVT-related history, diabetes, hypertension, history of atherosclerosis, drug, type of cancer, and chemotherapy regimen were carefully recorded.
Treatment and follow-up
All patients received chemotherapy or chemoradiotherapy. Cisplatin/carboplatin-based chemotherapy included gemcitabine, paclitaxel, docetaxel or pemetrexed. Fluorouracil-based chemotherapy was used in colorectal cancer patients as a component of the FOLFOX/FORFIRI/de Gramont protocol with or without bevacizumab or cetuximab. Anthracycline-based chemotherapy was used in several patients, including those with acute myeloid leukemia (AML) who received a 3+7 protocol containing daunorubicin or doxorubicin in the R-CHOP protocol. The patients were monitored for 3 months and evaluated during their follow-up visits at the Hematology and Medical Oncology Clinic by the attending staff. In the first, second and third months, evaluations of DVT occurrence were conducted clinically by examinations as well as the Wells’ pretest probability model to assess DVT. If the Wells’ score was ≥ 2, color duplex sonography was performed to establish the occurrence of DVT. Otherwise, color duplex sonography was conducted at the end of the third month.
Outcome measure: DVT
The endpoint of this study was the occurrence of DVT, either asymptomatic, symptomatic or fatal VTE, confirmed by a duplex ultrasound. Color duplex sonography was performed at the Radiology Department of Dr. Kariadi Hospital, Semarang, Indonesia. Patients with clinically suspected DVT and a Wells’ score ≥ 2 were assessed for DVT by using a Logiq 7 Pro US imaging system (Logiq 7 Pro; GE Healthcare, USA) with a 7-10 Hz linear probe. The diagnosis of DVT was based either on the presence of a noncompressible segment (compression ultrasound test – CUS) or flow impairment on color Doppler imaging. Patients were examined for both proximal (popliteal, femoral, and common femoral veins) and distal (peroneal and tibial veins) DVT. The first duplex ultrasound was performed within the first 7 days after inclusion and then during chemotherapy or if the Wells’ score was ≥ 2 or at the final observation in the third month.
Two prediction scores were used in this study: the Khorana risk score and Wells’ score. For each patient, we calculated the Khorana risk score to stratify the risk of VTE in cancer patients undergoing chemotherapy.22 Patients were assigned to three risk categories for VTE: low risk= 0, intermediate risk= 1-2, and high risk ≥ 3. A Wells’ score of 1 point each was given for active cancer, paralysis, paresis, recent plaster immobilization of the lower limb, recently bedridden for > 3 days, major surgery in the past 4 weeks, localized tenderness along the distribution of the deep venous system, entire leg swollen, calf swelling >3 cm compared to the asymptomatic leg, pitting edema and collateral superficial veins. Two points were subtracted from this score for an alternative diagnosis as likely DVT or more likely than DVT. A score of 3 or higher suggests that DVT is likely, the patient should receive a diagnostic US, and the results should be documented.23
Venous blood specimens were collected by sterile and atraumatic antecubital venipuncture into citrate vacutainer tubes (SST 5 mL) containing 0.5 mL of liquid anticoagulant. Measurements of plasma sP-selectin levels were carried out using a recombinant human P-selectin/CD62P immunoassay (catalog number ADP3; R&D Systems, Inc., 614 McKinley Place NE, Minneapolis, MN 55413, USA).24 vWF antigen was measured using an ELISA (catalog number 885_BcSD20121001; Sekisui Diagnostic, LLC, 500 West Avenue, Stamford).25 Plasma ADAMTS-13 levels were measured with a Quantikine ELISA human ADAMTS-13 immunoassay (R&D Systems, Inc., 614 McKinley Place NE, Minneapolis, MN 55413, USA).26
Blood samples were collected at the following time points: (i) baseline, before initial chemotherapy, and (ii) 3 months after initial chemotherapy. Samples for ELISA were immediately centrifuged at 2500 g for 15 minutes, and plasma samples were aliquoted, coded and stored at -80°C until the assays were performed. Samples were prepared according to the manufacturer’s instructions, and plates were read on an ELx808 plate reader (Biotek, Vermont) at a wavelength of 450 nm. The results were converted to total protein using a bicinchoninic acid (BCA) assay (Pierce Rockford, Illinois) and are reported as ng/mg total protein. Measurements were performed in a blinded manner. All samples were assayed in duplicate, and those showing values above the standard curve were retested with appropriate dilutions.
Cut-off points and normal reference values
sP-selectin, vWF:Ag and ADAMTS-13 activity cut-off points were determined based on fold changes. The minimum detectable dose for sP-selectin was 0.5 ng/mL, and the range in citrate plasma was 20 – 44 ng/mL.24 With regard to vWF:Ag, the reported mean level was 1.03±0.3 IU/mL in men and 1.08 ± 0.4 IU/mL in women.27 According to Green et al., the mean vWF:Ag/ADAMTS-13 ratio is 1.05 ± 0.30 IU/mL.28 The ADAMTS antigen level in noncancer patients was 0.70 – 1.42 IU/mL (median 1.08 IU/mL).29 According to the manufacturer, serum ADAMTS-13 levels range from 0.51 – 1.64 IU/mL (Quantikine ELISA Human ADAMTS Immunoassay, R&D Systems, Inc.).26 The cut-off values for sP-selectin, vWF:Ag, and ADAMTS-13 were set at 105.5 ng/mL, 2.35 IU/mL, and 1.03 IU/mL, respectively, according to the 75th percentiles of the levels observed in this cohort.
Response to chemotherapy and follow-up
All patients who underwent chemotherapy or chemoradiotherapy were followed up for 3 months. After obtaining informed consent, patients were evaluated either during routine visits at the Hematology and Medical Oncology Outpatient Clinic or at the medical ward every pre- and postchemotherapy cycle. The performance status, chemotherapy eligibility and Wells’ score were assessed at each visit. DVT occurring after enrollment was documented as a new event, first lifetime thrombosis. Median differences between the baseline and postchemotherapy levels of each independent variable were calculated and are reported as positive or negative delta values.
Quantitative variables were examined for normality with a Shapiro-Wilk test. Continuous variables are summarized as medians (minimum-maximum), whereas categorical data are described as absolute frequencies and percentages. Clinical and laboratory parameters, including sP-selectin, vWF:Ag, and ADAMTS-13 levels and their delta values between DVT and non-DVT subjects were compared using a nonparametric Mann-Whitney test. Correlations between two continuous variables were evaluated with Spearman’s rank correlation coefficient. To create positive and negative predictive values, we computed a logistic regression model with three independent variables together with other possible confounding factors. Dichotomous variables were created for all patients in our data set by comparing the probability of DVT associated with the selected cut-off point.
Stepwise multiple regression analysis was used to examine differences in sP-selectin, vWF:Ag, and ADAMTS-13 levels between baseline and postchemotherapy. The vWF:ADAMTS-13 ratio and other potential determinants, such as age, sex, smoking history, cardiovascular risk factors (such as overweight/obesity, hypertension and diabetes), chemotherapy regimen and other biological factors (Khorana risk score, D-dimer level, type of cancer and stage of disease) were also included. First, all potential predictors were entered simultaneously nto a multivariate logistic regression model that was reduced using a backward selection method as the final step. Multivariate logistic regression was adjusted for all independent predictors, and only variables with correlations to the outcome (defined as p<0.25 in the univariate model) were included. We generated the final multivariate model for DVT outcome using a backward stepwise approach, and p<0.05 from the likelihood ratio test was used to exclude excess factors. The precision of the specified model to detect DVT incidents was quantified by the Hosmer-Lemeshow goodness-of-fit statistic, where a value greater than 0.05 indicates adequate calibration for the corresponding area under the receiver operating characteristic (ROC) curve.30 The Statistical Package for Social Sciences (IBM v. 21; SPSS, Inc., USA) was used for all data analyses. All tests with p values < 0.05 were considered statistically significant.