Effect of prophylactic platelet transfusion on bleeding risk in patients with cirrhosis and thrombocytopenia undergoing ultrasound-guided liver biopsy
Article information
Abstract
Purpose
The utility of platelet transfusion for preventing post-biopsy bleeding in patients with cirrhosis and thrombocytopenia remains a topic of debate. This study investigated whether prophylactic transfusion reduces this risk.
Methods
This retrospective study included patients with cirrhosis and thrombocytopenia who underwent ultrasound-guided percutaneous liver biopsy (March 2018–October 2024) at a tertiary institution. Bleeding events were defined as major (requiring intervention or resulting in death) or minor (non-interventional hematoma). Pre- and post-transfusion platelet counts were compared, and bleeding incidence was examined by transfusion status. Logistic regression was used to identify factors associated with bleeding events. Propensity score matching (PSM) and inverse probability of treatment weighting (IPTW) were applied to assess bleeding incidence by transfusion status.
Results
This study analyzed 526 biopsies from 497 patients (mean age, 62.1±10.6 years; 365 men [73.4%]), of whom 42 (8.0%) received prophylactic platelet transfusion. While platelet counts increased after transfusion (from 54.5×109/L to 72.5×109/L, P<0.001), no significant difference was observed in bleeding incidence between the transfused and non-transfused groups (31.0% vs. 20.9%, P=0.128). Multivariate analysis indicated that platelet transfusion was not associated with all bleeding (odds ratio [OR], 1.318; P=0.449) or major bleeding (OR, 1.754; P=0.420). PSM and IPTW analyses similarly revealed no association between transfusion and all bleeding (PSM: OR, 1.173; P=0.739; IPTW: OR, 0.907; P=0.807) or major bleeding (PSM: OR, 0.658, P=0.716; IPTW: OR, 0.863; P=0.876).
Conclusion
Prophylactic platelet transfusion may not provide a universal benefit in patients with cirrhosis and thrombocytopenia, warranting a more tailored approach based on individual risk assessment.
Introduction
Ultrasonography-guided percutaneous liver biopsy is widely used for histopathological assessment of liver disease [1]. Although generally considered safe, it is an invasive procedure associated with a bleeding rate of 0.06% to 0.69% [2,3]. Thrombocytopenia, a complication in up to 76% of patients with chronic liver disease (CLD) or cirrhosis [4], is a potential risk factor for post-biopsy bleeding [5].
Traditionally, prophylactic measures such as platelet transfusion have been used in patients with CLD or cirrhosis to prevent bleeding before invasive procedures; however, emerging evidence has shifted the understanding of hemostasis in cirrhosis from a purely hypocoagulable state to the concept of “rebalanced hemostasis,” in which alterations in procoagulant and anticoagulant pathways result in a new equilibrium [6,7]. In this rebalanced state, despite thrombocytopenia and a prolonged prothrombin time (PT), compensatory mechanisms—such as increased von Willebrand factor and decreased levels of natural anticoagulants (protein C, protein S, and antithrombin)—help maintain hemostatic equilibrium [6,8]. This rebalancing challenges the traditional assumption that patients with cirrhosis are at high bleeding risk based solely on conventional laboratory parameters and has raised questions regarding the necessity and efficacy of prophylactic interventions in this population. Nevertheless, international guidelines regarding prophylactic platelet transfusion are inconsistent [9,10]. The Society of Interventional Radiology (SIR) [9] recommends considering platelet transfusion in patients with CLD when counts fall below 30×109/L [10]. The SIR does not provide separate recommendations specifically for cirrhosis, and this CLD threshold is generally applied to this patient population. The European Association for the Study of the Liver (EASL) suggests that prophylactic platelet transfusion should not be routinely performed even when counts are below 50×109/L but should instead be considered on a case-by-case basis. The American Association for the Study of Liver Diseases (AASLD) states that correction of thrombocytopenia may be reasonable when platelet counts are less than 30–50×109/L, based on clinical judgment [11]. Evidence supporting the clinical efficacy of prophylactic platelet transfusion in patients with thrombocytopenia—especially those with cirrhosis—remains limited. In patients with thrombocytopenia and liver disease, platelet transfusion typically results in only a modest increase in platelet count (approximately 5–10×109/L) and has a short half-life of 2–4 days [12]. Notably, there is a marked absence of evidence that this transient increase reduces bleeding risk.
Platelet transfusion has several disadvantages, including medical complications (febrile reaction, allergic reaction, transfusion-related infection, and graft-versus-host disease), as well as procedural delays and increased costs [13]. Furthermore, in patients with cirrhosis, transfusion may increase portal pressure, thereby increasing the risk of further bleeding [14]. The risk-benefit ratio of prophylactic platelet transfusion in this population remains uncertain, and the current lack of high-quality evidence represents a significant unmet clinical need, underscoring the importance of gathering sufficient evidence to inform appropriate management decisions. The primary aim of this study was to assess the clinical impact of prophylactic platelet transfusion on post-procedure bleeding in patients with cirrhosis and thrombocytopenia undergoing ultrasound-guided percutaneous liver biopsy. Additionally, the study sought to identify factors associated with bleeding events in this population.
Materials and Methods
Compliance with Ethical Standards
This retrospective, observational, single-institution study received approval from the institutional review board of Asan Medical Center (IRB No. 2022-1330), a tertiary referral hospital in South Korea, which waived the requirement for written informed consent.
Study Population
Electronic medical records were reviewed to identify patients with cirrhosis who underwent ultrasound-guided percutaneous liver biopsy between March 2018 and October 2024. Eligible patients were those with (1) evidence of cirrhosis confirmed by pathology or imaging (morphologic changes of the liver, including surface nodularity and redistribution of segmental volume [15], with either evidence of portal hypertension [i.e., splenomegaly, portal collaterals such as esophageal or gastric varices, and ascites] or fibrosis stage 4 on transient elastography [16]); (2) thrombocytopenia, defined as a platelet count below 150×109/L; and (3) available clinical information and laboratory data within 7 days before and after liver biopsy. Patients who underwent ultrasound-guided local ablation for liver lesions immediately following the biopsy were excluded. Of the 529 eligible biopsies obtained from 500 patients with cirrhosis and thrombocytopenia, three biopsies from three patients were excluded because they underwent radiofrequency ablation immediately after biopsy. The final study comprised 526 biopsies from 497 patients with thrombocytopenia (mean age, 62.1±10.6 years; 365 men) (Fig. 1).
Flow diagram of the study population.
Data Collection
Demographic and laboratory data obtained before liver biopsy, including platelet count, PT–international normalized ratio (PT-INR), and liver function tests (LFTs; serum aspartate aminotransferase, alanine aminotransferase, alkaline phosphatase, and total bilirubin), were recorded. When platelet transfusion was performed prior to biopsy, the post-transfusion platelet count was also documented. If fresh frozen plasma (FFP) was administered before biopsy, PT-INR values after FFP transfusion were obtained. The presence of malignancy was recorded. Measurements outside the normal ranges for aspartate aminotransferase (normal range, <40 IU/L), alanine aminotransferase (<40 IU/L), alkaline phosphatase (40–120 IU/L), or total bilirubin (0.2–1.2 mg/dL) were considered abnormal.
Thrombocytopenia was categorized as mild (75×109/L≤platelet count<150×109/L), moderate (50×109/L≤platelet count<75×109/L), or severe (platelet count<50×109/L) [4]. A prolonged PT-INR was defined as ≥1.5. At the authors’ institution, general laboratory thresholds for performing liver biopsy include a platelet count ≥50×109/L and PT-INR <1.5. Transfusion is considered when these thresholds are not met or when deemed necessary by the interventionist or referring clinician. Biopsy and prophylactic transfusion in patients who did not meet these criteria were performed only after an individualized risk-benefit assessment through discussion between the interventionist and the referring clinician. For platelet transfusion, platelet concentrates were administered 3–5 hours before biopsy, and the pre-biopsy platelet count was checked after transfusion.
To assess imaging signs of portal hypertension, this study applied previously validated imaging-based criteria for clinically significant portal hypertension [17,18], defined as splenomegaly accompanied by at least one of the following findings on pre-biopsy cross-sectional imaging: gastroesophageal varices, spontaneous portosystemic shunts, or ascites. The type of liver biopsy (targeted biopsy for a focal liver mass or blind biopsy), the number of needle passes, and the presence of perihepatic ascites at the needle insertion site were also recorded.
Liver Biopsy
Ultrasound-guided liver biopsies were performed by more than 10 board-certified radiologists, each with at least 4 years of experience in abdominal ultrasonography. Biopsies were performed using an intercostal or subxiphoid approach with local lidocaine injection, with patients placed in the supine or left lateral decubitus position. The biopsies utilized an 18-gauge biopsy needle (STARCUT, TSK Laboratory, Tochigi, Japan or MISSON Bard Biopsy System, Tempe, AZ, USA; 10–22 mm core length), enabling continuous monitoring of the needle path and avoidance of large intrahepatic vessels or bile ducts to ensure the safest route. Following biopsy, immediate grayscale and color Doppler scanning were routinely performed to detect signs of post-biopsy bleeding, including patent track signs (defined as a linear color-flow signal toward the liver capsule along the needle tract) or a new perihepatic hematoma. If these signs were noted, the biopsy site was compressed with the probe until they disappeared. A tight compression bandage was applied to the biopsy site, and patients were then positioned with the biopsied side down. Patients were closely monitored for signs of bleeding, and if bleeding was suspected, a complete blood count, along with ultrasonography and/or computed tomography, was performed.
Definition of Post-Biopsy Bleeding
All bleeding events occurring within 7 days of biopsy were recorded. Bleeding events were categorized as major or minor. A major bleeding event was defined as grade 3 or higher in accordance with the Common Terminology Criteria for Adverse Events, version 5.0 [19]. Grade 3 bleeding necessitates blood transfusion or radiologic, endoscopic, or surgical intervention. Grades 4 and 5 are life-threatening or fatal events, with the former requiring urgent intervention. Minor bleeding was defined as hematomas or intra-abdominal blood on post-biopsy imaging, with no changes in vital signs and no need for intervention.
Statistical Analysis
Baseline characteristics of patients with and without platelet transfusion were compared using the chi-square test or Fisher's exact test (categorical variables) and the Student’s t-test (continuous variables). Pre- and post-transfusion platelet counts were compared using the Wilcoxon signed-rank test. Bleeding incidence according to whether platelet transfusion was administered prior to biopsy was compared using the Fisher exact test.
Univariate and multivariate logistic regression analyses were utilized to identify independent risk factors for bleeding. Analyses were adjusted for age, sex, etiology of liver disease, presence of cancer, biopsy type (target vs. blind), number of needle passes, presence of perihepatic ascites, PT-INR prolongation, thrombocytopenia grade (mild/moderate vs. severe), LFT abnormality, and presence of portal hypertension. Variables with a P-value less than 0.1 in univariate analysis were entered into the multivariate models. To minimize the impact of potential confounders, propensity score analyses were performed using propensity score matching (PSM) and inverse probability of treatment weighting (IPTW). The above-mentioned variables were used to construct the propensity score model. PSM was conducted using nearest-neighbor matching without replacement, allowing a variable matching ratio of up to 1:4. A caliper width of 0.2 of the standard deviation of the logit of the propensity score was utilized to ensure adequate covariate balance. Covariate balance after adjustment was assessed with standardized mean differences (SMDs). The C-statistic and Hosmer-Lemeshow test were used to assess discrimination and calibration, respectively, of the propensity score model. Logistic regression models with robust standard errors accounting for the clustering of matched pairs and a weighted logistic regression model were used for the propensity score–matched set. All statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC, USA) and R version 4.1.1 (R Foundation for Statistical Computing, Vienna, Austria). P-values of less than 0.05 were considered to indicate statistical significance.
Results
Baseline Characteristics
Among the final study population of 526 biopsies from 497 patients, no patients were lost to follow-up, and no laboratory data within 7 days before or after biopsy were missing. All patients receiving anticoagulation or antiplatelet medication discontinued these medications 7 days before biopsy. The coaxial needle technique was used in eight of 526 biopsies (1.5%). No adjunctive hemostatic measures, including needle-tract plugging with gelatin sponge [20], were performed.
Baseline characteristics are summarized in Table 1. Prophylactic platelet transfusion was performed before 42 biopsies (8.0%). The distribution and degree of thrombocytopenia differed between the transfused and non-transfused groups (P<0.001). In the non-transfused group, most cases had mild or moderate thrombocytopenia (97.9%, 474/484), and only 2.1% (10/484) had severe thrombocytopenia. In the transfused group, 57.1% (24/42) had mild or moderate thrombocytopenia, while 42.9% (18/42) had severe thrombocytopenia.
Baseline characteristics of 526 biopsies according to platelet transfusion status
Changes in Platelet Count before and after Transfusion
Of the 42 biopsies in patients who received prophylactic platelet transfusion, 18 had severe thrombocytopenia and 24 had mild (nine biopsies) or moderate (15 biopsies) thrombocytopenia. The median platelet count increased significantly from 54.5×109/L (interquartile range [IQR], 45.5 to 72.0×109/L) before transfusion to 72.5×109/L (IQR, 57.8 to 93.8×109/L) after transfusion (P<0.001). When stratified by the degree of thrombocytopenia, both patients with mild/moderate thrombocytopenia and those with severe thrombocytopenia showed a significant increase in platelet count after transfusion (mild-moderate: from 68×109/L [IQR, 57.8 to 83.3×109/L] to 82.5×109/L [IQR, 73.8 to 103×109/L]; severe: from 43.5×109/L [IQR, 41.0 to 48.0×109/L] to 57.5×109/L [IQR, 53.3 to 61.0×109/L]; P<0.001) (Fig. 2).
Changes in platelet counts after platelet transfusion.
The number of patients is shown in parentheses.
Bleeding Incidence According to the Severity of Thrombocytopenia and Platelet Transfusion Status
Bleeding events occurred after 114 biopsies (21.7%), with major bleeding events observed after 18 biopsies (3.4%). Among patients with major bleeding events, 16 (88.9%) received a red blood cell transfusion and three (16.7%) underwent transarterial embolization to control the bleeding; one patient received both interventions.
When bleeding incidence was compared by platelet transfusion status (Fig. 3), the incidence was 31.0% (13/42) in patients who received platelet transfusion and 20.9% (101/484) in those who did not; this difference was not significant (P=0.132). Notably (and paradoxically), bleeding incidence in the mild/moderate thrombocytopenia group was higher among patients who received a platelet transfusion than among those who did not, although the between-group differences were not significant (all bleeding events: non-transfused, 20.5% [97/474] vs. transfused, 33.3% [8/24], P=0.131; major bleeding events: non-transfused, 2.7% [13/474] vs. transfused, 4.2% [1/24], P=0.681). In the severe thrombocytopenia group, bleeding incidence was numerically lower among patients who received a platelet transfusion than among those who did not, but again the differences were not statistically significant (all bleeding events: non-transfused, 40% [4/10] vs. transfused, 27.8% [5/18]; P=0.507; major bleeding events: non-transfused, 20% [2/10] vs. transfused, 11.1% [2/18]; P=0.520).
Bleeding incidence according to platelet transfusion status.
A. Incidence of all bleeding events was compared according to platelet transfusion status. B. Incidence of major bleeding events was compared according to platelet transfusion status.
Effect of Prophylactic Platelet Transfusion on Bleeding Incidence
Multivariate analysis revealed that neither platelet transfusion nor thrombocytopenia severity was significantly associated with bleeding. For all bleeding events, the presence of perihepatic ascites (odds ratio [OR], 2.332; 95% confidence interval [CI], 1.119 to 4.859; P=0.024) (Table 2) and portal hypertension (OR, 1.723; 95% CI, 1.047 to 2.834; P=0.032) were significant factors. For major bleeding, more than two needle passes (OR, 7.007; 95% CI, 1.575 to 31.178; P=0.011) and prolonged PT-INR (OR, 6.531; 95% CI, 2.328 to 18.324; P<0.001) were significantly associated.
Logistic regression analyses for all bleeding and major bleeding events
After PSM, all SMDs in the PSM dataset were below 0.2. In the IPTW dataset, all covariates except the etiology of liver disease and number of needle passes achieved an SMD <0.2, indicating relatively small residual imbalance. Data from 115 biopsies were used for 1:4 PSM (27 with platelet transfusion and 88 without transfusion). The constructed propensity score model showed excellent discrimination (C-statistic, 0.872) and calibration (Hosmer-Lemeshow, P=0.4924). After adjustment, no statistically significant association was observed between platelet transfusion and bleeding events (all bleeding and major bleeding) across all analytical approaches (Table 3). For all bleeding events, multivariate adjustment showed no significant association with platelet transfusion (OR, 1.318; 95% CI, 0.644 to 2.698; P=0.449), a finding confirmed by PSM (OR, 1.173; 95% CI, 0.458 to 3.005; P=0.739) and IPTW (OR, 0.907; 95% CI, 0.413 to 1.99; P=0.807) analyses. For major bleeding events, multivariate adjustment similarly showed no significant association with platelet transfusion (OR, 1.754; 95% CI, 0.447 to 6.874; P=0.420), which was again confirmed by both PSM (OR, 0.658; 95% CI, 0.068 to 6.316; P=0.716) and IPTW (OR, 0.863; 95% CI, 0.137 to 5.443; P=0.876).
Effect of platelet transfusion on bleeding risk: adjusted analyses
Discussion
This study evaluated the impact of prophylactic platelet transfusion in patients with cirrhosis and thrombocytopenia undergoing ultrasound-guided percutaneous liver biopsy. The findings revealed no significant reduction in bleeding incidence. Across all analytical approaches, including PSM and IPTW, there was no apparent association between platelet transfusion and post-biopsy bleeding. This finding is meaningful given the limited evidence assessing whether platelet transfusion provides clinical benefit in reducing bleeding after invasive procedures in patients with cirrhosis and thrombocytopenia. Liver disease causes complex changes in the hemostatic system [21], resulting in a new hemostatic balance. The combination of low platelet counts, prolonged PT or activated partial thromboplastin time, and low fibrinogen levels in patients with liver disease has traditionally been interpreted as indicating an increased bleeding risk, although these measures may not accurately reflect the rebalanced hemostatic state [22]. Therefore, the present study suggests that in patients with cirrhosis and thrombocytopenia, platelet transfusion may have limited clinical benefit in reducing bleeding risk after liver biopsy and may not provide universal benefit in this population. Rather than routine pre-procedural platelet transfusion, a tailored approach based on individualized bleeding risk assessment is warranted.
These results showed a significant increase in platelet counts across patients with varying thrombocytopenia severity, consistent with previous studies [23,24]. However, all post-transfusion platelet counts remained within the thrombocytopenic range, representing only modest improvements that may have minimal impact on hemostatic function. Consistent with this, platelet transfusion was not associated with a statistically significant reduction in bleeding incidence across thrombocytopenia severity strata, which aligns with prior studies [23,25,26].
Perihepatic ascites (P=0.024) and portal hypertension (P=0.032) were significantly associated with all bleeding events. Large-volume ascites is considered a relative contraindication to percutaneous liver biopsy [27] due to increased liver mobility and the risk of uncontrolled bleeding into the ascites, particularly because compression-based hemostasis may be ineffective in such cases. Thus, in patients with large-volume ascites in whom liver biopsy is indicated, percutaneous biopsy after substantial paracentesis or transjugular liver biopsy may be an appropriate option. Portal hypertension is a well-recognized contributor to hemorrhagic events in patients with cirrhosis [28] and has been implicated in increased bleeding risk during invasive procedures [29-31]. However, its role in post-liver biopsy bleeding remains incompletely characterized. One study evaluating complications after liver biopsy reported an OR of 2.1 for portal hypertension, although this did not reach statistical significance [32]. The present findings provide further evidence that portal hypertension is associated with post-biopsy bleeding.
More than two needle passes (P=0.011) and PT-INR prolongation (P<0.001) were significantly associated with major bleeding events. In general, the risk of hemorrhage increases with the number of liver capsule punctures, and the present findings align with previous studies showing that more aggressive biopsy with multiple needle passes increases bleeding risk [33,34]. Regarding PT-INR prolongation, current guidelines provide varying recommendations [9,10]. The SIR [9] considers a PT-INR <2.5 potentially acceptable for liver biopsy in patients with CLD, but emphasizes that plasma transfusion should be used judiciously in these patients given the concept of rebalanced hemostasis, as well as the potential for increased portal pressure and transfusion-related adverse events. The EASL and the AASLD [10,11] do not suggest specific PT-INR cutoffs for liver biopsy and do not recommend routine correction of prolonged PT-INR by transfusion. Despite this variability and the lack of consensus regarding PT-INR thresholds, the present results suggest that PT-INR prolongation remains a significant risk factor that warrants careful consideration during clinical decision-making.
Although the present study does not provide a clear framework for selecting candidates for prophylactic platelet transfusion among patients with cirrhosis, the findings—together with the prior literature [32-34]—indicate that certain patients may be at higher risk of post-biopsy bleeding. Factors such as prolonged PT-INR, substantial perihepatic ascites, the need for multiple needle passes, and severe thrombocytopenia, particularly when accompanied by additional coagulation abnormalities, have been associated with increased bleeding risk [32-34]. In such high-risk patients, preventive strategies may be warranted; however, because prophylactic platelet transfusion did not confer a universal benefit in the present study, transfusion alone may be insufficient, and alternative or adjunctive hemostatic measures may be required. Yoon et al. [20] recently proposed a coaxial technique with needle-tract plugging using gelatin sponge to reduce bleeding complications, which may represent a useful preventive approach. Further studies are required to identify which patient subgroups may benefit from specific prophylactic interventions.
This study has certain limitations. First, the retrospective study design could have introduced selection bias. For example, the paradoxical finding that patients with mild or moderate thrombocytopenia who received platelet transfusion had a higher incidence of bleeding than those who did not (although not statistically significant) may have been influenced by selection bias, as patients who received platelet transfusion were likely selected based on additional bleeding risk factors, such as greater liver disease severity. To mitigate this bias, PSM and IPTW were used to balance baseline characteristics between groups and obtain more robust estimates of the treatment effect. Second, the bleeding incidence in the present study (all bleeding events, 21.7%; major bleeding events, 3.4%) was higher than previously reported—particularly for major bleeding (≤0.69% in prior studies) [3,10,35]—likely because the present study included only thrombocytopenic patients with cirrhosis. In addition, illness severity in this cohort was relatively high, with 70% of patients having an underlying malignancy. This may have led to more proactive and aggressive management after biopsy. Third, the number of patients who received prophylactic platelet transfusion was relatively small (8%), which may have limited the ability to detect certain associations between platelet transfusion and bleeding events. Finally, although statistical significance was not achieved, the numerically lower bleeding rates after transfusion among patients with severe thrombocytopenia suggest that a clinical benefit may exist in a subset of this population, warranting further investigation in larger studies that consider a more extensive set of covariates.
In conclusion, while prophylactic platelet transfusion increased platelet counts, it may not provide universal benefit in patients with cirrhosis and thrombocytopenia, warranting a more tailored approach based on individual risk assessment.
Notes
Author Contributions
Conceptualization: Park HJ. Data acquisition: Jang HJ, Park HJ. Data analysis or interpretation: Jang HJ, Park HJ, Choi J, Heo S, Choi SJ, Kim SY, Lee SS. Drafting of the manuscript: Jang HJ. Critical revision of the manuscript: Park HJ, Choi J, Heo S, Choi SJ, Kim SY, Lee SS. Approval of the final version of the manuscript: all authors.
Conflict of Interest
No potential conflict of interest relevant to this article was reported.
References
Article information Continued
Notes
Key points
In patients with cirrhosis and thrombocytopenia undergoing ultrasound-guided percutaneous liver biopsy, prophylactic platelet transfusion did not reduce the incidence of overall or major bleeding events. Perihepatic ascites, portal hypertension, more than two needle passes, and prolonged prothrombin time-international normalized ratio were bleeding risk factors; thrombocytopenia severity and transfusion status were not. The findings support a tailored, individualized approach to bleeding risk assessment rather than routine prophylactic platelet transfusion in this patient population.