Early venous thromboembolism is a strong prognostic factor in patients with advanced pancreatic ductal adenocarcinoma

There are still controversial data regarding the prognostic value of Venous ThromboEmbolism (VTE) in advanced Pancreatic Ductal AdenoCarcinoma (PDAC) and thromboprophylaxis is poorly prescribed despite international recommendations. Medical charts of patients consecutively treated for advanced PDAC from 2010 to 2019 were retrospectively reviewed. Progression-free survival (PFS) and overall survival (OS) were estimated using Kaplan–Meier method. Prognostic Factors were identified using a multivariate Cox’s proportional hazard model. Early VTE was defined as VTE occurring within the three months following the PDAC diagnosis. A total of 174 patients were included (median age: 67 years; males: 55.2%; performance status (PS) 0–1: 88.5%) with metastatic disease in 74.7%. At baseline, Khorana score was high (≥ 3) in the vast majority of cases (93.7%). The cumulative incidences of VTE were 12.4% (95% CI 7.3–17.2) at 3 months, 20.4% (95% CI 13.9–26.4) at 6 months and 28.1% (95% CI 20.0–35.3) at 12 months. Patients who experienced early VTE had shorter PFS (3.8 months vs. 7.1 months; HR = 2.02; 95% CI 1.21–3.37; p = 0.006) and shorter OS (8.0 months vs. 14.1 months; HR = 2.42; 95% CI 1.37–4.30; p = 0.002) compared to the others, independently of prognostic factors such as PS, liver metastases, carcinomatosis, and chemotherapy regimen. early VTE is a strong prognostic factor in advanced PDAC and occurs in about one in 10 patients.


Introduction
The risk of Venous ThromboEmbolism (VTE) is increased in cancer patients due to a prothrombotic and proangiogenic state involving numerous mechanisms such as tumor overexpression of tissue factor (Khorana and Fine 2004). Chemotherapy, distant metastases and the primary cancer site are additional risk factors for VTE (Blom et al. 2006).
Pancreatic ductal adenocarcinoma (PDAC) is one of the cancers with a higher risk of VTE (Blom et al. 2006). In the majority of PDAC cohorts, the prevalence of VTE ranges from 10 to 30% (Epstein et al. 2012;Lee et al. 2016;Frere et al. 2019). However, VTE can occur at various times during the disease. Hence, the cumulative incidence of VTE appears more informative than the prevalence but has been less frequently reported in the literature (Lee et al. 2016;Frere et al. 2019;Gade et al. 2017Gade et al. , 2020Godinho et al. 2020;Chen et al. 2018).
The Khorana score was developed and validated in 2008 (Khorana et al. 2008) as a predictive model for chemotherapy-associated thrombosis. This score takes into account the primary cancer site, platelets count, hemoglobin level, use of erythropoiesis-stimulating agents, leucocytes count, and body mass index (BMI) (Khorana et al. 2008). All patients with PDAC have a Khorana score of at least 2 points (moderate risk) and a significant proportion of them have a high risk of VTE (score ≥ 3) and should undergoing primary thromboprophylaxis, as recommended since 2013 (Grade 1B evidence level).
However, the benefit of thromboprophylaxis in these patients continues to be under-recognized worldwide. Indeed, except in the perioperative setting and in hospitalized patients, ambulatory thromboprophylaxis remains underused because of high rates of VTE recurrence, and potential bleeding complications during VTE treatment (Key et al. 2020). Due to their short life expectancy, there is no strong evidence that inhibition of the coagulation cascade improves survival in patients with advanced PDAC (Frere et al. 2019). Moreover, there are controversial data regarding the prognostic value of VTE in PDAC (Epstein et al. 2012;Lee et al. 2016;Frere et al. 2019;Chen et al. 2018;Mandalà et al. 2007). Given VTE is a time-dependent factor, there is growing evidence that early-VTE predicts survivals (Chen et al. 2018;Mandalà et al. 2007;Kim et al. 2018). Therefore, this retrospective cohort study was conducted to assess the impact of early VTE on survivals in advanced PDAC.

Study design and objectives
All consecutive patients treated for PDAC from 2010 to 2019 at our University Hospital were reviewed. Until June 2018, patients were identified by using the International Classifications of Diseases (CIM10). Their data were collected from their electronic charts in agreement with the Declaration of Helsinki. The Ethics Committee of our hospital approved the study protocol (IRBN632017/CHUSTE). From June 2018 to June 2019, this retrospective cohort was pooled with an observational prospective cohort carried-out in the same centre (IRBN362018/CHUSTE).
The primary objective of this study was to assess the prognostic value of VTE and early VTE in advanced PDAC. Secondary objective was to report the cumulative incidence of VTE in patients with advanced PDAC. Inclusion criteria were patients with advanced PDAC at baseline (locally advanced or metastatic disease) or in recurrence after tumor resection.

Data collection and definitions
Patient data collection included demographic characteristics (age, sex, body mass index [BMI]), comorbidities (diabetes, tobacco consumption), the performance status (PS) using the Eastern Cooperative Oncology Group Performance Status Scale (ECOG-PS), and the nutritional status (albumin level).
Regarding disease characteristics, the following data were collected: primary tumor location, disease stage (locally advanced, metastatic or in recurrence), CA 19-9 level at baseline, and the first-line chemotherapy regimen. Baseline was defined as the time of diagnosis of advanced disease or as the time of disease recurrence after tumor resection.
Progression-free survival (PFS) was defined from the date of baseline to disease-progression [according to Response Evaluation Criteria in Solid Tumors (RECIST v1) criteria], or to death or loss to follow-up. Overall survival (OS) was defined from the date of baseline to death or loss to follow-up.
Regarding VTE events, the following data were collected: date of VTE occurrence, location of the thrombosis (visceral, extra-visceral), symptomatic character, treatment duration and response to the treatment. Time-to-progression and Time-to-death were defined from the date of VTE occurrence to the date of disease progression or death, respectively, or loss to follow-up. Early VTE was defined as the occurrence of VTE within the three months following baseline.

Statistical analysis
Survivals (PFS, OS) and time from VTE to the event (disease-progression, death) were estimated using Kaplan Meier Method. Prognostic factors were identified using Cox's proportional hazards regression. Multivariate analyses were performed by selecting variables according their corresponding p value (≤ 0.05) in univariate analyse. For each continuous variable, cut-offs used to evaluate the prognosis were in accordance with the literature (e.g.: BMI ≥ 25 or 30 kg/m 2 ( Physical status: the use and interpretation of anthropometry 1995; Mintziras et al. 2018), platelets count ≥ 350 × 10 9 /L, leucocytes count ≥ 11 × 10 9 /L, hemoglobin < 100 g/L) (Khorana et al. 2008;Veer et al. 2018). The cumulative incidence of VTE was estimated using the inverse Kaplan Meier method.
Quantitative variables were reported as median with corresponding interquartile range and were compared by Wilcoxon test. Qualitative variables were reported as numbers and percentages and compared using chi-square or Fisher test. All statistical analyses were performed using R ® version 3.2.2 (R project, Auckland, New Zealand).

Whole study population
A total of 174 patients matching with inclusion criteria were identified. Patients and disease characteristics were reported in Table 1, as well as the corresponding data for patients who experienced VTE during follow-up (n = 46) and for those who did not (n = 128). Forty-six patients (26.4%) had at least one VTE event during follow-up. There was no statistically significant difference between the two groups regarding their main characteristics. The median age was 67 years [60-75] with a sex ratio close to 1. At the time of diagnosis, most patients (88.5%) had an ECOG-PS of 0-1, about 1/3 (32.8%) had diabetes and 37.4% had a history of tobacco consumption. At baseline, the median albumin was 32.6 g/L [27.2-37.9].
About half of patients had a primary tumor located in the pancreatic head and 74.7% had metastases (17.2% after prior surgery of the primitive tumor). At baseline, median CA19-9 and median CRP were 353.5 UI/L [87.5-2761] and 17 mg/L [7.0-43.5], respectively. About 2/3 of patients received intensive chemotherapy (FOLFIRINOX or Gemcitabine/ Nab-Paclitaxel). About 1/3 were treated with Gemcitabine only. The Khorana score at baseline was similar between patients who did vs. those who did not experience VTE

Incidence of venous thromboembolism
During a median follow-up of 9.3 months [5.

Clinical presentation and treatment of venous thromboembolism events
Of the 46 patients who experienced VTE during followup, the majority of VTE was symptomatic (78.3%) and extra-abdominal (76.1%), including pulmonary embolism in 30.4%. Most of them were treated with Low Molecular Weight Heparins (LMWHs) (69.6%). Vitamin K antagonists (VKA) (4.3%), direct oral anticoagulants (DOAC) (2.1%), and unfractionated heparin (2.1%) were rarely used and 17.4% of patients did not receive specific treatment for VTE. One patient (2.1%) underwent vena cava filter. Under anticoagulant agents, VTE was considered as progressive (local extension, or by the occurrence of a second VTE site), stable or reduced in 17.4%, 67.4% and 13%, respectively. In the case of VTE extension and/or recurrence, the anticoagulation strategy was mainly based on the increase of LMWH dose (n = 4), or maintaining the same strategy (n = 4). Bleeding complication was observed in 15.2% with 1 related death (2.2%) by digestive bleeding.

Time from venous thromboembolism to disease progression or death
The median time from VTE to disease progression was 1.6 months (95% CI 0.0-2.8) (Supplementary Fig. 1, which means that half of the patients who had VTE experienced disease progression at the same time. Half of the remnant patients had disease progression about 3 months after the VTE occurrence. The median time from VTE to death was 5.8 months (95% CI 2.9-8.4) ( Supplementary Fig. 2, but there was any difference between patients with abdominal thrombosis (n = 11) compared to those who had extra-abdominal thrombosis (n = 35, p = 0.01), or regarding the symptomatic character of the VTE event (p = 0.85).

Discussion
The present study confirms the high risk for VTE in patients with advanced PDAC as about 1/4 to 1/2 of them experienced such events during follow-up. Previous large cohorts reported such outcomes (Lee et al. 2016;Frere et al. ;2019;Gade et al. 2017;Chen et al. 2018). Thereby, the interest of thromboprophylaxis is revived for all ambulatory patients treated for advanced PDAC as recommended since 2013 (Frere et al. 2019;Key et al. 2020;Farge et al. 2013).
However, VTE is a time-dependent factor and only early VTE, defined by the occurrence of VTE within the three months from advanced PDAC diagnosis, seems prognostic in this study which involves about 1 in 10 patients. Most studies have reported VTE without taking into account its time-dependent nature (Godinho et al. 2020;Shaib et al. 2010;Kruger et al. 2017;Oh et al. 2008). Only a few recent studies evaluated VTE either by considering VTE as timedependent factor in Cox model (Frere et al. , 2019), or as binary factor (Chen et al. 2018;Mandalà et al. 2007;Kim et al. 2018).
Time from VTE to disease progression or death has been rarely reported in the literature. The present study showed a probable relation between VTE and tumor aggressiveness (median time to progression: 1.6 months; median time to death: 5.8 months) which raises the question about the impact of anticoagulation therapy on survival. Moreover, 17.4% of patients treated for VTE experienced an extension of their thrombosis under anticoagulation therapy and coagulum stability was observed in 67.4% in the present study. Seven patients (15.2%) had bleeding complication including one (2.2%) death by digestive haemorrhage. There  is no strong evidence that thromboprophylaxis improves survival outcomes in advanced PDAC. In vitro, anticoagulation agents might have an anti-tumor effect by inhibiting metastatic spreading process. Most randomized clinical trials that evaluated the impact of anticoagulant agents on survival included all cancers (Kakkar et al. 2004;Klerk et al. 2005). Conclusions were controversial regarding the role of anticoagulation in the improvement of survivals, and outcomes in a subgroup of patients with PDAC should be interpreted with cautious (Frere et al. 2019). Corresponding data with DOAC are very scarce (Frere et al. 2019). Overall, anticoagulant use is recommended to reduce the risk of VTE (Frere et al. 2019;Key et al. 2020;Maraveyas et al. 2012;Pelzer et al. 2015) but not to improve survival in patients with cancer without VTE (Doormaal et al. 2011). For this last reason, thromboprophylaxis is rarely prescribed for ambulatory patients treated for PDAC in our center.
The present study is not designed to assess the impact of thromboprophylaxis on survival but underlines the importance to identify patients who will experience early VTE. By definition, all patients with pancreatic cancer have a Khorana score at least 2 and the vast majority of them have a score at 3, like in this cohort. Hence, the Khorana score does not allow the identification of patients who will truly experience early VTE. Thereby, if primary thrombophylaxis is done, we should consider that most of the patients are overtreated with the potential risk of anticoagulation-related complication. Other score should be developed to improve the identification of the patients who did from those who did not experience VTE within three months.
The ONKOTEV score is an emerging alternative to predict VTE (Godinho et al. 2020). To calculate the ONKOTEV score, vascular/lymphatic compression and previous history of VTE must be known. Hence, this score is not used in clinical practice due to its complexity, which supports the interest of the development of a new score more feasible and reproducible by clinicians.
In conclusion, VTE occurred in about ¼ to 1/3 of patients with advanced PDAC. Early VTE, defined by VTE occurrence within three months from advanced PDAC diagnostic, might be prognostic in this study which supports the international guidelines. However, failing to apply the primary thromboprophylaxis recommendations for all ambulatory patients treated for PDAC, further scores should be developed to identify better patients who are at very high-risk for early VTE from the others. Such score could avoid overtreatment of the majority of patients (9/10).