PART III - THE HIGH BLEEDING RISK PATIENT
TABLE OF CONTENTS
BOOKSHOP
Updated on May 14, 2021
PART III

The high bleeding risk patient

Róisín Colleran1,2, Philip Urban3,4
1 Cardiovascular Research Institute Dublin, Mater Private Hospital, Dublin, Ireland
2 Royal College of Surgeons in Ireland University of Medicine and Health Sciences, Dublin, Ireland
3 Hospital de la Tour, Geneva, Switzerland
4 CERC, Massy, Paris, France

Summary

Patients at high bleeding risk (HBR) comprise a growing subgroup of patients undergoing PCI. Nonetheless, such patients have historically been excluded from or under-represented in clinical trials in PCI patients. Moreover, completed and ongoing trials aiming to enrol HBR patients use heterogeneous definitions of HBR, limiting the interpretation and comparison of results. To address this issue, the Academic Research Consortium for high bleeding risk (ARC-HBR) proposed a standardised definition of HBR patients intended for use in trials of HBR patients undergoing PCI.

Using a cut-off of BARC 3-5 bleeding of ≥4% or intracranial haemorrhage of ≥1% at one year to denote major bleeding, a list of binary major and minor HBR criteria were created based on consensus and review of the available evidence. The criteria are pragmatic, objective, and comprehensive to both trialists, physicians and patients.

The definition is intended to provide consistency in defining the HBR population for clinical trials, to complement regulatory review and ultimately, clinical decision-making. The proposed ARC-HBR consensus definition represents the first pragmatic approach to a consistent definition of HBR in clinical trials evaluating safety and effectiveness of devices and drug regimens for patients undergoing PCI. It is hoped that use of the ARC-HBR definitions and trial design considerations may facilitate faster access to safer devices, drugs and optimal care for this vulnerable patient population.

Introduction

As interventional cardiologists, we are treating increasingly complex patients with PCI in every day clinical practice – including older patients, with more comorbidities, who previously may have been managed conservatively [1, 2]. Patients with risk factors for bleeding after PCI represent a growing subset among such patients. Although many risk factors for bleeding in patients undergoing PCI are well established, a standardised definition of high bleeding risk (HBR) patients was only recently proposed [3].

The optimal management of HBR patients undergoing PCI remains unknown because such patients have been excluded from the vast majority of randomised clinical trials investigating stent and antithrombotic strategies. Even so-called ‘all-comers’ trials have largely excluded patients deemed unsuitable for guideline-recommended antithrombotic therapy [4, 5, 6, 7, 8, 9].This initially led to the use of bare metal stents (BMS) in preference to drug-eluting stents (DES) in patients perceived to be at HBR to facilitate shorter DAPT-durations. While this is no longer the recommended practice in Europe, it remains common practice elsewhere.

In Europe, guidelines for clinical practice recommend the use of DES over BMS across the spectrum of patients undergoing stent implantation, irrespective of the clinical presentation, lesion type, planned non-cardiac surgery, anticipated DAPT duration, or concomitant anticoagulant therapy (class of recommendation [COR] I, level of evidence [LOE] A) [10]. However, US guidelines for clinical practice do not currently give such a recommendation.

In addition, European guidelines no longer differentiate between BMS and DES with respect to DAPT-duration: in non-HBR patients who undergo stent implantation, 6 months DAPT is recommended in stable coronary artery disease (CAD) and 12 months is recommended in ACS, irrespective of stent type (both COR I, LOE A) [11]. In contrast, in US clinical practice guidelines, the recommended DAPT durations in stable CAD differ for BMS and DES: after BMS implantation, a minimum of one month DAPT is recommended (COR I, LOE A), whereas after DES implantation, a minimum of six months DAPT is recommended (COR I, LOE B-RSR) [12]. In ACS patients, a minimum of 12 months DAPT is recommended, irrespective of the stent type (COR I, LOE B-R), consistent with European guidelines. Such differences inevitably result in preferential use of BMS in scenarios where shorter DAPT would be preferable, such as in patients at HBR. As recently as 2014, BMS accounted for 20% of implanted coronary stents in the US [14].

Furthermore, recommendations regarding DAPT-duration in HBR patients in both guidelines are largely based on expert consensus due to the relative paucity of available data. In stable CAD, European guidelines recommend that 3 months DAPT should be considered in patients considered to be at HBR (COR IIa, LOE B) and 1 month DAPT may be considered if 3 months DAPT poses safety concerns (COR IIb, LOE C) [11]. US guidelines address HBR patients treated with DES only, also recommending that DAPT discontinuation after 3 months may be reasonable in stable CAD (COR IIb, LOE C-LD), but in contrast to European guidelines, there is no recommendation for 1 month DAPT in such patients [12]. In patients with ACS, updated European guidelines recommend consideration of 3 months DAPT in HBR patients (defined as a PRECISE-DAPT score ≥ 25 or ARC-HBR criteria fulfilled) (COR IIa, LOE B).[13]In contrast, US guidelines recommend consideration of 6 months DAPT in such patients (COR IIb,LOE C-LD).

A total of five randomized trials of devices specifically enrolling HBR patients have been completed [15, 16, 17, 18, 19]. However, due to the lack of a standardized definition of HBR until recently, these trials have used a broad range of definitions for HBR, resulting in significant heterogeneity with respect to the included patient populations and their actual bleeding risk. This limits both interpretation of studies, cross-trial comparisons and pooling of data. Against this background, the Academic Research Consortium for high bleeding risk (ARC-HBR) proposed a standardised definition of HBR patients intended for use in clinical trial design as well as in clinical practice [3] and a consensus document regarding trial design considerations for trials enrolling such patients.[20]

FOCUS BOX 1 - High Bleeding Risk patients

High Bleeding Risk (HBR) patients are in many ways a “forgotten population” and have until recently been excluded from or under-represented in the majority of drug or device trials in patients undergoing PCI.

Given that HBR patients now constitute a growing proportion of PCI candidates, it has become urgent to create a standardised definition for this heterogeneous group, both for clinical trial design and for use at the bedside.

‘ALL-COMERS’ TRIALS IN PATIENTS UNDERGOING PCI EXCLUDE PATIENTS AT HBR

In general, trials of stents or antithrombotic therapies in patients undergoing PCI tend to be highly selective with respect to patient characteristics, including only patients considered suitable for guideline-recommended DAPT. Key patient-related characteristics and exclusion criteria in pivotal randomized trials of DES submitted for FDA review are shown in Table 1. Even recent DES trials have failed to include patients with prior bleeding, haematological abnormalities, recent stroke, and advanced renal impairment ( Table 1) [4, 5, 6, 7, 8, 9]. Furthermore, while trials claiming to include ‘all-comers’ tend to be more inclusive, patients at HBR continue to be under-represented and details on the proportions of patients with bleeding risk factors tend to be under-reported [21, 22, 23, 24, 25, 26, 27]. Results of such trials are, therefore, not generalisable to patients at HBR.

HETEROGENEITY IN TRIAL DESIGN BETWEEN COMPLETED AND ONGOING TRIALS IN PATIENTS DEEMED TO BE AT HBR

A number of trials in patients at HBR have been completed or are ongoing. These trials largely use exclusion criteria from ‘all-comers’ trials as inclusion criteria. Nonetheless, inclusion and exclusion criteria in such trials differ, resulting in enrolment of heterogeneous populations with varying levels of bleeding risk and thus, varying bleeding event rates. Such differences underline the need for standardisation of inclusion and exclusion criteria for trials of patients at HBR. These trials also differ with respect to other elements of their design, including the investigational and control therapies, timing of randomisation, the proposed antiplatelet regimens, the primary hypothesis tested, and the timing of the primary analysis, further highlighting the need for consensus definitions to guide the design of trials in patients at HBR undergoing PCI. Completed and ongoing trials in patients at HBR are discussed below and trial designs and inclusion criteria are shown in Table 2 and Table 3 respectively.

Completed trials in patients considered to be at HBR

To our knowledge, five randomised trials of patients at HBR undergoing PCI have been completed to date [15, 16, 17, 18, 19]. All compared two coronary devices with the same DAPT duration in both groups. Three trials compared a DES (investigational arm) to a BMS (control arm), one trial compared two DES platforms, and one trial compared a drug-coated balloon to a BMS, all on a background of one month of DAPT (with the exception of patients after unstable presentation in SENIOR, in whom 6 months DAPT was recommended).

The first and largest of these trials was LEADERS-FREE (n=2,466), which compared the polymer-free biolimus-eluting stent (BioFreedom, Biosensors Europe, Morges, Switzerland) to a similar backbone BMS. Patients were required to fulfill ≥1 bleeding risk inclusion criteria and on average, fulfilled 1.7 criteria [15].

The ZEUS trial compared the durable polymer zotarolimus-eluting stent (Endeavour Sprint, Medtronic Vascular, Minneapolis, MN, USA) to thin-strut BMS in “uncertain DES candidates” [28], with a pre-specified subgroup analysis of 828 patients who met ≥1 criteria for high bleeding risk (ZEUS-HBR) [16]. The investigational stent in the ZEUS trial is no longer available for clinical use.

The SENIOR trial (n=1,200) compared a bioresorbable polymer everolimus-eluting stent (Synergy, Boston Scientific, Marlborough, MA, USA) with a similar backbone BMS in elderly patients. Other than age, there were no pre-specified HBR inclusion criteria [17].

All three trials showed superiority of the investigated DES over BMS in HBR patients with respect to their primary endpoint, which differed among trials (shown in Table 2 ).

More recently, the ONYX ONE randomised trial (n=1,996) compared a current generation durable polymer DES (Resolute Onyx zotarolimus-eluting stent, Medtronic, Santa Rosa, CA, USA) to the polymer-free DES investigated in LEADERS FREE (BioFreedom biolimus-coated stent, Biosensors Europe, Morges, Switzerland) in patients deemed to be at HBR, defined using identical inclusion criteria to those in LEADERS FREE ( Table 2 ).[18] As was the case in LEADERS FREE, patients fulfilled an average of 1.7 criteria. The durable polymer DES was found to be non-inferior to the polymer-free DES with respect to the primary endpoint, which was identical to one of the co-primary endpoints in LEADERS FREE.

Finally, the DEBUT trial (n=208) compared the SeQuent Please paclitaxel-coated balloon (B. Braun Melsungen AG, Berlin, Germany) to BMS in de novo coronary stenoses in HBR patients.[19] DCB-angioplasty was found to be superior to BMS implantation with respect to the primary endpoint at 9 months. The primary endpoint and timing of the primary analysis in this trial differed to that in all other completed trials in HBR patients.

Four single-arm studies investigating short DAPT durations after stenting in patients at HBR have been completed: LEADERS FREE II compared the investigational stent from LEADERS FREE to the historical BMS control from the LEADERS FREE trial in patients enrolled in the US and showed consistent results.[29] Inclusion criteria were identical to those in the LEADERS FREE trial. In the other single-arm studies (EVOLVE SHORT DAPT,[30] XIENCE 28 [NCT03355742], and XIENCE 90 [NCT03218787]), inclusion criteria differed to those used in randomised trials in HBR patients. Moreover, although pre-specified performance goals were met (non-inferiority to propensity-matched historical controls) in all three studies with respect to thrombotic events, the results of these studies must be interpreted with caution. Patients were eligible for inclusion only if they remained free from ischaemic events and compliant with DAPT throughout the protocol-specified DAPT period. Accordingly, rates of definite or probable stent thrombosis and BARC 3-5 bleeding at one year in these studies were markedly lower than those observed in randomised trials in HBR patients, reflecting the exclusion of early events on DAPT and indicating inclusion of lower risk patients in the single-arm studies.

Ongoing randomised trials in patients at HBR

At present, there are a number of ongoing randomised trials Table 2 and Table 3 in patients at HBR. One trial is comparing two stents with differing DAPT durations, two trials are investigating shorter versus longer DAPT durations after PCI with the same DES in both treatment arms and two are comparing different stents with similar DAPT in both treatment arms.

The COBRA-REDUCE trial (n=996) is comparing two device and drug strategies in patients on OAC (VKA or NOAC). Strategy one is an investigational stent (the COBRA PzF, thin strut stent, coated with a nano-thin layer of the polymer Polyzene F, with no active drug-elution; Celonova Biosciences, San Antonio, TX, USA) with an investigational DAPT duration (14 days DAPT consisting of aspirin plus clopidogrel). Strategy two is a FDA-approved DES with guideline-recommended DAPT duration (3 or 6 months). The only pre-specified HBR criterion is OAC therapy. There is a bleeding co-primary endpoint, being tested for superiority and a thrombotic co-primary endpoint, being tested for non-inferiority: BARC ≥2 bleeding after hospital discharge or after 14 days (whichever is earlier) until 6 months post-PCI and the composite of all-cause death, myocardial infarction, definite or probable stent thrombosis, and ischaemic stroke 6 months post-PCI.

MASTER DAPT (NCT03023020) and TARGET SAFE (NCT03287167) are both investigating regimens of shorter versus longer DAPT duration after PCI. However, both trials differ markedly in their design. MASTER DAPT is comparing one month with 6-12 months DAPT after PCI with a biodegradable polymer sirolimus-eluting stent (SES, UltimasterTM TANSEITM, Terumo Corporation, Tokyo, Japan). TARGET SAFE is comparing one month to six months DAPT after PCI with a biodegradable polymer SES (Firehawk TM, Shanghai MicroPort Medical Group). Randomization occurs one month after PCI in MASTER DAPT and at the time of PCI in TARGET SAFE. In MASTER DAPT, participants must fulfil ≥1 ‘HBR’ inclusion criterion ( Table 3 ). An interim analysis has been published showing the distribution of these criteria in the first half of patients randomised [31]. Patients with a BARC ≥2 bleeding event, MI, symptomatic restenosis, stent thrombosis, stroke, or revascularization prior to the randomization visit are excluded.

In TARGET-SAFE, patients must be considered suitable for 6 months DAPT and meet at least one HBR inclusion criterion ( Table 3 ). OAC is an exclusion criterion. The trials also differ with respect to their primary endpoints ( shown in Table 2 ).

The Bioflow-DAPT trial (NCT04137510) and the COMPARE 60/80 HBR trial (NCT04500912) are both comparing two DES on a background of identical DAPT in both treatment groups ( Table 2). However, the background DAPT durations and primary hypotheses differ between the trials. Bioflow-DAPT is comparing the Orsiro Mission biodegradable polymer sirolimus-eluting stent (BIOTRONIK, Bülach, Switzerland) to the Resolute Onyx durable polymer zotarolimus-eluting stent followed by one month DAPT. Patients must meet at least one HBR inclusion criterion ( Table 3 ). The COMPARE 60/80 HBR trial is comparing two biodegradable polymer sirolimus-eluting stents with different strut thicknesses: the ultrathin (60 micron) Supraflex Cruz stent (Sahajanand Medical Technologies, Mumbai, India) and the Ultimaster Tansei stent (80 micron), followed by guideline-recommended DAPT (according to ESC guidelines). Patients are eligible for inclusion if ≥ 1 major or ≥ 2 minor ARC-HBR criteria are met.

These differences across all aspects of trial design highlight the unmet need for consensus definitions to guide the design of trials investigating devices and drugs in HBR patients, in order to promote consistency.

Inclusion criteria in completed and ongoing HBR randomised trials

Inclusion criteria in completed and ongoing HBR randomised trials are shown in ( Table 3 ). Age and indication for OAC are the two most commonly fulfilled HBR criteria across completed trials in HBR patients ( Table 2 ). Age is the most frequently met criterion in LEADERS FREE, ZEUS-HBR, SENIOR, and ONYX ONE (in 64%, 51%, 100%, and 62% of patients, respectively) and the interim analysis of the ongoing MASTER DAPT trial (almost 70% of patients). It is the second most commonly met criterion in the DEBUT and COBRA-REDUCE trials, in just over half of patients in both trials. However, the lower cut-off for age differs between trials (≥75 years in LEADERS FREE, SENIOR, and ONYX ONE versus >80 years in ZEUS-HBR and DEBUT).

Indication for OAC is the most frequently fulfilled criterion in the COBRA-REDUCE and DEBUT trials (in 100% and 60% of patients, respectively) and the second most frequently met characteristic in LEADERS FREE, ZEUS-HBR, SENIOR, and ONYX ONE (in 36%, 38%, 18%, and 39% of patients, respectively). In MASTER DAPT, OAC is the third most frequently met characteristic, present in over one third of patients in the interim analysis. Of all completed or ongoing randomised trials in HBR patients at this time, TARGET SAFE is the only trial where indication for OAC is not a HBR criterion.

Renal impairment is the third most commonly fulfilled criterion in LEADERS FREE (19%) and COBRA-REDUCE (in over one third of patients), and the fourth most common criterion in ONYX ONE (15%). This was not a pre-specified criterion for HBR-status in ZEUS-HBR, nor is it a criterion in MASTER DAPT but is in in all other ongoing trials. Moreover, in trials where renal impairment is a criterion, differing definitions are used: in LEADERS FREE and ONYX ONE, the definition used is creatinine clearance <40 ml/min, compared with eGFR <30 ml/min in COBRA-REDUCE, Bioflow-DAPT and COMPARE 60/80 HBR and eGFR <60 in TARGET-SAFE.

Early planned surgery was a HBR inclusion criterion in LEADERS FREE and ONYX ONE (met in 16% and 7% of patients, respectively) but this was an exclusion criterion in ZEUS-HBR and SENIOR and is a pre-specified criterion in two of five ongoing randomised trials. Prior haemorrhagic stroke was also an inclusion criterion in LEADERS FREE and ONYX ONE but was an exclusion criterion in SENIOR, and although this was not an exclusion criterion in ZEUS-HBR, no information on its prevalence is provided. Prior stroke is a HBR criterion in all ongoing trials other than COBRA-REDUCE.

Although a PRECISE-DAPT score of ≥25 is the second most commonly met HBR criterion in MASTER DAPT, only one other trial (Bioflow-DAPT) uses this as an inclusion criterion. Finally, TARGET SAFE is the only trial to use BMI (<18.5 kg/m2), gender (female with acute coronary syndrome), congestive cardiac failure with reduced ejection fraction (30-50%), HAS-BLED score (≥3) or gastric ulcers as HBR inclusion criteria.

These differences illustrate the significant heterogeneity in the definitions of HBR used in completed and ongoing trials. This is reflected in the heterogenous rates of BARC 3-5 bleeding at one year in completed HBR trials (7.2%, 4.2%, and 4.7% in patients treated with one-month DAPT in LEADERS FREE, ZEUS-HBR, and ONYX ONE, respectively, and circa 3.5% in the SENIOR trial in patients treated with one to six months DAPT). Although one-year outcomes are not yet available for the COBRA-REDUCE trial, the rate of BARC 3-5 bleeding at six months was extremely high, at > 6.0%. The marked differences across completed and ongoing trials highlight the need for a standardised definition of HBR for use in clinical trials.

EXISTING BLEEDING RISK SCORES

There are a number of risk scores to predict long-term bleeding risk in patients taking antiplatelet therapy [32, 33, 34, 35, 36]. One such score (DAPT) is not purely a bleeding prediction score, rather it is a tool to predict the benefit versus harm of prolonged DAPT (> one year) in patients after PCI. Characteristics of existing bleeding risk scores are shown in shown in Table 4.

Existing bleeding risk scores differ with respect to their derivation populations, their variables (and their definitions), and the type and severity of bleeding that they predict. With respect to derivation populations, three scores were derived in populations of patients undergoing PCI [32, 34, 35], one was developed in patients who were free from major bleeding or ischaemic events 12 months post-PCI [37], and two were created in non-PCI populations [24][24, 24][27]. With respect to variables, only one variable is common to all scores (age) and the age cut-offs predictive of bleeding as well as their relative weights vary between risk scores. In contrast, a number of variables feature in only one score (prior malignancy, congestive heart failure, BMI <25 or ≥35 kg/m2, hypercholesterolemia, and elevated white cell count) and all scores omit certain variables known to be associated with HBR, either because of low prevalence (e.g. severe liver disease, bleeding diatheses, and thrombocytopenia) or non-reporting in the derivation datasets (e.g. history of cancer or prior bleeding, use of non-steroidal anti-inflammatory drugs [NSAIDs], or planned surgery), or because their significance is underestimated due to collinearity with other variables tested.

Although the 2017 European Society of Cardiology (ESC) focused update on DAPT in coronary artery disease recommended that use of risk scores such as the PRECISE-DAPT and DAPT scores may be considered to guide antiplatelet therapy after PCI (Class IIb recommendation, level evidence A) [11], existing scores are of limited use for predicting bleeding in patients at HBR. To date, no score has been developed or validated in a HBR population and the accuracy of existing scores for prediction of bleeding even in the development cohorts was only moderate (C-statistics are shown in Table 4).

ARC-HBR DEFINITION OF HBR PATIENTS

The Academic Research Consortium for high bleeding risk (ARC-HBR) recently proposed a definition of high bleeding risk for use in clinical trials of patients undergoing PCI [3]. The definition consists of clinical criteria associated with increased bleeding risk categorised into major and minor criteria, depending on the associated risk of bleeding at one year. A major criterion for ARC-HBR is any criterion associated with a rate of BARC 3 or 5 bleeding of ≥4% or intracranial haemorrhage of ≥1% at 1 year and a minor criterion is any criterion associated with an increased risk of bleeding of <4% at 1 year [38]. The bleeding cut-offs for the definition are arbitrary, based on the consensus of the group, taking into account 1-year rates of moderate/major bleeding varying between 0.3% and 2.8% in 11 trials of DAPT duration in non-HBR populations ( Figure 1) [35, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48], as well as in trials enrolling HBR patients (BARC 3-5 bleeding rates of 7.2% in LEADERS FREE and 4.2% in ZEUS-HBR) [11, 12]. The criteria comprising the definition are shown in Figure 2 and Table 5. Given the limited amount of data available, which was often largely derived from registries, a pragmatic consensus-based approach was chosen to define major and minor HBR criteria, based on clinical characteristics. Moreover, although the relationship between many of the criteria and bleeding is continuous, binary criteria were used to simplify the definition and to facilitate its use in trial enrolment.

FOCUS BOX 2 - ARC-HBR major and minor criteria

Age

  • Age ≥75 years is considered a minor criterion for ARC-HBR.

Elderly patients, who represent the fastest growing patient subgroup undergoing PCI [1, 2], tend to have more comorbidities and co-existing risk factors for bleeding compared with younger patients [49]. Therefore, while bleeding risk increases with age, the increase is partially explained by confounding due to such factors. In the SENIOR trial, which included patients aged ≥75 years (mean age 81.4 ± 4.2), treated with one or six months DAPT after PCI, the rate of BARC 3-5 bleeding at one year was approximately 3.5%. A sub-group analysis of elderly patients (≥75 years) in the LEADERS FREE trial showed that patients who qualified for inclusion on the basis of age alone (n=562) had a similar rate of BARC 3-5 bleeding, which was lower than that of the overall elderly population (3.2% versus 7.8%, respectively) [50]. Despite the known confounding effect of co-existing bleeding risk factors, age remained an independent predictor of bleeding after adjustment for such factors in the development cohorts of all but one (CRUSADE) [51] bleeding risk score in PCI patients [51, 51, 51, 51, 51, 51, 51]. Although the relationship between age and bleeding risk is continuous, a pragmatic decision to use a binary variable in the ARC-HBR definition was made.

Renal disease

  • Severe (estimated glomerular filtration rate [eGFR] <30 mL/min) or end-stage (eGFR <15 mL/min or renal replacement therapy) chronic kidney disease (CKD) is considered a major criterion for ARC-HBR.
  • Moderate CKD (eGFR 30-59 mL/min) is considered a minor criterion for ARC-HBR.

Approximately 30% of patients undergoing PCI have at least moderate CKD [56]. Even mild CKD is an independent risk factor for bleeding after PCI [57, 58], and the risk increases with increasing severity of CKD () [57, 58, 59, 60, 61]. Patients with severe CKD tend to be excluded from randomised trials. In the LEADERS FREE trial, renal insufficiency (defined as creatinine clearance <40 ml/min) was a HBR inclusion criterion [15]. In the 19.1% of patients who met this criterion, the one year BARC 3-5 bleeding rate was 10.6% [3]. In the PRECISE DAPT bleeding risk score [35], isolated severe CKD places patients in the highest quartile for bleeding risk, whereas milder levels of CKD are associated with a low-to-moderately increased bleeding risk. Although the association between CKD and bleeding is most likely continuous, CKD stages rather than eGFR as a continuous variable were used in the ARC-HBR definition based on the available data ( Table 6).

The balance of bleeding and ischaemic risk is particularly sensitive in patients with CKD because the associated increase in bleeding and ischaemic risk are almost identical ( Table 6), which explains why CKD is not included as a variable in the DAPT score (a score to predict the difference between the anticipated reduction in ischaemic events and the anticipated increase in bleeding events with prolonged DAPT beyond 1 year after PCI) [37] in contrast with scores that purely assess long-term bleeding risk after PCI [32, 34, 35]. CKD was found to be almost equally predictive of ischaemic (HR 1.55 [95% CI 1.03-2.32]) and bleeding (HR 1.66 [1.04-2.66]) events and thus, did not help to identify discordant bleeding and ischaemic risk.

Cirrhosis with portal hypertension

  • Cirrhosis with portal hypertension is considered a major ARC-HBR criterion.

The reported prevalence of cirrhosis in patients undergoing PCI in the US is 1.2% [62]. The finding of obstructive coronary disease during transplant work-up in patients with end-stage liver disease is an increasingly common scenario: a single centre study of liver transplant patients in the US reported that 4.7% of patients in total over a ten-year period underwent PCI prior to transplant, with an increase in the rate over time [63]. Such patients are usually excluded from randomised trials. Although severe chronic liver disease was an inclusion criterion for high bleeding risk in the LEADERS FREE trial, only 0.9% of enrolled patients met this criterion [15]. The rate of BARC 3-5 bleeding at one year in this subgroup was 14.3% [3].

Although Child-Pugh and Mayo End-Stage Liver Disease (MELD) criteria are used as exclusion criteria in some DES and DAPT trials, these scores have not been validated for prediction of bleeding risk [64, 65, 66]. A large analysis of the US Nationwide Inpatient Sample (NIS) registry showed that liver disease was an independent predictor of in-hospital gastrointestinal bleeding in patients undergoing PCI (OR 2.59, 95% CI 2.22-3.02, p<0.001) [67]. Another NIS analysis showed that the most common post-procedural complications in such patients were hemorrhage (in 6.6%) and transfusion (in 11.3%) [62]. A retrospective study of patients with cirrhosis and CAD treated by either coronary stenting and DAPT or medical therapy with aspirin-monotherapy, showed gastrointestinal bleeding rates of 22% vs. 5%, respectively (p=0.003) [68]. An observational study of patients with chronic hepatitis B virus showed significantly higher rates of International Society on Thrombosis and Haemostasis (ISTH) [69, 70] major bleeding or clinically-relevant non-major bleeding in patients taking antiplatelet therapy compared with those without (9.5% vs. 1.8%, HR 3.28, 95% CI 1.98‐5.42; p<0.001) [71].

Active cancer

  • Active malignancy* (excluding non-melanoma skin cancer) is considered a major ARC-HBR criterion.

*Active malignancy is defined as diagnosis within the previous 12 months and/or ongoing active cancer treatment (surgery, radiotherapy, chemotherapy or immunotherapy) and does not include cancer that is considered to be in complete remission, or requiring only maintenance therapy (e.g. hormonal therapy for breast or prostate cancer).

The prevalence of active or previous cancer in patients undergoing PCI is increasing, with a prevalence of 9.5% in the US [72, 73]. Active cancer is associated with higher rates of in-hospital bleeding compared with both previous cancer and no history of cancer (9.7% vs. 4.2% vs. 3.1%; OR [active vs. no cancer] 1.92, 95% CI 1.82-2.04 and OR [previous vs. no cancer] 1.08, 95% CI 1.03-1.13) and ranged between 4.9% and 21.2% according to the type, site, and spread of the cancer [72]. Active cancer is also associated with a higher rate of readmission for bleeding within 90 days of PCI compared with no history of cancer (highest with colon cancer, at 4.2%, compared with 0.6% with no cancer), with the average time to bleeding readmission ranging from 38.2 to 42.7 days, depending on the type of cancer.[73]

The LEADERS FREE trial included 239 (9.7%) patients with non-skin cancer diagnosed or treated within 3 years prior to the index PCI [15]. The rate of BARC 3-5 bleeding at one year in this subgroup was 9.6% [3]. In an observational study of patients ≥65 years undergoing PCI, history of cancer was an independent predictor of late bleeding (HR 1.80, 95% CI 1.09–2.96, p=0.02), with 5.0% of patients being hospitalised for bleeding within 1 year after discharge [74].

In patients in the TRILOGY ACS trial newly diagnosed with cancer after PCI for ACS [75], GUSTO severe/life-threatening or moderate bleeding occurred substantially more frequently compared with patients without cancer (11.2 vs. 1.5%).

Anaemia

  • A haemoglobin level <11 g/dL is considered a major ARC-HBR criterion.
  • A haemoglobin level of 11-12.9 g/dL for men and 11-11.9 g/dL for women is considered a minor ARC-HBR criterion.

Anaemia is common in patients undergoing PCI, with a reported prevalence of WHO-defined anaemia (haemoglobin <13 g/dL in men and <12 g/dL in women) of 21.6% in the Bern DES Registry, 30.1% in the CREDO-KYOTO registry cohort-2, and 40% in an observational study in the US [76, 77, 78]. Anemia in patients undergoing PCI is associated with increased risk of subsequent bleeding. The one-year risk of BARC 3 or 5 bleeding in ACS patients who underwent PCI in the RENAMI registry was 5.4% in patients with WHO-defined anemia compared with 1.5% in those without (p=0.001) [79]. A meta-analysis including >230,000 patients undergoing PCI found anemia (WHO-defined in the majority of studies) to be associated with a more than two-fold risk of subsequent bleeding and bleeding risk increased with increasing severity of anemia [80].

Baseline anemia was found to be a strong predictor of bleeding in the development cohorts of a number of bleeding risk scores [32, 33, 34, 35]. In BleeMACS, haemoglobin <11 g/dL was the strongest predictor of serious spontaneous bleeding at one year (adjusted HR 2.41, 95% CI 1.29-4.50, p<0.001) and haemoglobin of 11.0-13.9 was also associated with a significantly increased bleeding risk (adjusted HR 1.59, 95% CI 1.14–2.21, p=0.006) compared with haemoglobin ≥14 g/dL [32]. In PARIS, baseline anaemia (haemoglobin <12 g/dL in men and <11 g/dL in women) was associated with a two-year BARC 3 or 5 bleeding rate of 9.5% compared with 2.7% in patients without anaemia (adjusted HR 2.72, CI 1.83-4.04, p<0.0001) [34].

Thrombocytopenia

  • Moderate or severe baseline thrombocytopenia* (platelet count <100 x 10[9]/L) is considered a major ARC-HBR criterion.

*Baseline thrombocytopenia is thrombocytopenia that is present prior to PCI (as opposed to acquired thrombocytopenia post-procedure).

The reported prevalence of baseline thrombocytopenia in patients undergoing PCI is approximately 2.5% in the US and 1.5% in Japan [81, 82]. Thrombocytopenia is classified as mild (100-149 x 109/L), moderate (50-99 x 109/L), or severe (<50 x 109/L) [83]. Even mild baseline thrombocytopenia is a risk factor for bleeding after PCI and bleeding risk increases with increasing severity of thrombocytopenia. Patients with moderate or severe thrombocytopenia are rarely represented in randomized trials of DES and DAPT as a platelet count <100 x 109/L is a common exclusion criterion. Although thrombocytopenia (< 100 x 109/L) was a HBR criterion in the LEADERS FREE trial [15], this criterion was fulfilled in only 1.6% of patients. The rate of BARC 3-5 bleeding at one year in this subgroup was 5.3% [3].

A post-hoc analysis of STEMI patients who underwent PCI in the HORIZONS-AMI trial showed a higher rate of 30-day ACUITY-HORIZONS major bleeding in patients with baseline mild thrombocytopenia compared with those without (15.4% vs. 9.1%, p=0.01) [84]. A large analysis from the NIS registry database found that in-hospital post-procedural bleeding (defined by International Classification of Diseases codes for in-hospital complications) was significantly higher in patients with baseline thrombocytopenia compared with propensity matched patients without thrombocytopenia (10.9% vs. 4.9%, odds ratio [OR] 2.40, 95% CI 2.05-2.72, p<0.0001) [81]. A Japanese study showed increased 3-year rates of GUSTO moderate or severe bleeding in PCI patients with baseline mild thrombocytopenia (9.9% vs. 6.9%, adjusted HR 1.20, 95% CI 1.03-1.40, p=0.02) and moderate/severe thrombocytopenia (23.1% vs. 6.9%, adjusted HR 2.35, 95% CI 1.80-3.08, p<0.001) compared with non-thrombocytopaenic patients [82].

Stroke, intra-cranial haemorrhage, brain arteriovenous malformation

  • Moderate or severe ischaemic stroke (defined as National Institutes of Health Stroke Scale [NIHSS] score ≥5 on presentation) within 6 months prior to PCI, previous intra-cranial hemorrhage (ICH) at any time, and the presence of a brain arteriovenous malformation (bAVM) are all considered major ARC-HBR criteria.
  • Ischaemic stroke at any time not meeting the major criterion is considered a minor ARC-HBR criterion.

Approximately 5-6% of patients undergoing PCI have a history of stroke [85]. Patients with a prior stroke within 6 months of enrolment were excluded from pivotal trials resulting in FDA-approval of DES ( Table 7 ). Although any stroke during the prior 12 months and any prior intra-cerebral bleed were HBR criteria in LEADERS-FREE, only 1.6% and 1.3% of patients met the criteria for stroke and prior ICH [15]. Respective rates of BARC 3-5 bleeding in these subgroups were 2.6% and 6.1% [3]. Rates of ICH at one year were not reported. In ZEUS-HBR, 8% of patients fulfilled the criteria for prior stroke or transient ischaemic attack (TIA) and although ICH was not an exclusion criterion, its prevalence was not reported [16]. In SENIOR, approximately 8% of the enrolled population had previous ischaemic stroke whereas prior ICH was an exclusion criterion [17]. Subsequent rates of bleeding or ICH were not reported in these subgroups.

Trials of DAPT in ACS patients have also included patients with prior ischaemic stroke/TIA but not those with prior ICH [86, 87, 88]. The TRITON-TIMI 38 trial compared DAPT with prasugrel vs. clopidogrel in ACS patients after PCI and resulted in a contraindication for prasugrel use in patients with prior TIA or stroke [88]. In the prasugrel arm, patients with prior TIA or stroke (>3 months prior to inclusion), had significantly higher rates of both ischaemic stroke and ICH at 15 months (6.5% and 2.3%, respectively) compared with patients without prior TIA/stroke (0.9% and 0.2%, respectively) [89]. In contrast, such an effect was not seen in patients in the clopidogrel arm, where rates of subsequent stroke and ICH were 1.2% and 0%, respectively, in patients with prior TIA/stroke and 1.0% and 0.3%, respectively, in those without. The PLATO trial compared DAPT with ticagrelor vs. clopidogrel in ACS patients treated with PCI. DAPT-treated patients with prior stroke/TIA had significantly higher rates of ICH at one year compared with those without prior stroke/TIA: 0.8% vs. 0.2% (unadjusted HR 3.95, 95% CI 1.82-8.55, p=0.0005) but in contrast to TRITON-TIMI 38, this did not significantly differ between treatment groups [90]. Rates of protocol-defined non-CABG related major bleeding at one year were also higher in patients with prior stroke/TIA compared with those without (6.3% vs. 4.0%, adjusted HR 1.38, 95% CI 1.03-1.85, p=0.03).

In contrast, in the development of bleeding risk scores, rates of non-ICH bleeding did not significantly differ according to previous stroke status in patients undergoing PCI. In PRECISE-DAPT and PARIS, patients with and without prior stroke had similar rates of TIMI major/minor bleeding (HR 1.16, 95% CI 0.54-2.48, p=0.70) and BARC 3 or 5 bleeding (4.1% and 3.5%, p=0.66), respectively [34, 35].

Although patients with previous stroke/TIA undergoing PCI have been included in the DES and DAPT trials discussed, no information regarding stroke timing or severity has been reported. To this end, the results of six major randomized trials investigating potent antiplatelet therapy for secondary stroke prevention in non-PCI patients ( Table 7 ) with index stroke ranging from acute minor stroke or TIA (<12-24 hours; NIHSS <3-5) [91, 92, 93], to recent stroke (≤90-120 days) of unspecified severity [94, 95], to recent (≤180 days) symptomatic lacunar infarct [96], may provide some insight into the risks of ICH according to stroke timing and severity.

There is a paucity of data regarding DAPT and bleeding risk in patients with bAVMs. Patients with bAVMs have a risk of ICH [97]. In a patient-level meta-analysis of 2,525 patients with bAVM, the annual risk of first and recurrent ICH in the absence of antiplatelet therapy was 1.3% (95% CI, 1.0 to 1.7) and 4.8% (95% CI, 3.9 to 5.9), respectively [98]. A randomized trial found an annual risk of first ICH of 2.0% in patients with bAVM (n=223) who did not undergo intervention of same [99].

Chronic bleeding diathesis

  • The presence of a clinically significant chronic bleeding diathesis is considered a major ARC-HBR criterion.

For the purpose of the ARC-HBR definition, chronic bleeding diatheses include inherited or acquired conditions known to be associated with increased bleeding risk, including platelet dysfunction (thrombocytopenia is discussed separately), von Willebrand disease (prevalence of 1-2% in the general population), inherited or acquired clotting factor deficiencies (including factors VII, VIII [hemophilia A], IX [hemophilia B], and XI) or acquired antibodies to clotting factors [100, 101, 102]. Given the low prevalence of such conditions in PCI patients and the lack of data regarding their outcomes, weighting the differential bleeding risks associated with different bleeding diatheses (and their levels of severity) is beyond the scope of the ARC-HBR definition. The most reliable predictor of bleeding in patients with bleeding diatheses is a personal history of bleeding, which may be assessed using a bleeding questionnaire [103].

There is a paucity of data regarding bleeding rates in patients with bleeding diatheses undergoing PCI as they are generally excluded from randomised trials of DES and DAPT. In ZEUS-HBR, haematological disorders or any known coagulopathy-associated bleeding diathesis (including prior or current thrombocytopenia, defined as platelet count <100 x 109/L) was a criterion for HBR status. However, although this criterion was met in 95 (11.5%) patients, bleeding outcomes were not reported for this subgroup [16]. In a small series of patients with haemophilia A or B undergoing coronary angiography with or without PCI, 6% of patients had major peri-procedural bleeding and 20% of patients had a bleeding event (predominantly minor) within one year [104].

Prior bleeding or transfusion

  • Spontaneous (non-intracranial) bleeding requiring hospitalisation and/or transfusion in the 6 months prior to PCI (or at any time if recurrent) is considered a major ARC-HBR criterion.
  • A first spontaneous (non-intracranial) bleed requiring hospitalisation and/or transfusion between 6 and 12 months prior to PCI is considered a minor ARC-HBR criterion.

There is a scarcity of data concerning the risk of future bleeding in patients undergoing PCI who have had a prior bleeding event or previous blood transfusion. Hospital admission for bleeding during the prior 12 months was a HBR criterion in the LEADERS FREE trial [15]. Of the 3.2% of patients who fulfilled this criterion, 12.7% experienced BARC 3-5 bleeding at one year [3]. Prior spontaneous bleeding at any time was found to be an important predictor of future bleeding in development of the PRECISE-DAPT score: this variable in isolation places patients in the highest quartile for bleeding risk [35]. Two randomised trials have investigated continued antithrombotic therapy in patients presenting with peptic ulcer bleeding on aspirin-monotherapy. In one trial, one-year rates of recurrent ulcer bleeding were 8.6% vs. 0.7% in patients randomized to further treatment with clopidogrel vs. aspirin plus esomeprazole, respectively, after confirmed ulcer-healing (difference 7.9%; 95% CI 3.4 to 12.4, p=0.001) [105]. In the second trial, recurrent ulcer bleeding at 30 days occurred in 10.3% vs. 5.4% of patients allocated to aspirin plus pantoprazole vs. aspirin-discontinuation (HR 1.9, 95% CI 0.6 to 6.0, p=0.25) [106].

With respect to previous blood transfusion, in a randomised trial in non-PCI patients comparing transfusion strategies after acute upper GI bleeding, the rate of further in-hospital bleeding was significantly lower in patients allocated to a restrictive strategy (maintain haemoglobin >7g/dL) vs. a liberal (maintain haemoglobin >9 g/dL) transfusion strategy (10% vs. 16%, adjusted HR 0.68, 95% 95% CI 0.47-0.98, p=0.03) [107]. The fact that the highest rates of recurrent bleeding occurred in the setting of acute blood transfusion suggests that timing of transfusion may be an important determinant of bleeding risk.

OAC

  • The anticipated use of long-term OAC (with a VKA or NOAC)* after PCI is considered a major ARC-HBR criterion.

*Although bleeding risk may differ between VKAs and NOACs, between different NOACs and their various doses, exposure times and variations in renal function, weighting the relative bleeding risk with different OAC regimens is considered beyond the scope of the ARC-HBR definition.

The most common indication for OAC in patients undergoing PCI is co-existing atrial fibrillation (AF), with a reported prevalence of 14.6% in an analysis of >3 million patients in the NIS in the US [108]. A number of trials have evaluated antithrombotic therapy in patients on OAC who undergo PCI. In the WOEST trial, one-year rates of BARC 3-5 bleeding in patients on vitamin K antagonists (VKA) after PCI were well over 4% irrespective of whether patients received double (VKA plus clopidogrel) or triple (VKA plus aspirin and clopidogrel) therapy (6.5% vs. 12.7%, HR 0·49, 95% CI 0·28–0·86, p=0.011) [109]. This was also the case with 9-month BARC rates in patients on VKA undergoing PCI in the ISAR TRIPLE trial, irrespective of the duration of triple therapy received (approximately 11.1% vs. 10.4% with six weeks vs. six months triple therapy, respectively) [110]. Bleeding rates were markedly lower in patients treated with triple therapy including a VKA in trials comparing non-VKA oral anticoagulants (NOAC) with VKA, probably indicating a lower bleeding risk profile in the populations enrolled. Nonetheless, BARC 3-5 bleeding rates at one year in PIONEER AF-PCI were >4% in all treatment groups: 4.1% with dual therapy (P2Y12 inhibitor and low-dose rivaroxaban [15 mg daily]), 4.4% with triple therapy (aspirin, P2Y12 inhibitor, and very low dose rivaroxaban [2.5 mg twice daily]), and 7.9% with triple therapy (aspirin, P2Y12 inhibitor and a VKA) [111]. While BARC bleeding was not reported in RE-DUAL PCI, rates of TIMI major or minor bleeding at 14 months were 3.0% vs. 7.0% in patients treated with dual therapy (P2Y12 inhibitor and dabigatran 110 mg twice daily) vs. triple therapy (aspirin, P2Y12 inhibitor and warfarin), (HR 0.41, 95% CI 0.26-0.63, p<0.001) and 3.5% vs. 6.3% in those treated with dual therapy (P2Y12 inhibitor and dabigatran 150 mg twice daily vs. triple therapy (aspirin, P2Y12 inhibitor and warfarin), (HR 0.53, 95% CI 0.33-0.85, p=0.009) [112].

Chronic NSAID- or steroid-use

  • Chronic* steroid- or oral NSAID-use is considered a minor ARC-HBR criterion.

* Chronic use is defined as planned daily intake for ≥4 days/week.

Oral non-steroidal anti-inflammatory drugs (NSAIDs) and steroids are associated with a risk of gastrointestinal toxicity, which is dose-dependent and increases with chronic use [113][113, 113][114]. Although non-steroidal anti-inflammatory drugs (NSAIDs) represent the most widely used class of medications worldwide [115, 116], there is a paucity of data regarding bleeding risk in patients with chronic oral NSAID- or steroid-use after PCI because of underrepresentation or underreporting in randomized trials. While chronic NSAID- or steroid-use was a HBR criterion in both LEADERS FREE and ZEUS-HBR, only 2.8% and 3% of patients, respectively, fulfilled this criterion [15, 16]. The rate of BARC 3-5 bleeding at one year was 8.3% in this patient subgroup in LEADERS FREE [3]. Bleeding rates were not reported for this subgroup in ZEUS-HBR. Nonetheless, studies in non-PCI patients confirm that concomitant use of NSAIDs and aspirin substantially increase bleeding risk compared with aspirin-monotherapy [33, 117]. This risk may differ between selective and non-selective NSAIDs, however. A 514-patient randomised trial comparing celecoxib and naproxen (in addition to aspirin and esomeprazole) in patients on NSAIDs with an indication for aspirin, who presented with upper gastrointestinal bleeding, reported recurrent gastrointestinal bleeding rates of 5.6% and 12.3% at 18 months, respectively (HR 0.44, 95% CI 0.23-0.82, p=0.008) [118]. In a subgroup analysis of patients taking aspirin in another randomised trial comparing celecoxib to either ibuprofen or diclofenac, the rates of symptomatic upper GI ulcers or complications (bleeding, perforation, and obstruction) at 6 months were 4.7% and 6.0%, respectively (p=0.49) [119].

Planned major non-cardiac surgery post-PCI

  • Planned non-deferrable major surgery on DAPT after PCI is considered a major ARC-HBR criterion.

A number of important questions need to be considered in patients requiring surgery after PCI. First, what is the thrombotic risk determined by time since PCI as well as patient, lesion, and procedural characteristics? Second, can the surgery be deferred? Third, what is the bleeding risk associated with the surgical procedure?

The risk of stent thrombosis is highest within the first month after PCI, irrespective of stent type [120, 121], and the magnitude of this risk and its impact on mortality rate are inversely proportional to the timing of DAPT-discontinuation after PCI [122, 123]. As such, European guidelines for clinical practice give a class III recommendation (level of evidence B) for discontinuation of DAPT within the first month after PCI in patients undergoing elective non-cardiac surgery, irrespective of the stent type and in patients with high risk ischaemic features requiring DAPT, such as recent MI, elective surgery should be postponed for up to 6 months (class IIb recommendation, level of evidence C) [11]. US guidelines for clinical practice are somewhat stricter, giving a class III recommendation (level of evidence B-NR) for elective surgery within 1 month after BMS implantation and within 3 months after DES implantation in patients in whom DAPT will need to be discontinued perioperatively [12].

In patients with recent PCI undergoing surgery that is non-deferrable, the potential risks of DAPT-discontinuation in the potentially pro-thrombotic perioperative period must be balanced against the increased bleeding risk with DAPT [124, 125, 124, 125]. Aspirin-monotherapy in patients undergoing non-cardiac surgery significantly increases the risk of major bleeding compared with placebo [126]. DAPT in the perioperative period further increases this risk. For this reason, both European and US guidelines for clinical practice recommend deferral of elective surgery until after the recommended DAPT period [11, 12].

There is a paucity of data regarding bleeding risk in patients who undergo major non-cardiac surgery after PCI. Although such patients are generally excluded from randomised trials, planned major surgery within 12 months of PCI was an inclusion criterion for HBR in the LEADERS-FREE trial [15]. This criterion was fulfilled by 16.1% of patients, who had a BARC 3-5 bleeding rate of 7.3% at one year [3]. The generalisability of such a bleeding rate is questionable, however, given that post-operative bleeding rates depend on the bleeding risk specific to the type of surgery performed. With this in mind, the bleeding risks associated with specific surgeries have been summarised in a detailed national multidisciplinary expert consensus document from Italy regarding perioperative management of antithrombotic therapy in patients with previous PCI [127]. National multidisciplinary expert consensus documents concerned with balancing patient-specific ischaemic risk with procedure-specific bleeding risk have also been published by French and Spanish consortia [128, 129].

Recent trauma or surgery

  • Major surgery or major trauma within 30 days prior to PCI are considered major ARC-HBR criteria.

There is a paucity of data regarding bleeding rates in patients with recent major surgery or trauma who undergo PCI. The estimated bleeding risks associated with different types of surgery may provide some insight into the risk of bleeding in the setting of urgent PCI after major non-cardiac surgery [127], which may be required in up to 10% of patients on account of peri-operative myocardial infarction [130].

FOCUS BOX 3 - Academic Research Consortium- High-Bleeding Risk (ARC-HBR app)

This mobile application is a useful clinical tool to help evaluate patient at high-bleeding risk

Other considerations

A number of other potential factors for bleeding that were not included in the proposed criteria include ethnicity, frailty, and ACS. Ethnicity may play a role in bleeding risk: Asian patients, for example, seem to be at higher bleeding risk than non-Asian patients [131], perhaps partly due to confounding from factors such as lower body weight and differences in patients characteristics and comorbidity profiles. However, more research is needed in this field. While frailty may increase bleeding risk [132], there is a paucity of data regarding bleeding risk in such patients post-PCI and there is no consensus at this time on how best to define frailty. ACS was not included as a HBR characteristic as it does probably not increase bleeding risk per se; any associated HBR is largely a result of more potent and longer duration antithrombotic therapies used to mitigate the high thrombotic risk associated with ACS.

It is important to recognise that the ARC-HBR definition focuses purely on bleeding risk and its use is therefore not intended to guide therapeutic decisions but as inclusion criteria in trials of patients at HBR undergoing PCI, to advance the consistency and quality of data collected and reported, to support the interpretability of clinical trials and thus, the evaluation of drug and device therapies by physicians and regulatory bodies [133], as well as to aid in the recognition and identification of HBR patients in clinical practice.

Limitations

The proposed ARC-HBR definition has a number of important limitations, some of which have been alluded to above. First, the chosen cut-off values for one-year BARC 3 or 5 bleeding (4%) and ICH (1%) are arbitrary, based on expert opinion. Second, while many criteria used in the definition are well established as risk factors for bleeding in PCI patients, with robust data to support their use in the ARC-HBR definition, data regarding rates of BARC 3 or 5 bleeding or ICH at one year was not available for a number of criteria, in some cases due to low prevalence in PCI patients and in other cases because such data regarding the criteria is rarely collected or reported in clinical studies. In such cases, justification for their use in the definition is based on consensus decision alone.

Finally, the differential bleeding risks associated with the criteria were not weighted beyond major and minor due to a lack of data to support such an approach.

Validation studies

A number of validation studies of the ARC-HBR criteria have been published using data from observational PCI registries. These studies have consistently shown a high prevalence of patients meeting ARC criteria for HBR in real-world settings in Europe, South Korea, and the US, ranging from 35% to 44%.[134, 135, 136, 137] All studies confirmed a one-year BARC 3-5 bleeding rate of ≥4% in patients who meet ARC-HBR criteria and a stepwise increase in bleeding rates in the presence of increasing numbers of ARC-HBR criteria.

In addition, these studies all showed that patients who met criteria for ARC-HBR had more than three-fold higher rates of study-defined major bleeding compared with patients who did not meet the criteria. Three of these studies also investigated thrombotic outcomes in their datasets and consistently showed that the three-fold increase in bleeding events was paralleled by an approximate doubling in the rate of study-defined thrombotic events (shown in Figure 3).[138] This underscores the importance of the bleeding-ischaemia trade-off in these patients, who are not only at higher risk of bleeding but also at higher risk of thrombotic events compared with non ARC-HBR patients.

All validation studies to date are limited by the retrospective application of the criteria to existing datasets. This means that it was not possible to assess all ARC-HBR criteria in any dataset and in all datasets, some criteria assessed were only available in modified form. Full validation of the criteria will only be possible with prospective collection of such data.

Trial design considerations for patients at HBR undergoing PCI

To promote consistency across trials investigating devices and drugs in HBR patients undergoing PCI, the ARC-HBR recently published a consensus document proposing recommendations to guide the design of such trials.[20] The document proposes recommendations regarding the preferred design, control device and drug regimens, and primary endpoint definitions in HBR trials.

The consortium recommends that an ARC-HBR trial includes only patients considered to be at HBR according to ARC-HBR criteria (i.e., fulfilling ≥1 major criterion or ≥2 minor criteria). For trials of interventions in HBR patients using alternative inclusion criteria, reporting the proportion of enrolled patients according to ARC-HBR criteria is encouraged. Although trials focusing on only one subgroup of ARC-HBR patients (e.g., patients taking OAC) qualify as ARC-HBR trials, it should be acknowledged that findings may not be generalizable to all HBR patients.

Randomized clinical trials are the preferred design for trials in HBR patients. In device trials, current generation DES may be considered the preferred control device. In drug trials comparing DAPT durations or strategies, guideline-recommended DAPT may be considered the preferred control strategy. In trials comparing two different devices and antithrombotic therapy durations or intensities, a 2x2 factorial design may be preferable to allow ascription of benefit/harm to individual components of the test strategy.

Selection of the primary endpoint depends on whether the trial is investigating devices or drugs. To avoid confusion, the ARC-HBR recommends the terminology ‘thrombotic/ischaemic and bleeding’ endpoints rather than ‘efficacy and safety’ endpoints. When both thrombotic/ischaemic and bleeding outcomes are captured in HBR trials, it is proposed that they are considered separately as a thrombotic/ischaemic co-primary endpoint and a bleeding co-primary endpoint.

The recommended primary bleeding endpoint in ARC-HBR trials evaluating drugs or devices is BARC 3 or 5 bleeding. The use of BARC 2, 3 or 5 bleeding as a primary endpoint may be justifiable to allow a reduction in the sample size needed to achieve statistical power.

The recommended thrombotic/ischaemic endpoint differs depending on whether the trial is a device or drug trial. Because comparative evaluation of devices in HBR patients is usually tested with same antithrombotic therapy in both treatment arms, it is reasonable to use primary thrombotic/ischaemic endpoints that specifically evaluate device failure. Therefore, the ARC-HBR endorses the ARC-2 device-oriented composite endpoint definition (DOCE, also referred to as target-lesion failure), defined as the composite of cardiovascular death (modified to exclude death caused by bleeding as this is taken into account in the bleeding co-primary endpoint), MI not clearly attributable to a non-target vessel, or clinically driven TLR.[139] In drug trials in HBR patients, the ARC-HBR endorses the primary thrombotic/ischaemic composite endpoint of cardiovascular death (modified as for device trials), any MI, and ischaemic stroke to capture major adverse events possibly related to the study intervention. Inclusion of additional thrombotic or ischaemic events in the primary thrombotic/ischaemic composite endpoint of drug trials may be acceptable to facilitate a reduction in the sample size needed to achieve statistical power.

The thrombotic/ischaemic and bleeding primary endpoints proposed by the ARC-HBR are summarized in Figure 4. The definitions of the individual components of these primary endpoints are provided in Table 8.

In drug trials, the objective is to reduce bleeding events without increasing thrombotic events. As such, superiority testing with respect to the bleeding co-primary endpoint is preferred. In addition, such trials should be adequately powered to at least test for non-inferiority with respect to the thrombotic/ischaemic co-primary endpoint. In device trials, non-inferiority testing with respect to the thrombotic/ischaemic primary endpoint is acceptable.

Assessing the risks of bleeding versus thrombotic events in HBR patients after PCI

The ARC-HBR bleeding risk trade-off model was created to assess the risk of major bleeding and major thrombotic events in HBR patients within one year after PCI (excluding periprocedural events).[140] Patients enrolled in 6 PCI studies who met modified ARC-HBR criteria (n=6,641) were included. Independent predictors of BARC 3-5 bleeding and MI and/or definite/probable ST and the prognostic relevance of these events were investigated. The bleeding and thrombotic risk for each patient was also plotted on a scatter plot, allowing visualisation of their bleeding-thrombosis risk trade-off.

The rates of BARC 3-5 bleeding and MI and/or ST at one year in this HBR cohort were 5.7% and 5.3%, respectively. Four criteria were predictive of both bleeding and thrombotic events, four criteria were predictive of bleeding events only and four criteria were predictive of thrombotic events only ( shown in Table 9 ). Both bleeding and thrombotic risk models had moderate discrimination when applied to the derivation (C-statistics of 0.68 and 0.69, respectively) and validation cohorts (C-statistic of 0.74 for both models). Overall, 44% of patients had a higher risk of BARC 3-5 bleeding than MI and/or ST; 23% of patients had a higher risk of MI and/or ST than BARC 3-5 bleeding; and 32% had an equal risk of both. The risk of mortality was higher in patients after thrombotic events (HR 6.1) compared with bleeding events (HR 3.7).

The ARC-HBR bleeding risk trade-off model allows assessment of the bleeding-thrombosis trade-off in individual patients undergoing PCI to help tailor decisions regarding treatment and adjunctive antithrombotic therapies in HBR patients undergoing PCI. A smartphone application can be downloaded (shown in Focus Box 4) to facilitate decision-making using this trade-off score at the bedside.

FOCUS BOX 4 - Academic Research Consortium- High-Bleeding Risk (ARC-HBR bleeding vs. thrombosis trade-off app)

This mobile application is a useful clinical tool to help evaluate patient at high-bleeding risk

Future directions

If widely adopted, it is hoped that the proposed standardised definition of HBR patients and trial design considerations will represent a significant step forward. Nonetheless, we anticipate re-calibration of the initial ARC-HBR definition as more data regarding bleeding risk criteria are reported. In addition, the ability to assess the bleeding-thrombotic risk trade-off of individual patients has the potential to significantly impact patient care by guiding clinical decision-making in HBR patients with respect to antithrombotic therapy duration and strategy after PCI.

However, these are only the first steps towards evaluation of the optimal therapies in this challenging patient subgroup. Dedicated randomised controlled trials are urgently needed to investigate gaps in evidence with respect to HBR patients undergoing PCI. With this in mind, the following specific clinical questions have priority for further investigation. First, the optimal duration and intensity of DAPT in HBR patients has not been investigated. Although one month DAPT has been used in a number of completed HBR trials, this has not been compared with longer DAPT durations. Perhaps alternative DAPT durations may result in an improved risk-benefit ratio in specific subgroups of HBR patients. Second, single antiplatelet therapy and its effect compared with other secondary prevention strategies that do not increase the risk of bleeding warrants further study in HBR patients. Third, whether or not long-term secondary prevention of coronary artery disease with antiplatelet therapy results in net clinical benefit in HBR patients is unknown. Fourth, additional device studies should investigate whether specific stent platforms may provide meaningful clinical benefits in combination with specific antiplatelet therapy regimens. Fifth, alternative approaches to stenting, such as drug-coated balloon angioplasty in de novo disease should be compared with current generation DES. In addition, further trials of revascularization versus medical therapy in HBR patients are likely justified.

Conclusion

The ARC-HBR definition addresses an unmet need by providing a homogeneous definition for HBR patients undergoing PCI for use in clinical trials, to facilitate collection and reporting of comprehensive, consistent, and comparable data regarding treatment options for such patients. Moreover, the proposed ARC-HBR trial design considerations provide guidance to promote consistency in trial design. It is hoped that trialists and trial sponsors will use the ARC-HBR definition, trial design considerations and endpoint definitions for trials of HBR patients to achieve this goal, and in doing so will help to stimulate scientific progress, innovation and quality control initiatives.

Personal perspective

  • HBR patients represent about 40% of PCI candidates today, but their numbers are growing, and they have specific needs that must be addressed:
    • Full evaluation of the patient prior to a non-urgent intervention, in consultation with other specialists when needed. Important considerations include: is PCI the best option? Do the benefits of revascularisation outweigh the risks of post-PCI bleeding? Are further investigations and/or treatment needed prior to PCI?
    • Planning the procedure (including consideration of access site, contrast load, choice of stent, avoidance of complex interventions when possible, completeness of revascularisation, etc)
    • Adapting intensity and duration of antithrombotic regimen post-PCI
    • The ARC definition of HBR and proposed trial design considerations should facilitate the design of trials as well as the comparison and pooling of trial results, thereby helping clinicians make more informed therapeutic decisions.
    • The ARC-HBR bleeding risk trade-off model allows assessment of the bleeding-thrombosis trade-off in individual patients undergoing PCI to help tailor decisions regarding treatment and adjunctive antithrombotic therapies in HBR patients undergoing PCI

SHARE YOUR COMMENT