GI bleeding is a serious and often fatal complication after PCI.5, 15, 16
Although post-PCI bleeding most frequently occurs at the access site, the GI tract is the second most common site of hemorrhage.10, 12, 17, 18
The overall observed incidence of GIB associated with PCI from 1998 to 2006 in this large, contemporary, national registry was 1.04%. This is lower compared with the reported incidence of GIB associated with PCI in previous studies which ranged from 1.1% to 3.0%.8–12
This finding is important since more intensive anti-platelet and anti-thrombotic therapies associated with increased GIB rates were introduced during this period.1, 12, 19, 20
During the observation period, bivalirudin, a direct thrombin inhibitor was approved for use in 2000. In the National Cardiovascular Data Registry (NCDR) Catheterization PCI (CathPCI) registry, the use of UFH, LMWH and glycoprotein IIb/IIIa inhibitors (GPI) significantly decreased and the use of bivalirudin increased among patients with ACS and non-ACS from 2005 to 2009.3
GIB occurred significantly more frequently in patients randomized to heparin + GPI when compared with bivalirudin monotherapy (0.6% vs. 0.1%).6, 12, 21, 22
A sharp decline in the incidence of GIB was noted from 2003 to 2006 in this study, which coincides with the publication of the REPLACE-2 results that brought the reduction in bleeding complications with the use of bivalirudin to the forefront. Similarly, the use of thienopyridines underwent dramatic changes in this period.
Among the NSTE-ACS patients enrolled in 4 large prospective, multicenter US registries from 1999 to 2008, the use of thienopyridines increased from 8.9% to 76.2% with a significant increase in dual anti-platelet therapy use.23
Dual antiplatelet therapy with clopidogrel and aspirin increases GI bleeding complications – GI is the site of major bleeding in 1.3% of patients treated with clopidogrel and aspirin versus 0.7% in those treated with aspirin alone.24
Prasugrel, a more potent thienopyridine, further increases the risk of major, life-threatening and fatal bleeding; and the GI site was one of the most frequent site for life-threatening bleeding.25
It is very encouraging that the wide adoption of intensive dual antiplatelet regimens did not result in an increase in GIB in the NIS patient population.
With an increase in PCI with stents, utilization of anticoagulants, dual oral anti-platelet medications, and GPI in patients with AMI and NSTE-ACS as noted from 1990 to 2008,23, 26
a parallel increase in the incidence of GIB would have been expected. The observation of risk factors for GIB associated with PCI, including age, AMI, and CRI () are consistent with other trials.10, 12
Focused efforts in identifying these high bleeding risk patients 5, 27
and tailoring therapies 23
to reduce GIB in these patients may have contributed partially to the lower observed GIB rates. Implementation of societal guidelines 26
and quality-improvement techniques targeting bleeding complications, such as renal function and weight-adjusted dosing of anti-platelet/anti-thrombotic therapies, preferential use of medications such as bivalirudin that cause less bleeding,4, 22
prophylactic treatment of high bleeding risk patients with proton pump inhibitors (PPIs),11, 28, 29
may have also contributed to the observed decrease in the rates of GIB in this period.
Previous studies have reported a 10% in-hospital mortality with development of GIB associated with PCI.8
The overall mortality in this NIS GIB group was 6.0%. The high mortality noted in patients developing GIB associated with PCI is likely multifactorial in etiology. The practice of discontinuing effective anti-platelet/anti-thrombotic agents in patients who develop GIB, has been associated with the increased risk of further ischemia, infarction, stent thrombosis and need for repeat PCI and death.5, 10, 30
Aspirin and/or thienopyridines were noted to be frequently discontinued in patients with GIB.12
Among patients triaged to PCI, 5.8% of patients with GIB developed stent thrombosis compared with 2.4% of patients without GIB.12
Other proposed mechanisms for the association between GIB and mortality are anemia and hypotension-related reduction in myocardial oxygen delivery, hemodynamic instability, blood transfusions and thrombocytopenia.8, 9, 12
The risk-adjusted in-hospital mortality did not change significantly among the ACS and non-ACS patients in the NCDR CathPCI registry from 2005 to 2009.3
A very concerning finding in this NIS PCI registry is that the adjusted risk of death after developing GIB associated with PCI remained high and largely unchanged from 1998 to 2006 (); this finding is consistent with the findings of the NCDR CathPCI registry and was noted over a longer study period.
The significant association between GI malignancies and GIB associated with PCI has not been previously observed. Patients with malignancies are generally excluded from clinical trials, and other large nationwide registries have not reported this particular information. Our findings indicate that the presence of an underlying GI malignancy is an additional significant independent predictor of GIB associated with PCI. Identifying patients with GI malignancies may reduce the rate of GIB associated with PCI. Currently, there are no systematic strategies to screen for GI malignancy in the elective PCI population. Such measures, however, may not be effective in the emergent STEMI or NSTEMI setting. While the rate of GIB in the elective PCI population was lower than that in the AMI population, elective PCI may allow substantial improvement in the processes of care and procedural outcomes. It is important to emphasize that the findings of this study does not imply causality but only an association of GI malignancy with GIB.
Data about management patterns, utilization rates or dosing of the various anti-platelet/anti-thrombotic medications and utilization rates of PPIs that have been associated with decreased GIB were not collected. The database does not allow for evaluation of practice patterns that account for the noted reduction in GIB. This precludes us from confirming an association between the trends in the utilization of specific anti-platelet/anti-thrombotic medications and PPI with the incidence of GIB. Data on other bleeding complications such as access site and retroperitoneal bleeds were not available in our dataset. The details on the type and temporal relationship of GIB to PCI were not available. The identification of GIB was determined by the local sites and not centrally adjudicated. The underreporting of smaller GIB cannot be excluded, and could have biased the estimation of the true incidence of GIB, in addition to the mortality rate associated with GIB. Importantly, the cause-effect relationship between PCI, anti-platelet/anti-thrombotic therapy, GIB and the observed outcome of mortality cannot be determined. Additional unmeasured confounders may have accounted for the observed differences. The race data was missing in about 29% of all cases. NIS does not collect long-term outcomes of GIB associated with PCI. Finally, participation in the NIS registry is voluntary and only selected centers may have participated in this registry, the results may not be generalized to all U.S. hospitals/population at large.
A temporal trend in GIB associated with PCI has not been studied, and as such this study addresses an important gap in the literature. This very large sample study represents the contemporary, real-world practice in the US and encompasses a broad range of patients who would have otherwise been excluded from randomized clinical trials. This study describes a period during which important changes in PCI management have occurred. From 1998 to 2006 the incidence of GIB associated with PCI decreased in the face of more aggressive therapies for ACS and CAD. The risk of GIB-associated death has remained high and unchanged. Underlying GI malignancy is a significant independent predictor of GIB associated with PCI. The findings of this study may help develop a basis for future investigation to better define causes of GIB associated with PCI and provide pertinent information regarding preventive management strategies to reduce morbidity and mortality.