Incidence of Bleeding after 15,181 Percutaneous Biopsies and the Role of Aspirin

Thomas D. Atwell; Ryan L. Smith; Gina K. Hesley; Matthew R. Callstrom; Cathy D. Schleck; W. Scott Harmsen; J. William Charboneau; Timothy J. Welch


Am J Roentgenol. 2010;194(3):784-789. 

In This Article


With the evolution of imaging guidance, percutaneous biopsy has become a valuable tool in obtaining tissue for diagnosis. Unfortunately, this procedure is not without risk, and hemorrhage is the most feared complication after biopsy.

The results of our study have shown that the overall risk of major bleeding and death due to percutaneous biopsy is very low, 0.5% and 0.02%, respectively. These results compare favorably with the data reported in the literature (Table 4). Although the overall number of studies in the literature regarding organ biopsy and complications is enormous, multiple large studies (> 1,000 patients) have reported similar bleeding and death rates of 0–1.1% and 0–0.22% after percutaneous biopsy.[1–5,7–13]

Our results showed a significant association of bleeding with both a lower serum platelet count and a higher INR, although the mean values for each of these criteria remain well within accepted ranges for percutaneous biopsy. We believe that these findings validate our normally accepted criteria for percutaneous biopsy of a serum platelet count greater than 50 × 109/L and an INR of less than 1.6. The fact that fewer needle passes were obtained in those with major hemorrhage is likely related to the organs that bled. Specifically, we routinely obtain two samples for core biopsies of both the kidney and liver, which make up most of the bleeding complications. For core biopsies of the other organs, we are frequently more aggressive in obtaining tissue.

When we break down the cases of bleeding complications by organ system, we can contrast the rates in our study with those reported in previously published studies, including both large and smaller studies. Specifically, the 0.7% incidence of bleeding after kidney biopsy in our study compares with published rates of 0.3–6.3%.[5,11,14–16] In a review of 750 renal biopsies, Whittier and Korbet[15] reported that 48 patients (6.4%) had a major bleeding complication. Of those 48 complications, the intervention most commonly required was an RBC transfusion, which was required in 38 patients or 5.1% of all patients in that series. This rather high rate of transfusion shows the difficulty in quantifying a true complication rate because the threshold for transfusion might differ among clinical practices.

Major hemorrhage occurred in 0.5% of liver biopsies in our study versus 0–3.4% in other studies.[1–5,9–22] Although the 3.4% incidence of important bleeding reported by Little et al.[17] lies outside the normal range, 15 of the 16 patients (94%) with a bleeding complication required a transfusion; only one patient required surgical intervention for hemothorax.[17] These findings contrast with those of Giorgio et al.[1] who reported performing 16,648 liver biopsies over 22 years without a single major complication.

The significant bleeding rate of 0.2% after lung biopsy in our study is similar to the results reported in two prior studies.[23,24] There was a single symptomatic hemothorax deemed severe in a series of 660 lung biopsies (0.2%).[23] In a large, multicenter review including 9,783 lung biopsies, the incidence of major hemorrhagic complications (hemothorax, severe pulmonary hemorrhage, or hemoptysis) was 0.15%.[24]

Hemorrhage after percutaneous biopsy of the pancreas is exceptionally rare. The incidence of 1.0% in our study is likely related to the proportion of pancreas transplants. When one reviews four large series detailing biopsy of the native pancreas,[25–28] only a single case of significant hemorrhage has been reported.[25] In contrast, we previously reported two major bleeding complications (CTCAE grade 3) after 232 biopsies (0.9%) of pancreas transplants.[29] To our fault, in this current independent study, we did not specify native or transplant pancreas at the time of initial data collection; however, all six bleeding complications that are presented herein occurred in transplants. Based on our clinical practice, we are comfortable stating that most of the pancreas biopsies we perform are on transplants.

In reviewing the biopsy-related complications in our study group and those reported in other articles, a few themes emerge. First, there is universal lack of standardization in reporting complications, with complications defined and stratified by authors. Second, many of these reviews including large patient populations are derived from multicenter studies with potential variations in reporting standards[4,8,10] or from experiences derived over an extended time interval.[1–3,7,13] In either case, these factors inherently reflect different biopsy techniques in the reported population. With this in mind, the strengths of our reported experience are, first, that our study uses accepted complication criteria when defining major hemorrhage and, second, that our study reflects a standard biopsy technique used over the 6-year study interval.

Delayed bleeding complications—that is, those presenting more than 24 hours after biopsy—represented 20% of the complication in our study, which is a greater proportion of patients than we expected. Given that 3-month follow-up was lacking in 4% of our patients, this complication rate could potentially be higher. With regard to liver biopsy, the 12% incidence of delayed bleeding in our study is much greater than that reported in two large series, where 0–5% of major hemorrhagic complications occurred more than 24 hours after biopsy.[3,4] In those two studies, patients were observed either for 4 hours or overnight after biopsy. In contrast, we routinely observe patients for 2 hours before dismissal from our hospital-based radiology practice. The duration of observation after liver biopsy used in our study also differs from the 1989 American Gastroenterological Association guidelines,[30] which suggests 6 hours of observation after outpatient liver biopsy. However, in our experience, we do not think that the threefold extended period of direct patient care is justified based on the incredibly small number of delayed complications, particularly given that patients are required to stay within 30 miles of our institution the night after the biopsy.

Of the 39 major bleeding complications after renal biopsy in our study, 10 (26%) presented more than 24 hours after biopsy. This compares with 9% of patients in a series of 750 biopsies reported by Whittier and Korbet.[15] In that series, patients were observed for 24 hours, compared with our 4-hour protocol period of observation after biopsy. The extended period of observation likely allows earlier detection of complications (within the 24-hour window). In fact, only 38% of the major complications reported by Whittier and Korbet were detected within 4 hours of biopsy.

Perhaps the most interest finding in our work is related to the role of aspirin in bleeding risk. We found only a very minor difference (0.2%) in significant bleeding rate of those taking aspirin compared with those who were not. However, this does represent a 50% increase in bleeding incidence. Interestingly, in a meta-analysis of the literature related to periprocedural aspirin use, Burger et al.[31] found an approximately 50% increase in the bleeding rate in those taking aspirin at the time of surgery or biopsy. This bleeding risk must be contrasted with the important implications of aspirin withdrawal—namely, the risk of coronary and cerebrovascular events and peripheral ischemia.[32–34]

There are some important limitations of our study. We used the National Institutes of Health's CTCAE. This 71-page document is specific to bleeding complications related to various organ systems. In considering only grade 3 or greater complications, we were forced to exclude some clinically important complications such as hematuria requiring bladder catheterization and pulmonary hemorrhage with secondary desaturation requiring hospitalization; such complications are considered grade 2 on the basis of CTCAE. In addition, the criteria for transfusion may differ among patients depending on baseline hemoglobin values, with a lower threshold for transfusion in those with baseline anemia. Importantly, however, the CTCAE provides a clear black-and-white framework on which we could define complications with little ambiguity. We encourage authors to consider using the CTCAE for their studies.

A second important limitation of our study is the use of a general database maintained over several years. With a more critical appraisal of nearly 5,000 kidney biopsies, we discovered five additional complications, or a 15% increase in the bleeding complication rate (raising the incidence for kidney biopsy from 0.6% to 0.7%). If we were to extrapolate this rate to the remaining 31 nonrenal bleeding complications, we might expect to find an additional five complications, yielding the same overall complication rate of 0.5%.

A third important limitation of our study is the definition of aspirin use. We considered aspirin use to be ingestion of aspirin within 10 days before the biopsy procedure. This definition is based on the permanent inactivation of platelets due to aspirin and the 7–10 days required for complete platelet turnover.[35] One could reasonably argue that new functional platelets are continuously produced and available immediately after aspirin withdrawal; thus, a shorter interval of time would be more reasonable. Based on our current database, we are unable to stratify a large number of patients to determine the optimum period of aspirin withdrawal.

In conclusion, what are the implications of this study in our radiology practice? We have shown that significant bleeding after percutaneous biopsy is exceptionally rare. In most cases, percutaneous biopsy can be performed in patients taking aspirin. In those patients undergoing elective, nonurgent deep organ biopsy, scheduling the biopsy 10 days after the last dose of aspirin is a reasonable, but not a necessary, precaution.


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.
Post as: