Effectiveness and safety of anticoagulants among patients with venous thromboembolism and active cancer who also had prior bleed or prior renal disease

Abstract Objective Patients with active cancer and venous thromboembolism (VTE) have elevated risk of recurrent VTE (rVTE) and major bleeding (MB). The risk is even higher within those with a prior bleeding event or renal disease. There is a need to understand the risk of rVTE and MB of commonly used anticoagulants among these high-risk patients. Methods VTE patients with active cancer and treated with apixaban, warfarin, or low molecular weight heparin (LMWH) within 30 days of VTE were identified from five claims databases in the United States. Inverse probability of treatment weighting (IPTW) was used to balance patient characteristics. The post-IPTW population was stratified by prior bleed or renal disease status. Cox proportional hazards models were used to evaluate interactions between treatment and prior bleed or renal disease on risk of rVTE and MB, with p value <.1 considered significant. Results Study criteria were met by 30,586 VTE cancer patients: 35.0% had prior bleed and 29.0% had renal disease. For apixaban, LMWH, and warfarin cohorts, the incidence (events per 100 person-years) of MB was higher in patients with prior bleed (17.48 vs 7.58, 25.61 vs 13.11, and 20.38 vs 8.97) or renal disease (15.79 vs 8.71, 22.11 vs 15.90, and 18.49 vs 10.39) vs those without the conditions. Generally, there were no significant interactions between anticoagulant use and prior bleed or renal disease on rVTE and MB (p for interaction >.1). Conclusion The incidence of MB was higher among those with prior bleed or renal disease. Effects of apixaban, warfarin, or LMWH were generally consistent regardless of prior bleed or renal disease status.


Introduction
Patients with cancer as well as venous thromboembolism (VTE) and renal disease are at a higher risk of a recurrent VTE (rVTE) and bleeding compared to those without renal disease 1 .The prevalence of renal disease and VTE increases with age and the accumulation of risk factors, including type 2 diabetes and hypertension [2][3][4] .Furthermore, the choice of therapy can be more limited in patients with severe renal disease owing to the clearance of direct oral anticoagulant (DOAC) medications by the kidneys 5 .
In addition, patients with cancer and a VTE event and a prior bleeding event are at increased risk of a major bleeding (MB) event, which can also influence the choice of anticoagulant therapy in these patients 6,7 .As patients with cancer already have a higher risk of bleeding compared to those without, this increased risk due to a prior bleeding event may further complicate anticoagulant therapy in this patient population 6,8 .
Low molecular weight heparin (LMWH) was once the standard of care for VTE among patients with cancer 9-11 .However, oral anticoagulants (OACs) such as warfarin have also been used for VTE treatment in patients with cancer 12 .The last decade has seen the emergence of DOACs including apixaban, dabigatran, edoxaban, and rivaroxaban as alternative VTE treatments 9, [13][14][15][16] .Randomized studies have shown that edoxaban, rivaroxaban, and apixaban are non-inferior to dalteparin (LMWH) in the treatment of VTE in patients with cancer 10,11,13,14 .Given that, a 2021 guideline released by the American Society of Hematology (ASH) recommended that a DOAC (apixaban or rivaroxaban) or LMWH be used for the initial treatment (within week 1) of VTE for patients with cancer 17 .In addition, the guideline suggests that DOACs are the preferred treatment over LMWH for the short-term (3-6 months) treatment of VTE.The ASH guidelines note that, compared with LMWH, DOACs have the potential to reduce the risk of rVTE and major bleeding (MB).However, the guideline panel indicated that evidence for their recommendations is uncertain due to the small number of controlled trials that have examined these treatments in patients with cancer.There is a paucity of clinical trial and real-world data regarding patients who are at particularly high risk of MB and rVTE.
Given the increased risk of rVTE or bleeding in patients with VTE and cancer who also had renal disease or prior bleed, there is a need to understand the effectiveness and safety of different anticoagulant treatments in these high-risk cancer subpopulations.The goal of this real-world study was to evaluate the effectiveness (rVTE) and safety (MB and clinically relevant non-major bleeding [CRNMB] events) of apixaban, LMWH, and warfarin use based on prior bleed status or renal disease status among patients with VTE and active cancer.

Data sources
Medical and pharmacy claims data were pooled from four commercial databases (Optum, Humana, PharMetrics, and MarketScan) and the Medicare Fee for Service (FFS) database in the United States.The study period ranged from March 1, 2014, through June 30, 2017 (MarketScan), December 31, 2017 (Optum, Humana, Medicare FFS), or March 31, 2018 (PharMetrics).Data were de-identified and data collection complied with the requirements of the Health Insurance Portability and Accountability Act (HIPAA).

Study design and patient selection
A retrospective longitudinal cohort design analysis was used to examine adults 18 years of age or older who were diagnosed with VTE (i.e.any medical claim with an International Classification of Diseases − 9/10 th Revision -Clinical Modification [ICD-9-CM or ICD-10-CM] diagnosis code for VTE in any position) with an identification period spanning September 1, 2014, through the end of available data in each dataset.The first VTE diagnosis was defined as the index VTE event.Patients were included if they had active cancer and initiated treatment with an anticoagulant (apixaban, LMWH, or warfarin) in the outpatient setting within 30 days after the index VTE event.The outpatient anticoagulant treatment initiation date was defined as the index date.Active cancer was defined as having at least one of the following between the 6 months pre and 30 days after the index VTE event date: (1) two or more claims at least 1 day apart for a cancer diagnosis, or (2) one cancer diagnosis claim and an additional cancer treatment claim.Patients with VTE and active cancer were also required to have had continuous health plan enrollment with medical and pharmacy benefits for �6 months prior to the index VTE event until the index date to allow for the determination of baseline comorbidities.Warfarin patients with LMWH bridging treatment who had evidence of MB or recurrent VTE between warfarin and the start of LMWH were excluded.The final study population excluded patients with evidence of atrial fibrillation/ flutter, mechanical heart valve, or outpatient anticoagulant use in the 6 months prior to the index date (baseline period).In addition, patients with evidence of VTE in the 6 months prior to the index VTE event and evidence of an inferior vena cava filter, pregnancy, or antiphospholipid syndrome during the study period were excluded.See Figure 1 for detailed inclusion and exclusion criteria.Index] score, and comorbid conditions), selected surgeries, and cancer-related characteristics (evidence of metastasis and cancer-related treatment) were measured for the baseline period.Renal disease was identified using evidence of ICD-9-CM and ICD-10-CM codes and included the following conditions: chronic renal disease (e.g.chronic kidney disease [CKD] stage 1-5, unknown stage of CKD, hypertensive CKD) and acute renal disease (includes nephritis and hepatorenal syndrome) in the baseline period.Patients with a prior bleed were defined as patients with a diagnosis (ICD-9 or ICD-10 codes) for any bleeding event (MB or not MB) in the baseline period.

Outcomes
Study outcomes included rVTE, MB, and CRNMB events.The follow-up period for study outcomes ranged from the index date through the earliest of death, disenrollment, switch to another anticoagulant, discontinuation of treatment (no evidence of anticoagulant use for at least 30 days after the end of the patient's day supply), or 6 months post-index date.A rVTE event was defined as an inpatient admission with a diagnosis for VTE in the primary or first-listed position.If the index VTE event was an inpatient admission, any admission within 7 days of the index VTE event was not considered a rVTE.An MB event was defined as an inpatient admission with a diagnosis for bleeding in the primary or first-listed position.A CRNMB event was defined as an inpatient admission for a bleeding event that did not qualify as an MB event (i.e. had a bleed diagnosis in the secondary position and the bleeding was for a non-critical site) or as an ambulatory care visit for non-critical bleeding.All diagnoses were identified using ICD-9 or ICD-10 codes.

Statistical analysis
Results were presented by initial anticoagulant used (warfarin, LMWH, and apixaban).Inverse probability of treatment weighting (IPTW) was used to balance patient characteristics between treatment cohorts among all patients with VTE and cancer.Weights were generated as the inverse of propensity score.A multinomial logistic model was used to generate the propensity score which is the probability of receiving treatment (LMWH, apixaban, or warfarin), with LMWH as the reference.The variables used to calculate propensity score included demographics, type of VTE diagnosis, NCI comorbidity index score, comorbidities, cancer type, cancer metastases, VTE risk level, and cancer-related treatment.Stabilized weights were used to reduce variability in the treatment weights caused by outliers.Standardized differences were used to determine clinically-meaningful differences between cohorts with a value of 10 or higher indicating a clinicallymeaningful difference.After IPTW, patients with VTE and cancer were further stratified by evidence of prior bleeding or renal disease.Standardized differences were checked again after stratification.
Outcomes were reported as incidence rates (IRs) per 100 person-years.Cox proportional hazards models were used to determine the interaction between prior bleed or between renal disease and treatments (apixaban vs warfarin, apixaban vs LMWH, and warfarin vs LMWH) on recurrent VTE, MB, or CRNMB and to compare incidence rates among treatment cohorts for each subgroup.Interactions with a p value <.1 were considered significant.The reason we used a less stringent cut-off (p <.1) is to avoid missing any significant subgroup interactions.This cut-off was based on a previous publication 18 .The Cox models were used to further adjust for variables that had a standardized difference of at least 10 after stratification by renal disease and prior bleed.

Patients characteristics by prior bleed status
Among the 30,586 patients with VTE and cancer who met all study criteria (Figure 1), 10,708 (35.0%) had a prior bleed, and 19,878 (65.0%) had no prior bleed.Use of warfarin, LMWH, and apixaban was similar between those with a prior bleed (37.8%, 36.7%, and 25.4%, respectively) and without a prior bleed (37.9%, 36.5%, and 25.6%, respectively) (Figure 1 and Table 1).Patients with and without a prior bleed were generally 68 to 72 years of age, and more than half were female (Table 1).Patients with a prior bleed were more likely to have their index VTE diagnosed in the inpatient setting (�60%) compared to patients without a prior bleed (�50%).A majority (�55%) of patients had their index VTE as DVT only, regardless of their prior bleed status.Patients with prior bleed had a higher mean NCI score (3.5-3.8) and higher prevalence of comorbidities (Table 1) compared to patients without prior bleed (mean NCI score 2.6-2.8).Patient characteristics were generally balanced among treatment groups after IPTW and when stratified by prior bleed status.

Evaluation of outcomes by prior bleed status
As represented in Figure 2, incidence rates of rVTE were similar or higher in patients with prior bleed versus patients without prior bleed for apixaban (8.43 vs 7.16 events per 100 person-years), LMWH (18.83 vs 13.15), and warfarin (9.98 vs 9.32) cohort.However, those with prior bleed versus those without prior bleed had a higher incidence rate across all three drug cohorts (apixaban, LMWH, and warfarin) for MB (17.48When comparing different anticoagulants, the incidence rates of rVTE, MB, and CRNM bleeding were numerically lower in the apixaban cohort compared to warfarin or LMWH regardless of prior bleed status.There was no significant interaction between treatment (apixaban vs warfarin, apixaban vs LMWH, and warfarin vs LMWH) and prior bleed status on any of the outcomes: rVTE, MB, and CRNMB (p >.1 for all interactions), i.e. treatment effects were not significantly different between those with and without prior bleed.S1.Those with renal disease were older (72-75 years of age across medication groups) and more than half were male compared to those without renal disease (Table 2).Patients with renal disease were more likely to have their index VTE diagnosed in the inpatient setting (�64%) compared to those without renal disease (�48%).A majority (52-63%) of patients had their index VTE as DVT only regardless of renal disease status.Patients with renal disease had a higher mean NCI score (�5.0) and higher prevalence of comorbidities (Table 2) compared to those without renal disease (mean NCI score �2.3).Patients with renal disease (vs without renal disease) were more likely to have a prior bleed (�41% vs �32.0%), but less likely to have metastasis (�44% vs �50%).

Evaluation of outcomes by renal disease status
As represented in Figure 3, incidence rates of rVTE were similar or higher among patients with renal disease vs patients without renal disease for apixaban (7.57vs 7.60), LMWH (18.48 vs 14.02), and warfarin (10.30vs 9.22) cohorts.However, those with renal disease had a higher incidence rate across all three drug cohorts (apixaban, LMWH, and warfarin) for MB (15.79

Discussion
This real-world study of patients with VTE and active cancer showed that those with prior bleed or renal disease were more likely to have comorbidities and had a higher incidence of MB and CRNM bleeding compared to those without prior bleed or without renal disease, respectively.The treatment effects of apixaban, warfarin, and LMWH were in general not significantly different between those with and without prior bleed or between those with or without renal disease.Patients with cancer are at increased risk for VTE (either new or recurrent) 19,20 .Patients with cancer and VTE are also predisposed to bleeding events due to various cancer treatments or due to the cancer itself, which makes it challenging to treat these patients with anticoagulants 21 .Compounding this problem is the history of bleeding in some of these patients, which can increase the risk of another major bleeding event 22 .Consistently our current study found that patients with prior bleed had much higher incidence rates of MB and CRNM bleeding compared to those without prior bleed regardless of the anticoagulant they received.Additionally, patients with a prior bleed tended to have a higher comorbidity index and higher prevalence of comorbidities, indicating potential challenges for anticoagulant therapy in this patient population.However, our study

CURRENT MEDICAL RESEARCH AND OPINION
showed no significant interactions between the treatment (apixaban vs LMWH, warfarin vs LMWH, and apixaban vs warfarin) and prior bleed status on rVTE, MB, and CRNM bleeding, suggesting that the treatment effects of apixaban, LMWH, and warfarin were not significantly different between patients with and without prior bleed.
Patients with cancer and VTE who also had renal disease are another highly vulnerable population.These patients have been shown to be at increased risk for mortality, rVTE, and MB 5 .In this study, we also found that patients with renal disease had higher incidence of MB and CRNM bleeding versus those without renal disease.Patients with renal disease were more likely to be 75 years of age or older, more likely to have had VTE diagnosed in the hospital and have a markedly higher comorbidity index score than those without renal disease.While these patients need careful considerations when being prescribed an anticoagulant, there is a knowledge gap on optimal anticoagulant treatment for these  17 .In this study, we generally found no significant interaction between the treatment (apixaban vs LMWH, warfarin vs LMWH, and apixaban vs warfarin) and renal disease status on rVTE, MB, and CRNM bleeding.The lack of a significant interaction indicates the treatment effects of apixaban, LMWH, and warfarin were not significantly different between patients with and without renal disease.Our results are consistent with the renal sub-analysis of the CARRIVAGIO trial which found no significant interaction between the treatment of apixaban vs. dalteparin and the presence of renal insufficiency (estimated glomerular filtration rate [eGFR] < 60 or <50) on rVTE or bleeding 1 .

Limitations
The present analysis includes limitations associated with retrospective claims analyses.Only associations and not causation can be inferred from this study and there may be residual confounding due to unmeasured variables (e.g. over-the-counter medications, lab values, etc.).There is a significant overlap between prior bleed and prior renal disease.As shown in Table 1, among patients with prior bleed, 20-40% had renal disease.However, sample size for patients with both conditions is small for some cohorts, limiting additional investigation on how both conditions together impact outcomes of anticoagulant treatments.Only about 50% of patients have information on renal stage.The other 50% either reported unknown stage or other renal disease.
Consequently, sample size for each renal stage was small and it is challenging to assess outcomes by renal stage.As such, any impact of the severity of renal stage was not able to be determined and results for patients with renal disease should be interpreted within the context that it is a heterogenous group comprising of acute and chronic renal disease.In addition, the commercial databases from which the data were derived do not have complete death information for the patients, and, consequently, mortality and fatal rVTE in patients were not assessed.The analysis assumes that medications were taken properly and that diagnosis codes were accurate.Nevertheless, coding errors and missing data may exist, which could contribute to underreporting or misclassification of outcomes.Transfusion has been included as part of the diagnosis codes to identify bleeding.But transfusions may be administered in the absence of active bleeding given the severe anemia characteristic of the study patients.The algorithm used to determine CRNMB events (including ICD-9 and ICD-10 diagnosis codes) has not been validated; however, this algorithm does follow the suggested definition by the International Society on Thrombosis and Haemostasis (ISTH) as closely as possible 23 .There was uncertainty whether rVTE events were indeed new events and not the follow-up care of previous VTE events.To minimize the uncertainty, any admission within 7 days of the index VTE event was not considered as rVTE.The data used in this study is a little bit out of date and the patterns of treatment may not fully reflect current guidelines.Finally, the findings from this analysis cannot be generalized to the entire US population of patients with VTE and active cancer as uninsured individuals, Medicaid recipients, and patients within the Veterans Affairs health system were not included.Additionally, the data were from US databases and may not generalize to other populations.

Conclusion
The results of this study add to the growing body of literature evaluating highly-vulnerable patients who not only had VTE and active cancer but also prior bleed or renal disease.Our study found that patients with prior bleed or renal disease had higher incidence rates of MB and CRNM bleeding versus those without prior bleed or without renal disease, respectively.However, treatment effects of apixaban, LMWH, and warfarin were consistent regardless of prior bleed or renal disease status.Further research is needed to assess the efficacy/effectiveness and safety of anticoagulant treatment in these special patient populations.

Declaration of funding
This work was supported by Pfizer Inc. and Bristol Myers Squibb/Pfizer Alliance.
Since this study does not involve the collection, use, or transmittal of individually identifiable data, it was deemed exempt from Institutional Review Board review by Solutions IRB.Both the datasets and the security of the offices where analysis was completed (and where the datasets are kept) meet the requirements of the Health Insurance Portability and Accountability Act of 1996.Solutions IRB determined this study to be EXEMPT from the Office for Human Research Protections (OHRP)'s Regulations for the Protection of Human Subjects (45 CFR 46) under Exemption 4: Research involving the collection or study of existing data, documents, records, pathological specimens, or diagnostic specimens, if these sources are publicly available or if the information is recorded by the investigator in such a manner that subjects cannot be identified, directly or through identifiers linked to the subjects.The HIPAA Authorization Waiver was granted in accordance with the specifications of 45 CFR 164.512(i).This project was conducted in full accordance with all applicable laws and regulations, and adhered to the project plan that was reviewed by Solutions Institutional Review Board.

Figure 2 .
Figure 2. Outcomes by prior bleed status.^Additional adjustment in the Cox model resulted in a direction shift of the hazard ratio for recurrent VTE (LMWH vs warfarin); � Reported as per 100 person-years.Abbreviations: CI, confidence interval; IR, incidence rate; VTE, venous thromboembolism.

Table 1 .
Patient characteristics by prior bleed (post IPTW) and by anticoagulant type.

Table 2 .
Patient characteristics by renal disease (post-IPTW) and by anticoagulant type.