FormalPara Key Points

Frail patients with cancer with cancer-associated thrombosis (CAT) are at higher risk of complications due to drug–drug interactions between direct oral anti-coagulants and anti-neoplastic drugs. Most of them are preventable, being associated to clinical situations and metabolic pathways.

A therapeutic decision algorithm taking in account patient frailties is proposed for the choice of an anti-coagulant for the treatment of patients with cancer-associated thrombosis.

A systematic collaboration between oncologists, vascular pathology specialists, pharmacologists and pharmacists is warranted to ensure an optimal patient management.

1 Introduction

Recent major therapeutic advances have led to an increase in the overall survival of patients with cancer. Population risk factors (e.g. elderly, smoking in women) are associated with an increase in the incidence of cancers. Also, anti-cancer drugs are used for longer treatment periods, resulting in a sharp increase in the prevalence of patients with cancer receiving anti-neoplastic treatments [1, 2].

The thrombotic risk in patients treated for cancer is increased seven-fold compared with the cancer-free population [3], while the bleeding risk is multiplied by two to three times [4]. Venous thromboembolic (VTE) disease represents the second leading cause of death after neoplastic progression [5]. The individual risk of VTE recurrence or bleeding must be assessed specifically for each patient to support the choice of anti-coagulation.

The 2019 American Society of Clinical Oncology (ASCO) [6] and 2022 International Initiative on Thrombosis and Cancer (ITAC) [7] guidelines recommend anti-Xa direct oral anti-coagulants (DOACs), including rivaroxaban, apixaban and edoxaban, for the treatment of cancer-associated thrombosis (CAT) in patients treated with systemic anti-cancer therapy, without bleeding risk and in the absence of drug–drug interactions with the anti-cancer treatment, or other concomitant therapy. DOACs are factor Xa inhibitors and/or factor IIa inhibitors. Caution is recommended in patients with gastrointestinal cancers, especially in unresected intra-luminal tumour and in genitourinary tumours, due to the high risk of bleeding observed with anti-Xa DOACs. Anti-Xa DOACs are not indicated in several haematologic cancers as thrombocytopaenia is frequently associated with the disease and/or the anti-neoplastic treatment. The recommended duration for the anti-coagulant treatment is at least 6 months and then extended as long as the cancer is active. A cancer is considered active when one of the three following conditions is present: the cancer has not received potentially curative treatment, the evidence shows that treatment has not been curative (recurrence, progression), or the cancer treatment is ongoing.

Like other drugs eliminated through metabolic pathways, the use of anti-Xa DOACs require the detection of the risk of drug–drug interactions (DDIs) with concomitant medications. Anti-Xa DOACs approved for the treatment of CAT are eliminated by the cytochrome P 450 (CYP) 3A4/5 enzymes and by P-glycoprotein (P-gp), leading to an increased bleeding or thrombotic risk when associated to strong CYP 3A4/5 and/or P-gp inhibitors or inductors, respectively.

Patients at high risk of VTE recurrence and high risk of bleeding are particularly vulnerable to DDIs.

Co-morbidities are common, and their incidence increases during the course of the disease. Patient frailty can lead to reduced DOAC elimination, resulting in an increased risk of bleeding or an increased likelihood of clinically relevant DDIs. There is no validated predictive score for bleeding in patients treated for cancer, whereas frailties are known to increase the bleeding risk in patients with atrial fibrillation treated with anti-coagulants [8]. In the absence of validated predictive bleeding scores, the RIETE bleeding risk score predicting fatal bleeding may be useful in patients with CAT, as cancer is known as a factor of frailty [9, 10]. Patient frailty increases with the number of co-morbidities and the severity of clinical consequences in the case of DDIs. The co-morbidities which seem to be the most relevant to be taken into account include renal insufficiency, hepatic function disorders, undernutrition and obesity, since they modify the pharmacokinetics of anti-coagulants. Caution is required in frail patient in oncology and geriatrics. In addition, mucosal diseases and thrombocytopaenia may increase the risk of bleeding complications. In this context, taking frailty into account, an age-related condition of increased vulnerability to acute endogenous or exogenous stressors is key to optimise treatment strategy. The gold standard to assess frailty in older adults with cancer is geriatric screening followed by geriatric assessment (GA) across essential GA-domains (social status, physical function, nutrition, cognition, emotion, co-morbidity, polypharmacy). GA enables tailoring of both oncological therapy and non-oncological interventions to the patient’s vulnerabilities.

This review aims to consider the impact of key factors that contribute to patient’s frailty on the anti-coagulation strategy and to discuss a decision-making algorithm for the management of patients with CAT and help with the choice between DOAC and LMWH.

2 Frailties and Risk Factors Influencing VTE Management

Main factors that contribute to patient’s frailty include the type of cancer and individual patient characteristics, such as creatinine clearance, liver function tests, age, BMI platelets rate and DDIs with anti-neoplastic treatments, and concomitant medications [11].

2.1 Cancer-Related Risk Factors

When prescribing anti-Xa DOACs, an additional risk of bleeding must be considered for unresected endoluminal digestive urothelial cancers and mucosal gynaecological cancers. Moreover, acute leukaemia, frequently associated with long-lasting thrombocytopaenia, exposes the patient to an important bleeding risk. Also, an increased thrombotic recurrence risk was reported in patients with multiple myeloma, pancreatic cancers, or lung cancer included in the RIETE registry [12]. However, no CAT recurrence risk score has been fully validated to date, including the Ottawa score.

2.2 Patient-Related Risk Factors

Metabolism is influenced by patient-specific factors, such as comorbidities and genetics, therapeutic-specific factors, including drug characteristics and interactions, and disease-specific factors, such as organ dysfunction [13].

2.2.1 Chronic Kidney Disease

The prevalence of chronic kidney disease (CKD) in patients with cancer is common and may depend on the type of cancer, being higher in patients with renal cancer (50%), urinary tract cancer (1 of 3 patients) and pancreatic cancer (1 of 5 patients) [14, 15]. The prevalence of CKD is lower in patients with colon cancers (5.3%) and brain tumours (2.5%) [16].

Renal elimination varies between anti-Xa DOACs: 50% for edoxaban, 36% for rivaroxaban and 27% for apixaban [17, 18]. Variations in anti-Xa DOACs plasma concentration in case of renal impairment are summarised in Table 1.

Table 1 Significant variations in anti-Xa DOACs plasma exposure (AUC*) in case of renal impairment and major drug–drug interactions with non-anti-neoplastic drugs (adapted from Bertoletti et al. 2017) [15]

Bleeding is more common in patients with higher anti-Xa DOACs peak levels [19]. In patients with intermediate clearance between 30 and 49 ml/min, anti-Xa DOACs are well tolerated without dose adjustment. Full doses of anti-Xa DOACs are contraindicated by European recommendations when creatinine clearance is below 30 ml/min for rivaroxaban and edoxaban, or 25 ml/min for apixaban. On the contrary, the US Food and Drug Administration (FDA) allows apixaban use up to 15 ml/min in VTE. However, studies are still ongoing, and efforts to detect drug–drug interactions must be maintained given the risk of drugs accumulation.

2.2.2 Liver Diseases

Patients with liver disease have an increased thrombotic risk and bleeding risk. Anti-Xa DOACs are primarily eliminated via the CYP 3A4 metabolic pathways. Among them, apixaban has the greatest hepatic elimination (75%), while rivaroxaban and edoxaban are eliminated at 65% and 50% by the liver, respectively [20]. The European Medicines Agency does not recommend the use of apixaban in patients with liver disease associated with coagulopathy and a clinically relevant bleeding risk, aspartate aminotransferase/alanine aminotransferase (AST/ALT) > 2 upper limit of normal (ULN), or liver disease expected to affect survival. In the general population, restrictions on their use in patients with cirrhosis are based on the Child–Pugh score [11, 20]. In case of cancer with cachexia, hypoalbuminemia Child–Pugh score could be overrated and may not provide a reliable evaluation of liver function.

All anti-Xa DOACs are contraindicated in the case of severe liver disease, but their use remains possible in the case of moderate liver disease without dose adjustment. In the event of a contraindication, the treatment option is LMWHs, as they are not affected by the liver metabolism.

2.2.3 Low Body Weight, Hypoalbuminemia and Malnutrition

Hypoalbuminemia, malnutrition and low body weight are frequently associated in patients with cancer. A proportion of 80% and 55% of patients with cancer have albumin plasma levels below 35 g/L and 20 g/L, respectively [21]. Anti-Xa DOACs are strongly bound to albumin. Hence, malnutrition and hypoalbuminemia are bleeding risk factors, not fully evaluated in specific studies in patients with CAT. Pharmacokinetic studies show a modification of the area under the curve (AUC) and Cmax. In patients weighing less than 60 kg, renal function assessment is important because conventional tests often overestimate renal function due to poor evaluation of muscle mass. Studies including atrial fibrillation patients or VTE reported fewer bleeding events in low weight patients (< 50 kg) with anti-Xa DOACs than with vitamin K antagonists (VKA) [11, 22].

2.2.4 Obesity

Various studies and one meta-analysis report comparable efficacy and safety of anti-Xa DOACs and low molecular weight heparin (LMWH) followed by VKA in patients with obesity without cancer [23]. Available data seem to show good safety and efficacy of anti-Xa DOACs for patients between 50 kg and 130 kg or for body mass index (BMI) up to 40–50 [22,23,24]. The use of doses adapted to body weight do not require monitoring of anti-Xa activity. LMWHs can be used for extreme weights up to 160 kg [11].

2.2.5 Thrombocytopaenia

Thrombocytopaenia can result from bone marrow infiltration or from anti-cancer medications and represents an important limitation for the use of anti-thrombotic treatments [25]. Post-chemotherapy thrombocytopaenia does not decrease the risk of VTE recurrence while it increases the risk of bleeding. Chemotherapy-induced thrombocytopaenia is a frequent complication of anti-neoplastic treatment. Severe thrombocytopaenia (< 10,000/µl) leads to an increased risk of bleeding [26]. Indeed, bleeding rates in this population range from 7 to 33% [27].

The risk of bleeding depends on the severity of thrombocytopaenia and its duration. According to an expert consensus, if the platelet count is > 50 × 109/L, there is no need to change the anti-coagulant therapy because the bleeding risk remains acceptable [28]. Clinical studies are warranted to precisely document the relationship between the risk of bleeding and thrombocytopenia.

If the platelet count is < 50 × 109/L, it may be necessary to transfuse platelets to maintain a full-dose anti-coagulant therapy with LMWH, if within 1 month of VTE diagnosis. After 1 month of anti-coagulant therapy, LMWH dose can be reduced without the need for platelet transfusion, because the risk of VTE recurrence is reduced [29]. There is no guidance to reduce anti-Xa DOAC doses in case of thrombocytopaenia, and a switch to LMWH is therefore recommended. Additionally, in all randomised clinical trials with DOACs in CAT, the cut-off for platelets rates at inclusion was 50 × 109/L.

2.2.6 Oral Drugs Malabsorption

Iterative nausea and vomiting induced by chemotherapy obviously alters the assimilation of drugs if they are rejected [30].

In the event of digestive surgical resection, the type of surgery influences the absorption of anti-Xa DOACs.

The mechanism of anti-Xa DOACs absorption varies from one drug to another. Rivaroxaban is absorbed primarily in the stomach and poorly in small intestine and colon; apixaban is absorbed primarily in the proximal small intestine with limited absorption in stomach and colon; and edoxaban is predominately absorbed in the proximal small intestine [31]. Gastrectomy reduces rivaroxaban plasma peak under therapeutic concentration, while apixaban concentration was shown to reach expected plasma concentration. Colectomy has no impact on anti-Xa DOACs concentration [24, 31, 32].

2.2.7 Elderly

Elderly patients frequently combine frailties: malnutrition with hypoalbuminemia, low muscle mass with risk of fall, renal insufficiency and co-medications, some of which may affect haemostasis. Elderly patients with cancer are a particularly frail population at risk for iatrogenic complications [33], especially as they are often poly-medicated [34].

Lesser-known but important drug–drug interactions with non-anti-neoplastic drugs

The use of anti-platelet agents is common in the population treated with concomitant anti-coagulants for atrial fibrillation leading to a higher rate of bleeding. In EINSTEIN [35, 36] and AMPLIFY [37] trials with rivaroxaban and apixaban, respectively, only low doses of aspirin below 165 mg daily were allowed and yet were associated with increased bleeding rates [hazard ratio (HR) 1.81].

Non-steroidal anti-inflammatory (NSAIDs) were also associated with a two-fold increase in bleeding risk, mainly related to gastric bleeding risk [38]. The concomitant use of anti-coagulants and selective serotonin reuptake inhibitors (SSRI) may be associated with a bleeding risk resulting from an interaction with platelets and serotonin [39]. This combination must be carefully considered as patients with cancer are frequently treated for depressive syndrome.

These accumulations of risk factors and frailties are additive to lesser-known DDIs. The risk of bleeding is amplified with the combination of both pharmacokinetic and pharmacodynamic interactions. This is the case with several anti-neoplastic agents that are associated with bleeding events, such as gastric petechiae, gastrointestinal and cerebral haemorrhage, which may occur rarely or very frequently depending on the drug. For example, in patients with chronic lymphoid leukaemia, the concomitant use of an anti-coagulant or a platelet inhibitor with a Bruton kinase inhibitor is associated with an increased major bleeding risk [40]. However, the concomitant use of an anti-cancer drug not metabolised by CYP3A4/5 or P-gp isoenzymes does not require any special precaution (e.g. fludarabine).

3 Drug–Drug Interactions in the Management of Patients with Cancer-Associated Thrombosis

The concomitant use of several drugs may potentially alter DOACs exposure thus increasing or decreasing their bioavailability and/or clearance. DDIs may therefore affect the efficacy and safety of DOAC therapy [41]. Anti-Xa DOACs are substrates of membrane P-glycoprotein (P-gp) which acts as an efflux pump by limiting the passage of drugs through the body’s main barriers, and therefore limiting their absorption. They do not significantly modify the metabolism of CYP450 3A4/5 or P-gp substrates, such as many anti-cancer drugs. This makes their use acceptable in patients with cancer. However, they may undergo variations to their own metabolism due to the concomitant use of inhibiting or inducing drugs. Their broad therapeutic index limits the impact of these pharmacokinetic interactions on their bioavailability, particularly in the case of moderate effect on metabolic pathways. Even though most of the known interactions are documented on pharmacological data, there is a lack of clinical data. However, in clinical practice, consideration should be given to the concomitant administration of moderate or potent inhibitors and/or inducers of CYP450 3A4/5 and P-gp as they may, depending on the clinical situation, significantly modify anti-Xa DOACs plasma concentrations [42]. Minor interactions are not discussed in this review as they have limited significant clinical impact.

3.1 Pharmacokinetic Interactions

The concomitant use of a DOAC and an anti-cancer drug that strongly inhibits or induces CYP3A4/5 or Pg-P metabolic pathways is not recommended, even though anti-Xa DOACs therapeutic index is broad. In this context the increase or the decrease in anti-Xa DOACs plasma levels may still result in increased bleeding or thrombotic risk, respectively. Examples of anti-cancer drugs effects on metabolic pathways are summarised in Table 2.

Table 2 Examples of anti-neoplastic drugs effects on metabolic pathways

The initiation of an anti-cancer drug that moderately inhibits or induces CYP-450 3A4/5 or P-gp in a patient treated with DOAC warrants a thorough analysis of the bleeding or thrombotic risk of multifactorial origin such as:

  • Individual risk factors (age, renal function, nutritional status).

  • Risk factors related to the cancer disease: increased risk of bleeding in gastrointestinal, gynaecological and urothelial cancers, potentiated thrombotic risk in multiple myeloma, lung and pancreatic cancers.

  • Drug risk factors, in particular the concomitant use of drugs that inhibit or induce CYP3A4/5 or P-gp: vigilance is required for strong inhibitors of CYP3A4/5 (azole antifungals, mostly macrolides, etc.) or of P-gp (antiarrhythmics, such as amiodarone, etc.) if they must be maintained (Table 1). The concomitant use of non-anti-neoplastic drugs that strongly inhibit both CYP 3A4/5 and P-gp (i.e. telithromycin, ritonavir, itraconazole, etc.) may modify anti-Xa DOACs metabolism, thus increasing the bleeding risk and making the patient management particularly complex.

  • Potent inducers of both CYP3A4/5 and P-gp, such as anti-epileptics (carbamazepine, phenytoin, etc.) or anti-tuberculosis medications (rifampicin, etc.), can be considered with caution as they may be associated with an increase of the thrombotic risk.

  • Drug half-lives and dosing regimen: based on pharmacological criteria, the risk of drug–drug interactions may be limited during short-term applications of short half-life anti-cancer drugs, even though published data are lacking.

In all cases, increased vigilance is required in poly-medicated patients. It is crucial to keep in mind that strong CYP 3A4/5 inhibitors may increase the AUC of substrates five-fold or more. Strong inducers may decrease the AUC of substrates by 80% or more. In poly-medicated patients with cancer receiving anti-cancer drugs and anti-Xa DOACs, caution is particularly required in case of the concomitant use of strong CYP-450 3A4/5 and Pg-P inhibitors. Some of these treatments are frequently prescribed as macrolides (clarithromycin) or azole antifungal agents (itraconazole) [43].

If a decision to co-administer such drugs is made due to clinical reasons, close clinical monitoring is essential, potentially associated to a dose reduction, given the heterogeneity of available databases. Moreover, considering individualized doses according to anti-Xa DOACs plasma levels might be a good option but difficult to implement in the clinical practice. In addition, an exhaustive drug assessment should be performed to detect the use of self-medication, over-the-counter drugs or herbal products with potential inhibiting (milk thistle, grapefruit juice) or inducing (Saint John’s wort) properties, to avoid other causes of drug–drug interactions. Attention should be paid at the use of herbal products at pharmacological doses higher than the usual diet. Doses of curcuma higher than usual cooking doses may increase ibrutinib plasma levels and the bleeding risk of patients with chronic lymphocytic leukaemia, while gingerol derived from ginger may have similar effects when combined with dabrafenib/trametinib in patients with melanoma [44].

3.2 Pharmacodynamic Interactions

The initiation of an anti-cancer drug in a patient receiving an anti-coagulant requires an assessment for potential pharmacodynamic interactions, possibly resulting in either an increase or a reduction of the anti-coagulant effect. However, we must keep in mind that pharmacodynamic interactions are equally important for anti-Xa DOACs and low molecular weight heparins (LMWH).

Several anti-cancer drugs are known to have an additive effect to DOAC’s thus increasing the risk of bleeding. They include injectable drugs such as the anti-angiogenic bevacizumab used in the treatment of numerous solid tumours [45] or oral drugs, such as the anti-vascular endothelial growth factor (VEGF) regorafenib in patients with colorectal cancer [46], BCR-Abl inhibitors which are also inhibitors of platelet-derived growth factor such as imatinib in patients with chronic myeloid leukaemia [47], or Bruton tyrosine kinase inhibitors such as ibrutinib in patients with chronic lymphocytic leukaemia [40]. There is a lack of clinical data. However, the occurrence of bleeding was reported in patients treated with anti-VEGF and anti-Xa DOACs [HR of 4.15 (95% CI 0.38–45.0; p = 0,24)] compared with a group of patients treated with anti-Xa DOACs only [48]. Grade 3–4 bleedings were reported in patients concomitantly treated with ibrutinib and anti-Xa DOACs [49]. These findings were according to the multicentre international study conducted by the Scientific and Standardization Committee of the ISTH based on the TacDOAC registry assessing bleeding and thrombotic outcomes in patients receiving anti-Xa DOACs concomitantly with targeted anti-cancer therapies [50]. Some of them can cause adverse events such as petechiae, as well as gastrointestinal or cerebral haemorrhage, with a varied frequency and very diverse clinical consequences (grades 1–4). Their concomitant administration with a DOAC could potentiate the risk of bleeding by addition of effect, which justifies reinforced clinical and biological monitoring of the patient, and appropriate management in case of other situations with associated bleeding risk such as surgery, trauma and ongoing healing. A history of peptic ulcer should be sought.

Other anti-cancer drugs are associated with a thrombotic risk. This is the case of immune modulators or imids for the treatment of multiple myeloma [51] or hormone therapy such as tamoxifen approved for the treatment of breast cancer as a selective inhibitor of oestrogen receptors [52]. Despite the pharmacological inhibition of anti-Xa DOACs by tamoxifen, there is no evidence of an increased thrombotic recurrence risk with this combination [53].

In addition, several studies have shown that 20–70% of patients with cancer use self-medication and complementary alternative therapies, including phytotherapy [53,54,55]. The concomitant administration of an oral anti-cancer drug that inhibits CYP3A4/5 or P-gp and an anti-Xa DOAC requires the investigation of self-medication, such as aspirin or nonsteroidal anti-inflammatory drugs, and the intake of castor oil, fish liver oil and vitamin E which may increase the risk of bleeding.

4 Therapeutic Decision Algorithm

Bleeding risk represents the major concern when using anti-coagulants in patients with cancer, and the therapeutic decision process may be particularly complex. The potential of a DDI with anti-Xa DOACs should be carefully considered. LMWH are the preferred choice in the case of strong CYP 3A4/5 or P-gp metabolic interaction. In the case of moderate metabolic interaction, patient-related risks factors must be identified before considering the use of anti-Xa DOACs, particularly in frail patients. A simplified therapeutic decision algorithm for the management of the anti-coagulant treatment in patients with CAT is proposed in Fig. 1.

Fig. 1
figure 1

Therapeutic decision algorithm for the anti-coagulation of patients with CAT: influence of drug–drug interactions. CAT cancer-associated thrombosis, CYP cytochrome P450, DOAC direct oral anti-coagulant, LMWH, low molecular weight heparin. *Cancer with bleeding risk intra-luminal digestive cancer, genitourinary cancer, acute leukaemia. **Risk factors related to patient: impaired renal function, impaired hepatic function, low or high BMI, thrombocytopenia. ***To be considered in case of recurrence of thromboembolism

5 Conclusion

Co-morbidities and poly-medication are important determinants of frailty in patients with cancer. LMWHs and anti-Xa DOACs are recommended for the management of patients with CAT. However, given the risk of DDIs, the use of anti-Xa DOACs is particularly complex in this population at high risk for venous thromboembolism and bleeding. Therapeutic drug monitoring with measurement of plasma levels for all factor Xa inhibitors and/or anti-cancer drugs or LMWH activity is not widely available and not recommended in routine, but it could be helpful to rule out or confirm a clinically relevant DDI in case of major bleeding or thromboembolic recurrence. Appropriate management of these patients requires a multi-disciplinary approach including oncologists, vascular pathologies specialists, pharmacologists and pharmacists.